<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body text="#000000" bgcolor="#ecca99">
    <p><font size="+1">I see, sort of a bound.</font></p>
    <p><font size="+1">so "understanding" could be a continuum..    and
        further dependent on whom we are interacting with... </font><br>
    </p>
    <div class="moz-cite-prefix">On 2/4/22 1:37 PM, Dietterich, Thomas
      wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:PH0P222MB00269D0BC488F2039771C908BE299@PH0P222MB0026.NAMP222.PROD.OUTLOOK.COM">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      <meta name="Generator" content="Microsoft Word 15 (filtered
        medium)">
      <!--[if !mso]><style>v\:* {behavior:url(#default#VML);}
o\:* {behavior:url(#default#VML);}
w\:* {behavior:url(#default#VML);}
.shape {behavior:url(#default#VML);}
</style><![endif]-->
      <style>@font-face
        {font-family:"Cambria Math";
        panose-1:2 4 5 3 5 4 6 3 2 4;}@font-face
        {font-family:Calibri;
        panose-1:2 15 5 2 2 2 4 3 2 4;}@font-face
        {font-family:"Lucida Console";
        panose-1:2 11 6 9 4 5 4 2 2 4;}@font-face
        {font-family:UICTFontTextStyleBody;}@font-face
        {font-family:UICTFontTextStyleItalicBody;}@font-face
        {font-family:"Courier New \;color\:\#1F497D";
        panose-1:0 0 0 0 0 0 0 0 0 0;}p.MsoNormal, li.MsoNormal, div.MsoNormal
        {margin:0in;
        margin-bottom:.0001pt;
        font-size:11.0pt;
        font-family:"Calibri",sans-serif;
        color:black;}a:link, span.MsoHyperlink
        {mso-style-priority:99;
        color:blue;
        text-decoration:underline;}a:visited, span.MsoHyperlinkFollowed
        {mso-style-priority:99;
        color:purple;
        text-decoration:underline;}p.msonormal0, li.msonormal0, div.msonormal0
        {mso-style-name:msonormal;
        mso-margin-top-alt:auto;
        margin-right:0in;
        mso-margin-bottom-alt:auto;
        margin-left:0in;
        font-size:11.0pt;
        font-family:"Calibri",sans-serif;
        color:black;}span.EmailStyle19
        {mso-style-type:personal;
        font-family:"Calibri",sans-serif;
        color:windowtext;}span.EmailStyle20
        {mso-style-type:personal-reply;
        font-family:"Calibri",sans-serif;
        color:windowtext;}.MsoChpDefault
        {mso-style-type:export-only;
        font-size:10.0pt;}div.WordSection1
        {page:WordSection1;}</style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]-->
      <div class="WordSection1">
        <p class="MsoNormal"><span style="color:windowtext">I mean that
            if you only say a system is “understanding” X if it can
            enumerate all of the consequences of X, then you have solved
            what is known as the “Ramification Problem”. And it is easy
            to show that this is impossible. Hence, our criteria for
            saying that a system “understandings” must lie somewhere
            between “doing the right thing in this one situation” and
            knowing all of the consequences of its beliefs.
            <o:p></o:p></span></p>
        <p class="MsoNormal"><span style="color:windowtext"><o:p> </o:p></span></p>
        <p class="MsoNormal"><span style="color:windowtext">--Tom<o:p></o:p></span></p>
        <p class="MsoNormal"><span style="color:windowtext"><o:p> </o:p></span></p>
        <div>
          <p class="MsoNormal"><span
              style="font-size:10.0pt;font-family:"Lucida
              Console";color:windowtext">Thomas G. Dietterich,
              Distinguished Professor Voice: 541-737-5559<o:p></o:p></span></p>
          <p class="MsoNormal"><span
              style="font-size:10.0pt;font-family:"Lucida
              Console";color:windowtext">School of Electrical
              Engineering              FAX: 541-737-1300<o:p></o:p></span></p>
          <p class="MsoNormal"><span
              style="font-size:10.0pt;font-family:"Lucida
              Console";color:windowtext">  and Computer
              Science                        URL:
              eecs.oregonstate.edu/~tgd<o:p></o:p></span></p>
          <p class="MsoNormal"><span
              style="font-size:10.0pt;font-family:"Lucida
              Console";color:windowtext">US Mail: 1148 Kelley
              Engineering Center
              <o:p></o:p></span></p>
          <p class="MsoNormal"><span
              style="font-size:10.0pt;font-family:"Lucida
              Console";color:windowtext">Office: 2067 Kelley
              Engineering Center<o:p></o:p></span></p>
          <p class="MsoNormal"><span
              style="font-size:10.0pt;font-family:"Lucida
              Console";color:windowtext">Oregon State Univ.,
              Corvallis, OR 97331-5501<o:p></o:p></span></p>
        </div>
        <p class="MsoNormal"><span style="color:windowtext"><o:p> </o:p></span></p>
        <div>
          <div style="border:none;border-top:solid #E1E1E1
            1.0pt;padding:3.0pt 0in 0in 0in">
            <p class="MsoNormal"><b><span style="color:windowtext">From:</span></b><span
                style="color:windowtext"> Stephen José Hanson
                <a class="moz-txt-link-rfc2396E" href="mailto:jose@rubic.rutgers.edu"><jose@rubic.rutgers.edu></a>
                <br>
                <b>Sent:</b> Friday, February 4, 2022 4:01 AM<br>
                <b>To:</b> Dietterich, Thomas
                <a class="moz-txt-link-rfc2396E" href="mailto:tgd@oregonstate.edu"><tgd@oregonstate.edu></a>; Gary Marcus
                <a class="moz-txt-link-rfc2396E" href="mailto:gary.marcus@nyu.edu"><gary.marcus@nyu.edu></a>; Danko Nikolic
                <a class="moz-txt-link-rfc2396E" href="mailto:danko.nikolic@gmail.com"><danko.nikolic@gmail.com></a><br>
                <b>Cc:</b> AIhub <a class="moz-txt-link-rfc2396E" href="mailto:aihuborg@gmail.com"><aihuborg@gmail.com></a>;
                <a class="moz-txt-link-abbreviated" href="mailto:connectionists@mailman.srv.cs.cmu.edu">connectionists@mailman.srv.cs.cmu.edu</a><br>
                <b>Subject:</b> Re: Connectionists: Stephen Hanson in
                conversation with Geoff Hinton<o:p></o:p></span></p>
          </div>
        </div>
        <p class="MsoNormal"><o:p> </o:p></p>
        <p><span style="color:#D73F09">[This email originated from
            outside of OSU. Use caution with links and attachments.]</span><o:p></o:p></p>
        <div>
          <p><span style="font-size:13.5pt">Tom, understanding is a
              theorem?</span><o:p></o:p></p>
          <p><span style="font-size:13.5pt">you mean it should be a
              theorem?</span><o:p></o:p></p>
          <p><span style="font-size:13.5pt">and yes, if you are having
              brain surgery.. you hope your surgeon, "understands" what
              they are doing..</span><o:p></o:p></p>
          <p><span style="font-size:13.5pt">Steve</span><o:p></o:p></p>
          <div>
            <p class="MsoNormal">On 2/3/22 12:31 PM, Dietterich, Thomas
              wrote:<o:p></o:p></p>
          </div>
          <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt">
            <p class="MsoNormal">“Understanding” is not a Boolean. It is
              a theorem that no system can enumerate all of the
              consequences of a state of affairs in the world.
