<div dir="ltr"><div dir="ltr"><div>Dear Gary and everyone,</div><div><br></div><div>I am continuing the discussion from where we left off a few months ago. Back then, some of us agreed that the problem of understanding remains unsolved.</div><div><br></div><div>As a reminder, the challenge for connectionism was to 1) learn with few examples and 2) apply the knowledge to a broad set of situations.</div><div><br></div><div>I am happy to announce that I have now finished a draft of a paper in which I propose how the brain is able to achieve that. The manuscript requires a bit of patience for two reasons: one is that the reader may be exposed for the first time to certain aspects of brain physiology. The second reason is that it may take some effort to understand the counterintuitive implications of the new ideas (this requires a different way of thinking than what we are used to based on connectionism).</div><div><br></div><div>In short, I am suggesting that instead of the connectionist paradigm, we adopt transient selection of subnetworks. The mechanisms that transiently select brain subnetworks are distributed all over the nervous system and, I argue, are our main machinery for thinking/cognition. The surprising outcome is that neural activation, which was central in connectionism, now plays only a supportive role, while the real 'workers' within the brain are the mechanisms for transient selection of subnetworks.</div><div><br></div><div>I also explain how I think transient selection achieves learning with only a few examples and how the learned knowledge is possible to apply to a broad set of situations.</div><div><br></div><div>The manuscript is made available to everyone and can be downloaded here: <a href="https://bit.ly/3IFs8Ug">https://bit.ly/3IFs8Ug</a></div><div>(I apologize for the neuroscience lingo, which I tried to minimize.)</div><div><br></div><div>It will likely take a wide effort to implement these concepts as an AI technology, provided my ideas do not have a major flaw in the first place. Does anyone see a flaw?</div><div><br></div><div>Thanks.</div><div><br></div><div>Danko</div><div><br></div><br clear="all"><div><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr">Dr. Danko Nikolić<br><a href="http://www.danko-nikolic.com" target="_blank">www.danko-nikolic.com</a><br><a href="https://www.linkedin.com/in/danko-nikolic/" target="_blank">https://www.linkedin.com/in/danko-nikolic/</a><div><br></div></div></div></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, Feb 3, 2022 at 5:25 PM Gary Marcus <<a href="mailto:gary.marcus@nyu.edu">gary.marcus@nyu.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div style="overflow-wrap: break-word;">Dear Danko,<div><br></div><div>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://static.googleusercontent.com/media/research.google.com/en//archive/unsupervised_icml2012.pdf" target="_blank">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. </div><div><br></div><div>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.</div><div><div><br></div><div>Gary<br><div><br><blockquote type="cite"><div>On Feb 3, 2022, at 3:19 AM, Danko Nikolic <<a href="mailto:danko.nikolic@gmail.com" target="_blank">danko.nikolic@gmail.com</a>> wrote:</div><br><div><div dir="ltr">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."<div><br></div><div>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".</div><div>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.</div><div><br></div><div><div>What would happen to the hat if the hamster rolls on its back? (Would the hat fall off?)</div></div><div>What would happen to the red hat when the hamster enters its lair? (Would the hat fall off?)</div><div>What would happen to that hamster when it goes foraging? (Would the red hat have an influence on finding food?)</div><div>What would happen in a situation of being chased by a predator? (Would it be easier for predators to spot the hamster?)</div><div><br></div><div>...and so on.</div><div><br></div><div>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). <br></div><div><br></div><div>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.</div><div><br></div><div>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. </div><div><br></div><div>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.</div><div><br></div><div>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.</div><div><br></div><div>So, for the case of drawing a hamster wearing a red hat, understanding perhaps would have taken place if the following happened before that:</div><div><br></div><div>1) first, the network learned about hamsters (not many examples)</div><div>2) after that the network learned about red hats (outside the context of hamsters and without many examples) </div><div>3) finally the network learned about drawing (outside of the context of hats and hamsters, not many examples)</div><div><br></div><div>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.