              <o:p></o:p></p>
            <p class="MsoNormal"> <o:p></o:p></p>
            <p class="MsoNormal">For low-stakes application work, we can
              be satisfied by a system that “does the right thing”. If
              the system draws a good picture, that’s sufficient. It
              “understood” the request.<o:p></o:p></p>
            <p class="MsoNormal"> <o:p></o:p></p>
            <p class="MsoNormal">But for higher-stakes
              applications---and for advancing the science---we seek a
              causal account of how the components of a system cause it
              to do the right thing. We are hoping that a small set of
              mechanisms can produce broad coverage of intelligent
              behavior. This gives us confidence that the system will
              respond correctly outside of the narrow tasks on which we
              have tested it.
              <o:p></o:p></p>
            <p class="MsoNormal"> <o:p></o:p></p>
            <p class="MsoNormal">--Tom <o:p></o:p></p>
            <p class="MsoNormal"> <o:p></o:p></p>
            <div>
              <p class="MsoNormal"><span
                  style="font-family:"Courier New
                  ;color:#1F497D",serif">Thomas G. Dietterich,
                  Distinguished Professor Emeritus</span><o:p></o:p></p>
              <p class="MsoNormal"><span
                  style="font-family:"Courier New
                  ;color:#1F497D",serif">School of Electrical
                  Engineering and Computer
                  Science                       
                </span><o:p></o:p></p>
              <p class="MsoNormal"><span
                  style="font-family:"Courier New
                  ;color:#1F497D",serif">US Mail: 1148 Kelley
                  Engineering Center      
                </span><o:p></o:p></p>
              <p class="MsoNormal"><span
                  style="font-family:"Courier New
                  ;color:#1F497D",serif"> </span><o:p></o:p></p>
              <p class="MsoNormal"><span
                  style="font-family:"Courier New
                  ;color:#1F497D",serif">Office: 2067 Kelley
                  Engineering Center      
                </span><o:p></o:p></p>
              <p class="MsoNormal"><span
                  style="font-family:"Courier New
                  ;color:#1F497D",serif">Oregon State Univ.,
                  Corvallis, OR 97331-5501</span><o:p></o:p></p>
              <p class="MsoNormal"><span
                  style="font-family:"Courier New
                  ;color:#1F497D",serif">Voice: 541-737-5559; FAX:
                  541-737-1300</span><o:p></o:p></p>
              <p class="MsoNormal"><span
                  style="font-family:"Courier New
                  ;color:#1F497D",serif">URL:
                  <a
href="https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fweb.engr.oregonstate.edu%2F~tgd%2F&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=0GMBZ29dC97h0bfhQXJ7yS47%2FTFUI9b8POzuhz1SZT4%3D&reserved=0"
                    moz-do-not-send="true">
                    http://web.engr.oregonstate.edu/~tgd/</a></span><o:p></o:p></p>
            </div>
            <p class="MsoNormal"> <o:p></o:p></p>
            <div>
              <div style="border:none;border-top:solid #E1E1E1
                1.0pt;padding:3.0pt 0in 0in 0in">
                <p class="MsoNormal"><b>From:</b> Connectionists <a
                    href="mailto:connectionists-bounces@mailman.srv.cs.cmu.edu"
                    moz-do-not-send="true">
<connectionists-bounces@mailman.srv.cs.cmu.edu></a> <b>On Behalf
                    Of </b>Gary Marcus<br>
                  <b>Sent:</b> Thursday, February 3, 2022 8:26 AM<br>
                  <b>To:</b> Danko Nikolic <a
                    href="mailto:danko.nikolic@gmail.com"
                    moz-do-not-send="true"><danko.nikolic@gmail.com></a><br>
                  <b>Cc:</b> <a
                    href="mailto:connectionists@mailman.srv.cs.cmu.edu"
                    moz-do-not-send="true">connectionists@mailman.srv.cs.cmu.edu</a>;
                  AIhub
                  <a href="mailto:aihuborg@gmail.com"
                    moz-do-not-send="true"><aihuborg@gmail.com></a><br>
                  <b>Subject:</b> Re: Connectionists: Stephen Hanson in
                  conversation with Geoff Hinton<o:p></o:p></p>
              </div>
            </div>
            <p class="MsoNormal"> <o:p></o:p></p>
            <p><span style="color:#D73F09">[This email originated from
                outside of OSU. Use caution with links and attachments.]</span><o:p></o:p></p>
            <div>
              <p class="MsoNormal">Dear Danko, <o:p></o:p></p>
              <div>
                <p class="MsoNormal"> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">Well said. I had a somewhat similar
                  response to Jeff Dean’s 2021 TED talk, in which he
                  said (paraphrasing from memory, because I don’t
                  remember the precise words) that the famous 200 Quoc
                  Le unsupervised model [<a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstatic.googleusercontent.com%2Fmedia%2Fresearch.google.com%2Fen%2F%2Farchive%2Funsupervised_icml2012.pdf&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=%2Fttn8tKZ9K6zekF3kO7qxz3xLirDNc%2Bt4aQFht6mcQw%3D&reserved=0"
                    moz-do-not-send="true">https://static.googleusercontent.com/media/research.google.com/en//archive/unsupervised_icml2012.pdf</a>]
                  had learned the concept of a ca. In reality the model
                  had clustered together some catlike images based on
                  the image statistics that it had extracted, but it was
                  a long way from a full, counterfactual-supporting
                  concept of a cat, much as you describe below. <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal"> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">I fully agree with you that the
                  reason for even having a semantics is as you put it,
                  "to 1) learn with a few examples and 2) apply the
                  knowledge to a broad set of situations.” GPT-3
                  sometimes gives the appearance of having done so, but
                  it falls apart under close inspection, so the problem
                  remains unsolved.<o:p></o:p></p>
              </div>
              <div>
                <div>
                  <p class="MsoNormal"> <o:p></o:p></p>
                </div>
                <div>
                  <p class="MsoNormal">Gary<o:p></o:p></p>
                  <div>
                    <p class="MsoNormal"><br>
                      <br>
                      <br>
                      <o:p></o:p></p>
                    <blockquote
                      style="margin-top:5.0pt;margin-bottom:5.0pt">
                      <div>
                        <p class="MsoNormal">On Feb 3, 2022, at 3:19 AM,
                          Danko Nikolic <<a
                            href="mailto:danko.nikolic@gmail.com"
                            moz-do-not-send="true">danko.nikolic@gmail.com</a>>
                          wrote:<o:p></o:p></p>
                      </div>
                      <p class="MsoNormal"> <o:p></o:p></p>
                      <div>
                        <div>
                          <p class="MsoNormal">G. Hinton wrote: "I
                            believe that any reasonable person would
                            admit that if you ask a neural net to draw a
                            picture of a hamster wearing a red hat and
                            it draws such a picture, it understood the
                            request."
                            <o:p></o:p></p>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">I would like to suggest
                              why drawing a hamster with a red hat does
                              not necessarily imply understanding of the
                              statement "hamster wearing a red hat".<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">To understand that
                              "hamster wearing a red hat" would mean
                              inferring, in newly emerging situations of
                              this hamster, all the real-life
                              implications that the red hat brings to
                              the little animal.<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <div>
                              <p class="MsoNormal">What would happen to
                                the hat if the hamster rolls on its
                                back? (Would the hat fall off?)<o:p></o:p></p>
                            </div>
                          </div>
                          <div>
                            <p class="MsoNormal">What would happen to
                              the red hat when the hamster enters its
                              lair? (Would the hat fall off?)<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">What would happen to
                              that hamster when it goes foraging? (Would
                              the red hat have an influence on finding
                              food?)<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">What would happen in a
                              situation of being chased by a predator?
                              (Would it be easier for predators to spot
                              the hamster?)<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">...and so on.<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">Countless many
                              questions can be asked. One has understood
                              "hamster wearing a red hat" only if one
                              can answer reasonably well many of such
                              real-life relevant questions. Similarly, a
                              student has understood materias in a class
                              only if they can apply the materials in
                              real-life situations (e.g., applying
                              Pythagora's theorem). If a student gives a
                              correct answer to a multiple choice
                              question, we don't know whether the
                              student understood the material or whether
                              this was just rote learning (often, it is
                              rote learning). <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">I also suggest that
                              understanding also comes together with
                              effective learning: We store new
                              information in such a way that we can
                              recall it later and use it effectively 
                              i.e., make good inferences in newly
                              emerging situations based on this
                              knowledge.<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">In short: Understanding
                              makes us humans able to 1) learn with a
                              few examples and 2) apply the knowledge to
                              a broad set of situations. <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">No neural network today
                              has such capabilities and we don't know
                              how to give them such capabilities. Neural
                              networks need large amounts of
                              training examples that cover a large
                              variety of situations and then
                              the networks can only deal with what the
                              training examples have already covered.
                              Neural networks cannot extrapolate in that
                              'understanding' sense.<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">I suggest that
                              understanding truly extrapolates from a
                              piece of knowledge. It is not about
                              satisfying a task such as translation
                              between languages or drawing hamsters with
                              hats. It is how you got the capability to
                              complete the task: Did you only have a few
                              examples that covered something different
                              but related and then you extrapolated from
                              that knowledge? If yes, this is going in
                              the direction of understanding. Have you
                              seen countless examples and then
                              interpolated among them? Then perhaps it
                              is not understanding.<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">So, for the case of
                              drawing a hamster wearing a red hat,
                              understanding perhaps would have taken
                              place if the following happened before
                              that:<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">1) first, the network
                              learned about hamsters (not many examples)<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">2) after that the
                              network learned about red hats (outside
                              the context of hamsters and without many
                              examples) <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">3) finally the network
                              learned about drawing (outside of the
                              context of hats and hamsters, not many
                              examples)<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">After that, the network
                              is asked to draw a hamster with a red hat.
                              If it does it successfully, maybe we have
                              started cracking the problem of
                              understanding.<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">Note also that this
                              requires the network to learn sequentially
                              without exhibiting catastrophic forgetting
                              of the previous knowledge, which is
                              possibly also a consequence of human
                              learning by understanding.<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal">Danko<o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                          <div>
                            <div>
                              <div>
                                <div>
                                  <p class="MsoNormal">Dr. Danko Nikolić<br>
                                    <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttp-3A__www.danko-2Dnikolic.com%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DwaSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj%26s%3DHwOLDw6UCRzU5-FPSceKjtpNm7C6sZQU5kuGAMVbPaI%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=Kq%2Fs2VyICbR%2FnZ0Td0zRANqaHcwCpom4%2FkXoF7Hq69M%3D&reserved=0"
                                      target="_blank"
                                      moz-do-not-send="true">www.danko-nikolic.com</a><br>
                                    <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__www.linkedin.com_in_danko-2Dnikolic_%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DwaSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj%26s%3Db70c8lokmxM3Kz66OfMIM4pROgAhTJOAlp205vOmCQ8%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=zJmnBM8VSwQnU9U412msEPpTxdmj3OTZmHFwUoyYz9s%3D&reserved=0"
                                      target="_blank"
                                      moz-do-not-send="true">https://www.linkedin.com/in/danko-nikolic/</a>
                                    <o:p></o:p></p>
                                  <div>
                                    <p class="MsoNormal">--- A progress
                                      usually starts with an insight ---<o:p></o:p></p>
                                  </div>
                                </div>
                              </div>
                            </div>
                            <p class="MsoNormal"> <o:p></o:p></p>
                          </div>
                        </div>
                        <div id="DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2">
                          <p class="MsoNormal"> <o:p></o:p></p>
                          <table class="MsoNormalTable"
                            style="border:none;border-top:solid #D3D4DE
                            1.0pt" cellspacing="3" cellpadding="0"
                            border="1">
                            <tbody>
                              <tr>
                                <td
                                  style="width:41.25pt;border:none;padding:9.75pt
                                  .75pt .75pt .75pt" width="55">
                                  <p class="MsoNormal"><a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__www.avast.com_sig-2Demail-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dlink-26utm-5Fcampaign-3Dsig-2Demail-26utm-5Fcontent-3Dwebmail%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DwaSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj%26s%3DAo9QQWtO62go0hx1tb3NU6xw2FNBadjj8q64-hl5Sx4%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=Oeg28bYrH1y3a80AXcK2hmPWzSPvuidnsXah%2Bsal%2F3k%3D&reserved=0"
                                      target="_blank"
                                      moz-do-not-send="true"><span
                                        style="text-decoration:none"><img
style="width:.4833in;height:.3in" id="_x0000_i1025"
src="https://ipmcdn.avast.com/images/icons/icon-envelope-tick-round-orange-animated-no-repeat-v1.gif"
                                          moz-do-not-send="true"
                                          width="46" height="29"
                                          border="0"></span></a><o:p></o:p></p>
                                </td>
                                <td
                                  style="width:352.5pt;border:none;padding:9.0pt
                                  .75pt .75pt .75pt" width="470">
                                  <p class="MsoNormal"
                                    style="line-height:13.5pt"><span
style="font-size:10.0pt;font-family:"Arial",sans-serif;color:#41424E">Virus-free.