</div><div><br></div><div>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.</div><div><br></div><div><br></div><div>Danko</div><div><br></div><div> </div><div><br></div><div><br></div><div><br></div><div><br></div><div><div><div dir="ltr"><div dir="ltr">Dr. Danko Nikolić<br><a href="https://urldefense.proofpoint.com/v2/url?u=http-3A__www.danko-2Dnikolic.com&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=waSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj&s=HwOLDw6UCRzU5-FPSceKjtpNm7C6sZQU5kuGAMVbPaI&e=" target="_blank">www.danko-nikolic.com</a><br><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__www.linkedin.com_in_danko-2Dnikolic_&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=waSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj&s=b70c8lokmxM3Kz66OfMIM4pROgAhTJOAlp205vOmCQ8&e=" target="_blank">https://www.linkedin.com/in/danko-nikolic/</a><div>--- A progress usually starts with an insight ---</div></div></div></div><br></div></div><div id="gmail-m_-1177896716680425671DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2"><br>
<table style="border-top:1px solid rgb(211,212,222)">
<tbody><tr>
<td style="width:55px;padding-top:13px"><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__www.avast.com_sig-2Demail-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dlink-26utm-5Fcampaign-3Dsig-2Demail-26utm-5Fcontent-3Dwebmail&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=waSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj&s=Ao9QQWtO62go0hx1tb3NU6xw2FNBadjj8q64-hl5Sx4&e=" target="_blank"><img src="https://ipmcdn.avast.com/images/icons/icon-envelope-tick-round-orange-animated-no-repeat-v1.gif" alt="" width="46" height="29" style="width: 46px; height: 29px;"></a></td>
<td style="width:470px;padding-top:12px;color:rgb(65,66,78);font-size:13px;font-family:Arial,Helvetica,sans-serif;line-height:18px">Virus-free. <a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__www.avast.com_sig-2Demail-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dlink-26utm-5Fcampaign-3Dsig-2Demail-26utm-5Fcontent-3Dwebmail&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=waSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj&s=Ao9QQWtO62go0hx1tb3NU6xw2FNBadjj8q64-hl5Sx4&e=" style="color:rgb(68,83,234)" target="_blank">www.avast.com</a>
</td>
</tr>
</tbody></table><a width="1" height="1"></a></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, Feb 3, 2022 at 9:55 AM Asim Roy <<a href="mailto:ASIM.ROY@asu.edu" target="_blank">ASIM.ROY@asu.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div lang="EN-US">
<div><p class="MsoNormal">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.<u></u><u></u></p><p class="MsoNormal"><u></u> <u></u></p><p class="MsoNormal"><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><u></u><u></u></p><p class="MsoNormal"><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><u></u><u></u></p><p class="MsoNormal"><u></u> <u></u></p><p class="MsoNormal">Asim Roy<u></u><u></u></p><p class="MsoNormal">Professor, Information Systems<u></u><u></u></p><p class="MsoNormal">Arizona State University<u></u><u></u></p><p class="MsoNormal"><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__lifeboat.com_ex_bios.asim.roy&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=waSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj&s=oDRJmXX22O8NcfqyLjyu4Ajmt8pcHWquTxYjeWahfuw&e=" target="_blank">Lifeboat Foundation Bios: Professor Asim Roy</a><u></u><u></u></p><p class="MsoNormal"><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__isearch.asu.edu_profile_9973&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=waSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj&s=jCesWT7oGgX76_y7PFh4cCIQ-Ife-esGblJyrBiDlro&e=" target="_blank">Asim Roy | iSearch (asu.edu)</a><u></u><u></u></p><p class="MsoNormal"><u></u> <u></u></p><p class="MsoNormal"><u></u> <u></u></p>
<div>
<div style="border-right:none;border-bottom:none;border-left:none;border-top:1pt solid rgb(225,225,225);padding:3pt 0in 0in"><p class="MsoNormal"><b>From:</b> Connectionists <<a href="mailto:connectionists-bounces@mailman.srv.cs.cmu.edu" target="_blank">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">geoffrey.hinton@gmail.com</a>><br>
<b>Cc:</b> AIhub <<a href="mailto:aihuborg@gmail.com" target="_blank">aihuborg@gmail.com</a>>; <a href="mailto:connectionists@mailman.srv.cs.cmu.edu" target="_blank">connectionists@mailman.srv.cs.cmu.edu</a><br>
<b>Subject:</b> Re: Connectionists: Stephen Hanson in conversation with Geoff Hinton<u></u><u></u></p>
</div>
</div><p class="MsoNormal"><u></u> <u></u></p>
<div>
<div><p class="MsoNormal">Dear Geoff, and interested others,<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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:<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal"><span id="gmail-m_-1177896716680425671cid:17ebf223d8f4cff311"><image001.png></span><u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">One could <u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">a. avert one’s eyes and deem the anomalous outputs irrelevant<u></u><u></u></p>
</div>
<div><p class="MsoNormal">or<u></u><u></u></p>
</div>
<div><p class="MsoNormal">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.<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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://urldefense.com/v3/__https:/arxiv.org/abs/2201.02387__;!!IKRxdwAv5BmarQ!INA0AMmG3iD1B8MDtLfjWCwcBjxO-e-eM2Ci9KEO_XYOiIEgiywK-G_8j6L3bHA$" target="_blank">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).<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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. <u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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. <u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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?)<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">(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.)<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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. <u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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.”<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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.<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">Gary<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<blockquote style="margin-top:5pt;margin-bottom:5pt"><p class="MsoNormal" style="margin-bottom:12pt">On Feb 2, 2022, at 11:22, Geoffrey Hinton <<a href="mailto:geoffrey.hinton@gmail.com" target="_blank">geoffrey.hinton@gmail.com</a>> wrote:<u></u><u></u></p>
</blockquote>
</div>
<blockquote style="margin-top:5pt;margin-bottom:5pt">
<div><p class="MsoNormal"><u></u><u></u></p>
<div><p class="MsoNormal">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.<u></u><u></u></p>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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.<u></u><u></u></p>
</div>
</div>
</div>
</blockquote><p class="MsoNormal"><br>
<br>
<u></u><u></u></p>
<blockquote style="margin-top:5pt;margin-bottom:5pt">
<div>
<div>
<div><p class="MsoNormal">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?<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">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.<u></u><u></u></p>
</div>
<div>
<div><p class="MsoNormal">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.<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">Geoff<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
</div>
</div>
</div><p class="MsoNormal"><u></u> <u></u></p>
<div>
<div><p class="MsoNormal">On Wed, Feb 2, 2022 at 1:38 PM Gary Marcus <<a href="mailto:gary.marcus@nyu.edu" target="_blank">gary.marcus@nyu.edu</a>> wrote:<u></u><u></u></p>
</div>
<blockquote style="border-top:none;border-right:none;border-bottom:none;border-left:1pt solid rgb(204,204,204);padding:0in 0in 0in 6pt;margin-left:4.8pt;margin-right:0in">
<div>
<div>
<div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">Dear AI Hub, cc: Steven Hanson and Geoffrey Hinton, and the larger neural network community,</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">Hinton says “In 2015 he [Marcus] made a prediction that computers wouldn’t be able to do machine translation.”</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">I never said any such thing. </span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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:13pt;font-family:UICTFontTextStyleItalicBody,serif">understand</span></i><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif"> language.
Seven years later, they still haven’t, except in the most superficial way. </span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div style="margin-left:27pt;font-stretch:normal"><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div style="margin-left:27pt;font-stretch:normal;min-height:22.9px"><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div style="margin-left:0.75in;font-stretch:normal"><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div style="margin-left:27pt;font-stretch:normal;min-height:22.9px"><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div style="margin-left:27pt;font-stretch:normal"><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">It does </span><i><span style="font-size:13pt;font-family:UICTFontTextStyleItalicBody,serif">not</span></i><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif"> 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:13pt;font-family:UICTFontTextStyleItalicBody,serif">understanding </span></i><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">novel sentences.