                                      <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__www.avast.com_sig-2Demail-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dlink-26utm-5Fcampaign-3Dsig-2Demail-26utm-5Fcontent-3Dwebmail%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DwaSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj%26s%3DAo9QQWtO62go0hx1tb3NU6xw2FNBadjj8q64-hl5Sx4%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=Oeg28bYrH1y3a80AXcK2hmPWzSPvuidnsXah%2Bsal%2F3k%3D&reserved=0"
                                        target="_blank"
                                        moz-do-not-send="true">
                                        <span style="color:#4453EA">www.avast.com</span></a>
                                    </span><o:p></o:p></p>
                                </td>
                              </tr>
                            </tbody>
                          </table>
                        </div>
                        <p class="MsoNormal"> <o:p></o:p></p>
                        <div>
                          <div>
                            <p class="MsoNormal">On Thu, Feb 3, 2022 at
                              9:55 AM Asim Roy <<a
                                href="mailto:ASIM.ROY@asu.edu"
                                moz-do-not-send="true">ASIM.ROY@asu.edu</a>>
                              wrote:<o:p></o:p></p>
                          </div>
                          <blockquote
                            style="border:none;border-left:solid #CCCCCC
                            1.0pt;padding:0in 0in 0in
6.0pt;margin-left:4.8pt;margin-top:5.0pt;margin-right:0in;margin-bottom:5.0pt">
                            <div>
                              <div>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Without
                                  getting into the specific dispute
                                  between Gary and Geoff, I think with
                                  approaches similar to GLOM, we are
                                  finally headed in the right direction.
                                  There’s plenty of neurophysiological
                                  evidence for single-cell abstractions
                                  and multisensory neurons in the brain,
                                  which one might claim correspond to
                                  symbols. And I think we can finally
                                  reconcile the decades old dispute
                                  between Symbolic AI and Connectionism.<o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
                                    style="background:yellow">GARY:
                                    (Your GLOM, which as you know I
                                    praised publicly, is in many ways an
                                    effort to wind up with encodings
                                    that effectively serve as symbols in
                                    exactly that way, guaranteed to
                                    serve as consistent representations
                                    of specific concepts.)</span><o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
                                    style="background:yellow">GARY: I
                                    have
                                    <i>never</i> called for dismissal of
                                    neural networks, but rather for some
                                    hybrid between the two (as you
                                    yourself contemplated in 1991); the
                                    point of the 2001 book was to
                                    characterize exactly where
                                    multilayer perceptrons succeeded and
                                    broke down, and where symbols could
                                    complement them.</span><o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Asim
                                  Roy<o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Professor,
                                  Information Systems<o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Arizona
                                  State University<o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__lifeboat.com_ex_bios.asim.roy%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DwaSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj%26s%3DoDRJmXX22O8NcfqyLjyu4Ajmt8pcHWquTxYjeWahfuw%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=WcjLo5CKdZHVYMrODcH4m1t8P1D1vRP9Kk6r0HMEp5o%3D&reserved=0"
                                    target="_blank"
                                    moz-do-not-send="true">Lifeboat
                                    Foundation Bios: Professor Asim Roy</a><o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__isearch.asu.edu_profile_9973%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DwaSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj%26s%3DjCesWT7oGgX76_y7PFh4cCIQ-Ife-esGblJyrBiDlro%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=VcsruBhcuGiqE8PWjadj90uFUnsUgXMsUTHuxmFfxKk%3D&reserved=0"
                                    target="_blank"
                                    moz-do-not-send="true">Asim Roy |
                                    iSearch (asu.edu)</a><o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                <div>
                                  <div
                                    style="border:none;border-top:solid
                                    #E1E1E1 1.0pt;padding:3.0pt 0in 0in
                                    0in">
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><b>From:</b>
                                      Connectionists <<a
                                        href="mailto:connectionists-bounces@mailman.srv.cs.cmu.edu"
                                        target="_blank"
                                        moz-do-not-send="true">connectionists-bounces@mailman.srv.cs.cmu.edu</a>>
                                      <b>On Behalf Of </b>Gary Marcus<br>
                                      <b>Sent:</b> Wednesday, February
                                      2, 2022 1:26 PM<br>
                                      <b>To:</b> Geoffrey Hinton <<a
href="mailto:geoffrey.hinton@gmail.com" target="_blank"
                                        moz-do-not-send="true">geoffrey.hinton@gmail.com</a>><br>
                                      <b>Cc:</b> AIhub <<a
                                        href="mailto:aihuborg@gmail.com"
                                        target="_blank"
                                        moz-do-not-send="true">aihuborg@gmail.com</a>>;
                                      <a
                                        href="mailto:connectionists@mailman.srv.cs.cmu.edu"
                                        target="_blank"
                                        moz-do-not-send="true">connectionists@mailman.srv.cs.cmu.edu</a><br>
                                      <b>Subject:</b> Re:
                                      Connectionists: Stephen Hanson in
                                      conversation with Geoff Hinton<o:p></o:p></p>
                                  </div>
                                </div>
                                <p class="MsoNormal"
                                  style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                <div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Dear
                                      Geoff, and interested others,<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">What,
                                      for example, would you make of a
                                      system that often drew the
                                      red-hatted hamster you requested,
                                      and perhaps a fifth of the time
                                      gave you utter nonsense?  Or say
                                      one that you trained to create
                                      birds but sometimes output stuff
                                      like this:<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><image001.png><o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">One
                                      could <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">a.
                                      avert one’s eyes and deem the
                                      anomalous outputs irrelevant<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">or<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">b.
                                      wonder if it might be possible
                                      that sometimes the system gets the
                                      right answer for the wrong reasons
                                      (eg partial historical
                                      contingency), and wonder whether
                                      another approach might be
                                      indicated.<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Benchmarks
                                      are harder than they look; most of
                                      the field has come to recognize
                                      that. The Turing Test has turned
                                      out to be a lousy measure of
                                      intelligence, easily gamed. It has
                                      turned out empirically that the
                                      Winograd Schema Challenge did not
                                      measure common sense as well as
                                      Hector might have thought. (As it
                                      happens, I am a minor coauthor of
                                      a very recent review on this very
                                      topic: <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.com%2Fv3%2F__https%3A%2Farxiv.org%2Fabs%2F2201.02387__%3B!!IKRxdwAv5BmarQ!INA0AMmG3iD1B8MDtLfjWCwcBjxO-e-eM2Ci9KEO_XYOiIEgiywK-G_8j6L3bHA%24&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=GW0nhKsLRzdzUwXga9mgmMHOII98elKIMxX5%2FJ7ZNnU%3D&reserved=0"
                                        target="_blank"
                                        moz-do-not-send="true">https://arxiv.org/abs/2201.02387</a>)
                                      But its conquest in no way means
                                      machines now have common sense;
                                      many people from many different
                                      perspectives recognize that
                                      (including, e.g., Yann LeCun, who
                                      generally tends to be more aligned
                                      with you than with me).<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">So:
                                      on the goalpost of the Winograd
                                      schema, I was wrong, and you can
                                      quote me; but what you said about
                                      me and machine translation remains
                                      your invention, and it is
                                      inexcusable that you simply
                                      ignored my 2019 clarification. On
                                      the essential goal of trying to
                                      reach meaning and understanding, I
                                      remain unmoved; the problem
                                      remains unsolved. <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">All
                                      of the problems LLMs have with
                                      coherence, reliability,
                                      truthfulness, misinformation, etc
                                      stand witness to that fact. (Their
                                      persistent inability to filter out
                                      toxic and insulting remarks stems
                                      from the same.) I am hardly the
                                      only person in the field to see
                                      that progress on any given
                                      benchmark does not inherently mean
                                      that the deep underlying problems
                                      have solved. You, yourself, in
                                      fact, have occasionally made that
                                      point. <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">With
                                      respect to embeddings: Embeddings
                                      are very good for natural language
                                      <i>processing</i>; but NLP is not
                                      the same as NL<i>U</i> – when it
                                      comes to <i>understanding</i>,
                                      their worth is still an open
                                      question. Perhaps they will turn
                                      out to be necessary; they clearly
                                      aren’t sufficient. In their
                                      extreme, they might even collapse
                                      into being symbols, in the sense
                                      of uniquely identifiable
                                      encodings, akin to the ASCII code,
                                      in which a specific set of numbers
                                      stands for a specific word or
                                      concept. (Wouldn’t that be
                                      ironic?)<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">(Your
                                      GLOM, which as you know I praised
                                      publicly, is in many ways an
                                      effort to wind up with encodings
                                      that effectively serve as symbols
                                      in exactly that way, guaranteed to
                                      serve as consistent
                                      representations of specific
                                      concepts.)<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Notably
                                      absent from your email is any kind
                                      of apology for misrepresenting my
                                      position. It’s fine to say that
                                      “many people thirty years ago once
                                      thought X” and another to say
                                      “Gary Marcus said X in 2015”, when
                                      I didn’t. I have consistently felt
                                      throughout our interactions that
                                      you have mistaken me for Zenon
                                      Pylyshyn; indeed, you once (at
                                      NeurIPS 2014) apologized to me for
                                      having made that error. I am still
                                      not he. <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Which
                                      maybe connects to the last point;
                                      if you read my work, you would see
                                      thirty years of arguments
                                      <i>for</i> neural networks, just
                                      not in the way that you want them
                                      to exist. I have ALWAYS argued
                                      that there is a role for them;
                                       characterizing me as a person
                                      “strongly opposed to neural
                                      networks” misses the whole point
                                      of my 2001 book, which was
                                      subtitled “Integrating
                                      Connectionism and Cognitive
                                      Science.”<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">In
                                      the last two decades or so you
                                      have insisted (for reasons you
                                      have never fully clarified, so far
                                      as I know) on abandoning
                                      symbol-manipulation, but the
                                      reverse is not the case: I have <i>never</i>
                                      called for dismissal of neural
                                      networks, but rather for some
                                      hybrid between the two (as you
                                      yourself contemplated in 1991);
                                      the point of the 2001 book was to
                                      characterize exactly where
                                      multilayer perceptrons succeeded
                                      and broke down, and where symbols
                                      could complement them. It’s a
                                      rhetorical trick (which is what
                                      the previous thread was about) to
                                      pretend otherwise.<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Gary<o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <p class="MsoNormal"
                                      style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                  </div>
                                  <div>
                                    <blockquote
                                      style="margin-top:5.0pt;margin-bottom:5.0pt">
                                      <p class="MsoNormal"
                                        style="mso-margin-top-alt:auto;margin-bottom:12.0pt">On
                                        Feb 2, 2022, at 11:22, Geoffrey
                                        Hinton <<a
                                          href="mailto:geoffrey.hinton@gmail.com"
                                          target="_blank"
                                          moz-do-not-send="true">geoffrey.hinton@gmail.com</a>>
                                        wrote:<o:p></o:p></p>
                                    </blockquote>
                                  </div>
                                  <blockquote
                                    style="margin-top:5.0pt;margin-bottom:5.0pt">
                                    <div>
                                      <p class="MsoNormal"
                                        style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><o:p></o:p></p>
                                      <div>
                                        <p class="MsoNormal"
                                          style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Embeddings
                                          are just vectors of soft
                                          feature detectors and they are
                                          very good for NLP.  The quote
                                          on my webpage from Gary's 2015
                                          chapter implies the opposite.<o:p></o:p></p>
                                        <div>
                                          <p class="MsoNormal"
                                            style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                        </div>
                                        <div>
                                          <p class="MsoNormal"
                                            style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">A
                                            few decades ago, everyone I
                                            knew then would have agreed
                                            that the ability to
                                            translate a sentence into
                                            many different languages was
                                            strong evidence that you
                                            understood it.<o:p></o:p></p>
                                        </div>
                                      </div>
                                    </div>
                                  </blockquote>
                                  <p class="MsoNormal"
                                    style="mso-margin-top-alt:auto;margin-bottom:12.0pt"> <o:p></o:p></p>
                                  <blockquote
                                    style="margin-top:5.0pt;margin-bottom:5.0pt">
                                    <div>
                                      <div>
                                        <div>
                                          <p class="MsoNormal"
                                            style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">But
                                            once neural networks could
                                            do that, their critics moved
                                            the goalposts. An exception
                                            is Hector Levesque who
                                            defined the goalposts more
                                            sharply by saying that the
                                            ability to get pronoun
                                            references correct in
                                            Winograd sentences is a
                                            crucial test. Neural nets
                                            are improving at that but
                                            still have some way to go.
                                            Will Gary agree that when
                                            they can get pronoun
                                            references correct in
                                            Winograd sentences they
                                            really do understand? Or
                                            does he want to reserve the
                                            right to weasel out of that
                                            too?<o:p></o:p></p>
                                        </div>
                                        <div>
                                          <p class="MsoNormal"
                                            style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                        </div>
                                        <div>
                                          <p class="MsoNormal"
                                            style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Some
                                            people, like Gary, appear to
                                            be strongly opposed to
                                            neural networks because they
                                            do not fit their
                                            preconceived notions of how
                                            the mind should work.<o:p></o:p></p>
                                        </div>
                                        <div>
                                          <div>
                                            <p class="MsoNormal"
                                              style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">I
                                              believe that any
                                              reasonable person would
                                              admit that if you ask a
                                              neural net to draw a
                                              picture of a hamster
                                              wearing a red hat and it
                                              draws such a picture, it
                                              understood the request.<o:p></o:p></p>
                                          </div>
                                          <div>
                                            <p class="MsoNormal"
                                              style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                          </div>
                                          <div>
                                            <p class="MsoNormal"
                                              style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Geoff<o:p></o:p></p>
                                          </div>
                                          <div>
                                            <p class="MsoNormal"
                                              style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                          </div>
                                          <div>
                                            <p class="MsoNormal"
                                              style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                          </div>
                                          <div>
                                            <p class="MsoNormal"
                                              style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                            <div>
                                              <p class="MsoNormal"
                                                style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                            </div>
                                          </div>
                                        </div>
                                      </div>
                                      <p class="MsoNormal"
                                        style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                      <div>
                                        <div>
                                          <p class="MsoNormal"
                                            style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">On
                                            Wed, Feb 2, 2022 at 1:38 PM
                                            Gary Marcus <<a
                                              href="mailto:gary.marcus@nyu.edu"
                                              target="_blank"
                                              moz-do-not-send="true">gary.marcus@nyu.edu</a>>
                                            wrote:<o:p></o:p></p>
                                        </div>
                                        <blockquote
                                          style="border:none;border-left:solid
                                          #CCCCCC 1.0pt;padding:0in 0in
                                          0in
6.0pt;margin-left:4.8pt;margin-top:5.0pt;margin-right:0in;margin-bottom:5.0pt">
                                          <div>
                                            <div>
                                              <div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Dear AI Hub,
                                                      cc: Steven Hanson
                                                      and Geoffrey
                                                      Hinton, and the
                                                      larger neural
                                                      network community,</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">There has
                                                      been a lot of
                                                      recent discussion
                                                      on this list about
                                                      framing and
                                                      scientific
                                                      integrity. Often
                                                      the first step in
                                                      restructuring
                                                      narratives is to
                                                      bully and
                                                      dehumanize
                                                      critics. The
                                                      second is to
                                                      misrepresent their
                                                      position. People
                                                      in positions of
                                                      power are
                                                      sometimes tempted
                                                      to do this.</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">The
                                                      Hinton-Hanson
                                                      interview that you
                                                      just published is
                                                      a real-time
                                                      example of just
                                                      that. It opens
                                                      with a needless
                                                      and largely
                                                      content-free
                                                      personal attack on
                                                      a single scholar
                                                      (me), with the
                                                      explicit intention
                                                      of discrediting
                                                      that person.