</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div style="margin-left:27pt;font-stretch:normal;min-height:22.9px"><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div style="margin-left:27pt;font-stretch:normal"><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">Google Translate does yet not </span><i><span style="font-size:13pt;font-family:UICTFontTextStyleItalicBody,serif">understand</span></i><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif"> 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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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://urldefense.proofpoint.com/v2/url?u=https-3A__aclanthology.org_2020.acl-2Dmain.463.pdf&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=xnFSVUARkfmiXtiTP_uXfFKv4uNEGgEeTluRFR7dnUpay2BM5EiLz-XYCkBNJLlL&s=K-Vl6vSvzuYtRMi-s4j7mzPkNRTb-I6Zmf7rbuKEBpk&e=" target="_blank">https://aclanthology.org/2020.acl-main.463.pdf</a>,
also emphasizing issues of understanding and meaning:</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div style="margin-left:27pt;font-stretch:normal"><p class="MsoNormal"><i><span style="font-size:13pt;font-family:UICTFontTextStyleItalicBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">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><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">More broadly, Hinton’s ongoing dismissiveness of research from perspectives other than his own (e.g. linguistics) have done the field a disservice. </span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">As Herb Simon once observed, science does not have to be zero-sum.</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt"><u></u> <u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">Sincerely,</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">Gary Marcus</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">Professor Emeritus</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
<div><p class="MsoNormal"><span style="font-size:13pt;font-family:UICTFontTextStyleBody,serif">New York University</span><span style="font-size:13pt"><u></u><u></u></span></p>
</div>
</div>
<div><p class="MsoNormal"><br>
<br>
<u></u><u></u></p>
<blockquote style="margin-top:5pt;margin-bottom:5pt"><p class="MsoNormal" style="margin-bottom:12pt">On Feb 2, 2022, at 06:12, AIhub <<a href="mailto:aihuborg@gmail.com" target="_blank">aihuborg@gmail.com</a>> wrote:<u></u><u></u></p>
</blockquote>
</div>
<blockquote style="margin-top:5pt;margin-bottom:5pt">
<div><p class="MsoNormal"><u></u><u></u></p>
<div>
<div><p class="MsoNormal">Stephen Hanson in conversation with Geoff Hinton<u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">In the latest episode of this video series for <a href="https://urldefense.proofpoint.com/v2/url?u=http-3A__AIhub.org&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=xnFSVUARkfmiXtiTP_uXfFKv4uNEGgEeTluRFR7dnUpay2BM5EiLz-XYCkBNJLlL&s=eOtzMh8ILIH5EF7K20Ks4Fr27XfNV_F24bkj-SPk-2A&e=" target="_blank">
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.<u></u><u></u></p>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal">You can watch the discussion, and read the transcript, here:<br clear="all">
<u></u><u></u></p>
<div><p class="MsoNormal"><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__aihub.org_2022_02_02_what-2Dis-2Dai-2Dstephen-2Dhanson-2Din-2Dconversation-2Dwith-2Dgeoff-2Dhinton_&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=OY_RYGrfxOqV7XeNJDHuzE--aEtmNRaEyQ0VJkqFCWw&e=" target="_blank">https://aihub.org/2022/02/02/what-is-ai-stephen-hanson-in-conversation-with-geoff-hinton/</a><u></u><u></u></p>
</div>
<div><p class="MsoNormal"><u></u> <u></u></p>
</div>
<div><p class="MsoNormal"><span style="font-family:Arial,sans-serif">About AIhub: </span><u></u><u></u></p>
</div>
<div><p class="MsoNormal"><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://urldefense.proofpoint.com/v2/url?u=http-3A__AIhub.org&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=xnFSVUARkfmiXtiTP_uXfFKv4uNEGgEeTluRFR7dnUpay2BM5EiLz-XYCkBNJLlL&s=eOtzMh8ILIH5EF7K20Ks4Fr27XfNV_F24bkj-SPk-2A&e=" target="_blank">
AIhub.org</a> (</span><span style="font-family:Arial,sans-serif"><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__aihub.org_&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=IKFanqeMi73gOiS7yD-X_vRx_OqDAwv1Il5psrxnhIA&e=" target="_blank">https://aihub.org/</a><span>).