                                                      Worse, the only
                                                      substantive thing
                                                      it says is false.</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Hinton says
                                                      “In 2015 he
                                                      [Marcus] made a
                                                      prediction that
                                                      computers wouldn’t
                                                      be able to do
                                                      machine
                                                      translation.”</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">I never said
                                                      any such thing. </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">What I
                                                      predicted, rather,
                                                      was that
                                                      multilayer
                                                      perceptrons, as
                                                      they existed then,
                                                      would not (on
                                                      their own, absent
                                                      other mechanisms) </span><i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleItalicBody">understand</span></i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody"> language.
                                                      Seven years later,
                                                      they still
                                                      haven’t, except in
                                                      the most
                                                      superficial way. 
                                                       </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">I made no
                                                      comment whatsoever
                                                      about machine
                                                      translation, which
                                                      I view as a
                                                      separate problem,
                                                      solvable to a
                                                      certain degree by
                                                      correspondance
                                                      without
                                                      semantics. </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">I
                                                      specifically tried
                                                      to clarify
                                                      Hinton’s confusion
                                                      in 2019, but,
                                                      disappointingly,
                                                      he has continued
                                                      to purvey
                                                      misinformation
                                                      despite that
                                                      clarification.
                                                      Here is what I
                                                      wrote privately to
                                                      him then, which
                                                      should have put
                                                      the matter to
                                                      rest:</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div
                                                  style="margin-left:27.0pt;font-stretch:normal">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">You have
                                                      taken a single out
                                                      of context quote
                                                      [from 2015] and
                                                      misrepresented it.
                                                      The quote, which
                                                      you have
                                                      prominently
                                                      displayed at the
                                                      bottom on your own
                                                      web page, says:</span><o:p></o:p></p>
                                                </div>
                                                <div
                                                  style="margin-left:27.0pt;font-stretch:normal;min-height:22.9px">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div
                                                  style="margin-left:.75in;font-stretch:normal">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Hierarchies
                                                      of features are
                                                      less suited to
                                                      challenges such as
                                                      language,
                                                      inference, and
                                                      high-level
                                                      planning. For
                                                      example, as Noam
                                                      Chomsky famously
                                                      pointed out,
                                                      language is filled
                                                      with sentences you
                                                      haven't seen
                                                      before. Pure
                                                      classifier systems
                                                      don't know what to
                                                      do with such
                                                      sentences. The
                                                      talent of feature
                                                      detectors --
                                                      in  identifying
                                                      which member of
                                                      some category
                                                      something belongs
                                                      to -- doesn't
                                                      translate into
                                                      understanding
                                                      novel  sentences,
                                                      in which each
                                                      sentence has its
                                                      own unique
                                                      meaning. </span><o:p></o:p></p>
                                                </div>
                                                <div
                                                  style="margin-left:27.0pt;font-stretch:normal;min-height:22.9px">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div
                                                  style="margin-left:27.0pt;font-stretch:normal">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">It does </span><i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleItalicBody">not</span></i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody"> say "neural
                                                      nets would not be
                                                      able to deal with
                                                      novel sentences";
                                                      it says that
                                                      hierachies of
                                                      features detectors
                                                      (on their own, if
                                                      you read the
                                                      context of the
                                                      essay) would have
                                                      trouble </span><i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleItalicBody">understanding </span></i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">novel sentences.
                                                       </span><o:p></o:p></p>
                                                </div>
                                                <div
                                                  style="margin-left:27.0pt;font-stretch:normal;min-height:22.9px">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div
                                                  style="margin-left:27.0pt;font-stretch:normal">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Google
                                                      Translate does yet
                                                      not </span><i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleItalicBody">understand</span></i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody"> the content
                                                      of the sentences
                                                      is translates. It
                                                      cannot reliably
                                                      answer questions
                                                      about who did what
                                                      to whom, or why,
                                                      it cannot infer
                                                      the order of the
                                                      events in
                                                      paragraphs, it
                                                      can't determine
                                                      the internal
                                                      consistency of
                                                      those events, and
                                                      so forth.</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Since then, a
                                                      number of
                                                      scholars, such as
                                                      the the
                                                      computational
                                                      linguist Emily
                                                      Bender, have made
                                                      similar points,
                                                      and indeed current
                                                      LLM difficulties
                                                      with
                                                      misinformation,
                                                      incoherence and
                                                      fabrication all
                                                      follow from these
                                                      concerns. Quoting
                                                      from Bender’s
                                                      prizewinning 2020
                                                      ACL article on the
                                                      matter with
                                                      Alexander Koller, <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__aclanthology.org_2020.acl-2Dmain.463.pdf%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DxnFSVUARkfmiXtiTP_uXfFKv4uNEGgEeTluRFR7dnUpay2BM5EiLz-XYCkBNJLlL%26s%3DK-Vl6vSvzuYtRMi-s4j7mzPkNRTb-I6Zmf7rbuKEBpk%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=B4j1RHz%2B9YHApNdcbTZytuS7CdLqhhELB%2BcZTt48xms%3D&reserved=0"
                                                        target="_blank"
moz-do-not-send="true">https://aclanthology.org/2020.acl-main.463.pdf</a>,
                                                      also emphasizing
                                                      issues of
                                                      understanding and
                                                      meaning:</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div
                                                  style="margin-left:27.0pt;font-stretch:normal">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><i><span
style="font-size:13.0pt;font-family:UICTFontTextStyleItalicBody">The
                                                        success of the
                                                        large neural
                                                        language models
                                                        on many NLP
                                                        tasks is
                                                        exciting.