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
</span><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__aaai.org_&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=wBvjOWTzEkbfFAGNj9wOaiJlXMODmHNcoWO5JYHugS0&e=" target="_blank"><span>AAAI</span></a><span>,
</span><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__neurips.cc_&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=3-lOHXyu8171pT_UE9hYWwK6ft4I-cvYkuX7shC00w0&e=" target="_blank"><span>NeurIPS</span></a><span>,
</span><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__icml.cc_imls_&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=JJyjwIpPy9gtKrZzBMbW3sRMh3P3Kcw-SvtxG35EiP0&e=" target="_blank"><span>ICML</span></a><span>,
</span><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__www.journals.elsevier.com_artificial-2Dintelligence&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=eWrRCVWlcbySaH3XgacPpi0iR0-NDQYCLJ1x5yyMr8U&e=" target="_blank"><span>AIJ</span></a><span>/</span><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__www.journals.elsevier.com_artificial-2Dintelligence&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=eWrRCVWlcbySaH3XgacPpi0iR0-NDQYCLJ1x5yyMr8U&e=" target="_blank"><span>IJCAI</span></a><span>,
</span><a href="https://urldefense.proofpoint.com/v2/url?u=http-3A__sigai.acm.org_&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=7rC6MJFaMqOms10EYDQwfnmX-zuVNhu9fz8cwUwiLGQ&e=" target="_blank"><span>ACM
SIGAI</span></a><span>, EurAI/AICOMM, </span><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__claire-2Dai.org_&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=66ZofDIhuDba6Fb0LhlMGD3XbBhU7ez7dc3HD5-pXec&e=" target="_blank"><span>CLAIRE</span></a><span>
and </span><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__www.robocup.org__&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=yl7-VPSvMrHWYKZFtKdFpThQ9UTb2jW14grhVOlAwV21R4FwPri0ROJ-uFdMqHy1&s=bBI6GRq--MHLpIIahwoVN8iyXXc7JAeH3kegNKcFJc0&e=" target="_blank"><span>RoboCup</span></a><span>.</span></span><u></u><u></u></p>
</div>
<div><p class="MsoNormal"><span style="font-family:Arial,sans-serif">Twitter: @aihuborg</span><u></u><u></u></p>
</div>
</div>
</div>
</div>
</div>
</blockquote>
</div>
</div>
</blockquote>
</div>
</div>
</blockquote>
</div>
</div>
</div>
</blockquote></div><div id="gmail-m_-1177896716680425671DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2"><br>
<table style="border-top:1px solid rgb(211,212,222)">
<tbody><tr>
<td style="width:55px;padding-top:13px"><a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__www.avast.com_sig-2Demail-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dlink-26utm-5Fcampaign-3Dsig-2Demail-26utm-5Fcontent-3Dwebmail&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=waSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj&s=Ao9QQWtO62go0hx1tb3NU6xw2FNBadjj8q64-hl5Sx4&e=" target="_blank"><img src="https://ipmcdn.avast.com/images/icons/icon-envelope-tick-round-orange-animated-no-repeat-v1.gif" alt="" width="46" height="29" style="width: 46px; height: 29px;"></a></td>
<td style="width:470px;padding-top:12px;color:rgb(65,66,78);font-size:13px;font-family:Arial,Helvetica,sans-serif;line-height:18px">Virus-free. <a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__www.avast.com_sig-2Demail-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dlink-26utm-5Fcampaign-3Dsig-2Demail-26utm-5Fcontent-3Dwebmail&d=DwMFaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=waSKY67JF57IZXg30ysFB_R7OG9zoQwFwxyps6FbTa1Zh5mttxRot_t4N7mn68Pj&s=Ao9QQWtO62go0hx1tb3NU6xw2FNBadjj8q64-hl5Sx4&e=" style="color:rgb(68,83,234)" target="_blank">www.avast.com</a>
</td>
</tr>
</tbody></table><a width="1" height="1"></a></div>
</div></blockquote></div><br></div></div></div></blockquote></div></div>