                                                        However, we find
                                                        that these
                                                        successes
                                                        sometimes lead
                                                        to hype in which
                                                        these models are
                                                        being described
                                                        as
                                                        “understanding”
                                                        language or
                                                        capturing
                                                        “meaning”. In
                                                        this position
                                                        paper, we argue
                                                        that a system
                                                        trained only on
                                                        form has a
                                                        priori no way to
                                                        learn meaning.
                                                        .. a clear
                                                        understanding of
                                                        the distinction
                                                        between form and
                                                        meaning will
                                                        help guide the
                                                        field towards
                                                        better science
                                                        around natural
                                                        language
                                                        understanding. </span></i><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Her later
                                                      article with Gebru
                                                      on language models
                                                      “stochastic
                                                      parrots” is in
                                                      some ways an
                                                      extension of this
                                                      point; machine
                                                      translation
                                                      requires mimicry,
                                                      true understanding
                                                      (which is what I
                                                      was discussing in
                                                      2015) requires
                                                      something deeper
                                                      than that. </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Hinton’s
                                                      intellectual error
                                                      here is in
                                                      equating machine
                                                      translation with
                                                      the deeper
                                                      comprehension that
                                                      robust natural
                                                      language
                                                      understanding will
                                                      require; as Bender
                                                      and Koller
                                                      observed, the two
                                                      appear not to be
                                                      the same. (There
                                                      is a longer
                                                      discussion of the
                                                      relation between
                                                      language
                                                      understanding and
                                                      machine
                                                      translation, and
                                                      why the latter has
                                                      turned out to be
                                                      more approachable
                                                      than the former,
                                                      in my 2019 book
                                                      with Ernest
                                                      Davis).</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">More broadly,
                                                      Hinton’s ongoing
                                                      dismissiveness of
                                                      research from
                                                      perspectives other
                                                      than his own (e.g.
                                                      linguistics) have
                                                      done the field a
                                                      disservice. </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">As Herb Simon
                                                      once observed,
                                                      science does not
                                                      have to be
                                                      zero-sum.</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt"> </span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Sincerely,</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Gary Marcus</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">Professor
                                                      Emeritus</span><o:p></o:p></p>
                                                </div>
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
style="font-size:13.0pt;font-family:UICTFontTextStyleBody">New York
                                                      University</span><o:p></o:p></p>
                                                </div>
                                              </div>
                                              <div>
                                                <p class="MsoNormal"
                                                  style="mso-margin-top-alt:auto;margin-bottom:12.0pt"> <o:p></o:p></p>
                                                <blockquote
                                                  style="margin-top:5.0pt;margin-bottom:5.0pt">
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;margin-bottom:12.0pt">On
                                                    Feb 2, 2022, at
                                                    06:12, AIhub <<a
href="mailto:aihuborg@gmail.com" target="_blank" moz-do-not-send="true">aihuborg@gmail.com</a>>
                                                    wrote:<o:p></o:p></p>
                                                </blockquote>
                                              </div>
                                              <blockquote
                                                style="margin-top:5.0pt;margin-bottom:5.0pt">
                                                <div>
                                                  <p class="MsoNormal"
                                                    style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><o:p></o:p></p>
                                                  <div>
                                                    <div>
                                                      <p
                                                        class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">Stephen
                                                        Hanson in
                                                        conversation
                                                        with Geoff
                                                        Hinton<o:p></o:p></p>
                                                    </div>
                                                    <div>
                                                      <p
                                                        class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                                    </div>
                                                    <div>
                                                      <p
                                                        class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">In the latest
                                                        episode of this
                                                        video series for
                                                        <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttp-3A__AIhub.org%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DxnFSVUARkfmiXtiTP_uXfFKv4uNEGgEeTluRFR7dnUpay2BM5EiLz-XYCkBNJLlL%26s%3DeOtzMh8ILIH5EF7K20Ks4Fr27XfNV_F24bkj-SPk-2A%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=Wpm7HzprwI4R6J2C4qBtBgEs7qwvFWYFqA4yxEMx6aU%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          AIhub.org</a>,
                                                        Stephen Hanson
                                                        talks to  Geoff
                                                        Hinton about
                                                        neural networks,
                                                        backpropagation,
overparameterization, digit recognition, voxel cells, syntax and
                                                        semantics,
                                                        Winograd
                                                        sentences, and
                                                        more.<o:p></o:p></p>
                                                      <div>
                                                        <p
                                                          class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                                      </div>
                                                      <div>
                                                        <p
                                                          class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">You can watch
                                                          the
                                                          discussion,
                                                          and read the
                                                          transcript,
                                                          here:<br
                                                          clear="all">
                                                          <o:p></o:p></p>
                                                        <div>
                                                          <p
                                                          class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__aihub.org_2022_02_02_what-2Dis-2Dai-2Dstephen-2Dhanson-2Din-2Dconversation-2Dwith-2Dgeoff-2Dhinton_%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3DOY_RYGrfxOqV7XeNJDHuzE--aEtmNRaEyQ0VJkqFCWw%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=PpSkqWx1E%2BKzNyAmsGPuwtMDBcsKvaFCcPIrygyAexs%3D&reserved=0"
target="_blank" moz-do-not-send="true">https://aihub.org/2022/02/02/what-is-ai-stephen-hanson-in-conversation-with-geoff-hinton/</a><o:p></o:p></p>
                                                        </div>
                                                        <div>
                                                          <p
                                                          class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"> <o:p></o:p></p>
                                                        </div>
                                                        <div>
                                                          <p
                                                          class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
                                                          style="font-family:"Arial",sans-serif">About
                                                          AIhub: </span><o:p></o:p></p>
                                                        </div>
                                                        <div>
                                                          <p
                                                          class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
                                                          style="font-family:"Arial",sans-serif">AIhub
                                                          is a
                                                          non-profit
                                                          dedicated to
                                                          connecting the
                                                          AI community
                                                          to the public
                                                          by providing
                                                          free,
                                                          high-quality
                                                          information
                                                          through
                                                          <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttp-3A__AIhub.org%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DxnFSVUARkfmiXtiTP_uXfFKv4uNEGgEeTluRFR7dnUpay2BM5EiLz-XYCkBNJLlL%26s%3DeOtzMh8ILIH5EF7K20Ks4Fr27XfNV_F24bkj-SPk-2A%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=Wpm7HzprwI4R6J2C4qBtBgEs7qwvFWYFqA4yxEMx6aU%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          AIhub.org</a>
                                                          (<a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__aihub.org_%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3DIKFanqeMi73gOiS7yD-X_vRx_OqDAwv1Il5psrxnhIA%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=N3f9ErvzAq8MlpKOxpwLrbhONI6beg4Zsr7N%2BsHSLYU%3D&reserved=0"
target="_blank" moz-do-not-send="true">https://aihub.org/</a>). We help
                                                          researchers
                                                          publish the
                                                          latest AI
                                                          news,
                                                          summaries of
                                                          their work,
                                                          opinion
                                                          pieces,
                                                          tutorials and
                                                          more.  We are
                                                          supported by
                                                          many leading
                                                          scientific
                                                          organizations
                                                          in AI, namely
                                                          <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__aaai.org_%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3DwBvjOWTzEkbfFAGNj9wOaiJlXMODmHNcoWO5JYHugS0%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=S40AqsxKJjT%2Bwru92O45JPkdDVRn%2FMN19TBJdw0Hyek%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          AAAI</a>, <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__neurips.cc_%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3D3-lOHXyu8171pT_UE9hYWwK6ft4I-cvYkuX7shC00w0%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372759823%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=bI9N4NQKOKzPiwiWVtZDTqANuC9bVJcTSRfVwAi1hWk%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          NeurIPS</a>, <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__icml.cc_imls_%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3DJJyjwIpPy9gtKrZzBMbW3sRMh3P3Kcw-SvtxG35EiP0%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372916012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=liSRQikObkozAApIsD8bsxfUIIxqyfsIeP01TXDPOkM%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          ICML</a>, <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__www.journals.elsevier.com_artificial-2Dintelligence%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3DeWrRCVWlcbySaH3XgacPpi0iR0-NDQYCLJ1x5yyMr8U%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372916012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=vBzk1t7z%2BkTnDpnbpoPtOcebMr27SgB2gPqcVW%2FbSKM%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          AIJ</a>/<a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__www.journals.elsevier.com_artificial-2Dintelligence%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3DeWrRCVWlcbySaH3XgacPpi0iR0-NDQYCLJ1x5yyMr8U%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372916012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=vBzk1t7z%2BkTnDpnbpoPtOcebMr27SgB2gPqcVW%2FbSKM%3D&reserved=0"
target="_blank" moz-do-not-send="true">IJCAI</a>,
                                                          <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttp-3A__sigai.acm.org_%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3D7rC6MJFaMqOms10EYDQwfnmX-zuVNhu9fz8cwUwiLGQ%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372916012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=DCepgdqWy7qgCR80FbdA7BuKvC9ztFmQMgoRL%2FiZYD0%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          ACM SIGAI</a>,
                                                          EurAI/AICOMM,
                                                          <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__claire-2Dai.org_%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3D66ZofDIhuDba6Fb0LhlMGD3XbBhU7ez7dc3HD5-pXec%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372916012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=mnpEa%2FEHrPDFhh04S3AjyZpnV%2FP%2Ba8CSAyDORR6wuHo%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          CLAIRE</a> and
                                                          <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__www.robocup.org__%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3Dyl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1%26s%3DbBI6GRq--MHLpIIahwoVN8iyXXc7JAeH3kegNKcFJc0%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372916012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=X9%2F5TO74rSEv2o7UTpgFQBP7e4pgPRZ18Eczn0zalEA%3D&reserved=0"
target="_blank" moz-do-not-send="true">
                                                          RoboCup</a>.</span><o:p></o:p></p>
                                                        </div>
                                                        <div>
                                                          <p
                                                          class="MsoNormal"
style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto"><span
                                                          style="font-family:"Arial",sans-serif">Twitter:
                                                          @aihuborg</span><o:p></o:p></p>
                                                        </div>
                                                      </div>
                                                    </div>
                                                  </div>
                                                </div>
                                              </blockquote>
                                            </div>
                                          </div>
                                        </blockquote>
                                      </div>
                                    </div>
                                  </blockquote>
                                </div>
                              </div>
                            </div>
                          </blockquote>
                        </div>
                        <div id="DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2">
                          <p class="MsoNormal"> <o:p></o:p></p>
                          <table class="MsoNormalTable"
                            style="border:none;border-top:solid #D3D4DE
                            1.0pt" cellspacing="3" cellpadding="0"
                            border="1">
                            <tbody>
                              <tr>
                                <td
                                  style="width:41.25pt;border:none;padding:9.75pt
                                  .75pt .75pt .75pt" width="55">
                                  <p class="MsoNormal"><a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__www.avast.com_sig-2Demail-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dlink-26utm-5Fcampaign-3Dsig-2Demail-26utm-5Fcontent-3Dwebmail%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DwaSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj%26s%3DAo9QQWtO62go0hx1tb3NU6xw2FNBadjj8q64-hl5Sx4%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372916012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=cggC9NXZbatKaQXXuhTozGHI5UKNpSVpcSwYtLxcJvc%3D&reserved=0"
                                      target="_blank"
                                      moz-do-not-send="true"><span
                                        style="text-decoration:none"><img
style="width:.4833in;height:.3in" id="_x0000_i1026"
src="https://ipmcdn.avast.com/images/icons/icon-envelope-tick-round-orange-animated-no-repeat-v1.gif"
                                          moz-do-not-send="true"
                                          width="46" height="29"
                                          border="0"></span></a><o:p></o:p></p>
                                </td>
                                <td
                                  style="width:352.5pt;border:none;padding:9.0pt
                                  .75pt .75pt .75pt" width="470">
                                  <p class="MsoNormal"
                                    style="line-height:13.5pt"><span
style="font-size:10.0pt;font-family:"Arial",sans-serif;color:#41424E">Virus-free.
                                      <a
href="https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__www.avast.com_sig-2Demail-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dlink-26utm-5Fcampaign-3Dsig-2Demail-26utm-5Fcontent-3Dwebmail%26d%3DDwMFaQ%26c%3DslrrB7dE8n7gBJbeO0g-IQ%26r%3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ%26m%3DwaSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj%26s%3DAo9QQWtO62go0hx1tb3NU6xw2FNBadjj8q64-hl5Sx4%26e%3D&data=04%7C01%7Ctgd%40oregonstate.edu%7Cfe236f1b41ad46b207f208d9e7d5fbe7%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C637795730372916012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=cggC9NXZbatKaQXXuhTozGHI5UKNpSVpcSwYtLxcJvc%3D&reserved=0"
                                        target="_blank"
                                        moz-do-not-send="true">
                                        <span style="color:#4453EA">www.avast.com</span></a>
                                    </span><o:p></o:p></p>
                                </td>
                              </tr>
                            </tbody>
                          </table>
                        </div>
                      </div>
                    </blockquote>
                  </div>
                  <p class="MsoNormal"> <o:p></o:p></p>
                </div>
              </div>
            </div>
          </blockquote>
          <div>
            <p class="MsoNormal">-- <br>
              <img style="width:4.1833in;height:1.025in"
                id="_x0000_i1027"
                src="cid:part40.96573C95.D97F2FF0@rubic.rutgers.edu"
                class="" width="402" height="98" border="0"><o:p></o:p></p>
          </div>
        </div>
      </div>
    </blockquote>
    <div class="moz-signature">-- <br>
      <img src="cid:part41.6107DA9C.A9AD3D47@rubic.rutgers.edu"
        border="0"></div>
  </body>
</html>