<div dir="ltr"><div>Hi Wlodek</div><div><br></div><div>Thanks for your very thought-provoking reply and the great reading suggestions. We have known for a long time that the brain has both modal and amodal representations of concepts. There is also evidence that abstract concepts are built on the scaffolding of concrete ones (such as directions and shapes in physical space), even in non-human animals. This is just a conjecture but I think that the ability to build abstractions is just meta-representation made possible by hierarchical depth with the evolution of the cortex. So the representations such as hippocampal place codes, built as a direct result of embodied experience, become "the world" for higher layers of processing doing essentially the same thing but with a different level of grounding - with abstract concepts as "place codes" in a more abstract space. I call it "multi-level grounding". When you became grounded at the level of group theory, your grounding in embodiment was temporarily obscured because it was several level down. Of course, this is hardly a new idea, but worth keeping in mind.</div><div><br></div><div>To go on a bit of a tangent but not unrelated, did you see this new paper:<br><br><a href="https://arxiv.org/abs/2202.07206">https://arxiv.org/abs/2202.07206</a></div><div><br></div><div>Apparently, GPT-3 does rely more than people admit on regurgitation. I don't think any language model build on the distributional hypothesis can ever be sufficiently grounded to have "understanding", but  it should be possible in highly formalized domains such as computer programming, where the truth is so constrained and present wholly in the patterns of symbols. Natura; language less so. Actual experience, hardly at all.</div><div><br></div><div>Ali<br></div><div><br></div><div><br></div><div><br></div><div> </div><div><div><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr"><div><b>Ali A. Minai, Ph.D.</b><br>Professor and Graduate Program Director<br>Complex Adaptive Systems Lab<br>Department of Electrical Engineering & Computer Science<br></div><div>828 Rhodes Hall<br></div><div>University of Cincinnati<br>Cincinnati, OH 45221-0030<br><br></div><div>Past-President (2015-2016)<br></div><div>International Neural Network Society<br></div><div><br>Phone: (513) 556-4783<br>Fax: (513) 556-7326<br>Email: <a href="mailto:Ali.Minai@uc.edu" target="_blank">Ali.Minai@uc.edu</a><br>          <a href="mailto:minaiaa@gmail.com" target="_blank">minaiaa@gmail.com</a><br><br>WWW: <a href="http://www.ece.uc.edu/%7Eaminai/" target="_blank">https://eecs.ceas.uc.edu/~aminai/</a></div></div></div></div></div></div></div></div></div></div></div></div></div></div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Fri, Feb 18, 2022 at 12:45 PM Wlodzislaw Duch <<a href="mailto:wduch@umk.pl">wduch@umk.pl</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>
    <p><font face="Arial">Ali,</font></p>
    <p><font face="Arial">certainly for many people identification with
        "being like us" is important - this covers fertilized eggs and
        embryos, but not orangutans. </font><font face="Arial">John
        Locke wrote 300 years ago: "Consciousness is the perception of
        what passes in a Man's own mind". Physical states and processes
        that represent imagery, and the ability to create symbolic
        narratives describing what goes on inside cognitive system,
        should be the hallmark of consciousness. Of course more people
        will accept it if we put it in a baby robot -:)</font></p>
    <p><font face="Arial">This is why I prefer to focus on a simple
        requirement: inner world and the ability to describe it. </font><br>
      <font face="Arial"><font face="Arial">The road to create robots
          that can feel has been described by Kevin O'Regan in the book:
          <br>
        </font>
      </font></p>
    <p>O’Regan, J.K. (2011). Why Red Doesn’t Sound Like a Bell:
      Understanding the Feel of Consciousness. Oxford University Press,
      USA.</p>
    <p><font face="Arial">Inner worlds may be based on different
        representations, not always deeply grounded in experience.
        Binder made a step toward a brain-based semantics:<br>
      </font></p>
    Binder, J. R., Conant, L. L., Humphries, C. J., Fernandino, L.,
    Simons, S. B., Aguilar, M., & Desai, R. H. (2016). Toward a
    brain-based componential semantic representation. Cognitive
    Neuropsychology, 33(3–4), 130–174.<br>
    Fernandino, L., Tong, J.-Q., Conant, L. L., Humphries, C. J., &
    Binder, J. R. (2022). Decoding the information structure underlying
    the neural representation of concepts. PNAS 119(6). <br>
    <font face="Arial"><br>
      This does not solve the symbol grounding problem (Harnad, 1990),
      but goes half the way, mimicking embodiment by decomposing
      symbolic concepts into attributes that are relevant to the brain.
      It should be sufficient to add human-like </font><font face="Arial"><font face="Arial">semantics to bots</font>. As you
      mention yourself, embodiment could be more abstract, and I can
      imagine that a copy of a robot brain that has grounded its
      representations in interactions with environment will endow a new
      robot with similar experience. Can we simply implant it in the
      network?  <br>
    </font>
    <p>I wonder if absorption in abstract thinking can leave space for
      the use of experientially grounded concepts. I used to focus on
      group theory for hours and was not able to understand what was
      said to me for brief moments. Was I not conscious? Or should we
      consider continuous transition from abstract semantics to fully
      embodied, human-like semantics in artificial systems? <br>
    </p>
    <p>Wlodek</p>
    <p><br>
    </p>
    <div>On 18/02/2022 16:36, Ali Minai wrote:<br>
    </div>
    <blockquote type="cite">
      
      <div dir="ltr">
        <div>Wlodek</div>
        <div><br>
        </div>
        <div>I think that the debate about consciousness in the strong
          sense of having a conscious experience like we do is sterile.
          We will never have a measuring device for whether another
          entity is "conscious", and at some point, we will get to an AI
          that is sufficiently complex in its observable behavior that
          we will either accept its inner state of consciousness on
          trust - just as we do with humans and other animals - or admit
          that we will never believe that a machine that is "not like
          us" can ever be conscious. The "like us" part is more
          important than many of us in the AI field think: A big part of
          why we believe other humans and our dogs are conscious is
          because we know that they are "like us", and assume that they
          must share our capacity for inner conscious experience. We
          already see this at a superficial level where, as ordinary
          humans, we have a much easier time identifying with an
          embodied, humanoid AI like Wall-E or the Terminator than with
          a disembodied one like Hal or Skynet. This is also why so many
          people find the Boston Dynamics "dog" so disconcerting.<br>
        </div>
        <div><br>
        </div>
        <div>The question of embodiment is a complex one, as you know,
          of course, but I am with those who think that it is necessary
          for grounding mental representations - that it is the only way
          that the internal representations of the system are linked
          directly to its experience. For example, if an AI system
          trained only on text (like GPT-3) comes to learn that touching
          something hot results in the fact of getting burned, we cannot
          accept that as sufficient because it is based only on the
          juxtaposition of abstractions, not the actual painful
          experience of getting burned. For that, you need a body with
          sensors and a brain with a state corresponding to pain -
          something that can be done in an embodied robot. This is why I
          think that all language systems trained purely on the
          assumption of the distributional hypothesis of meaning will
          remain superficial; they lack the grounding that can only be
          supplied by experience. This does not mean that systems based
          on the distributional hypothesis cannot learn a lot, or even
          develop brain-like representations, as the following extremely
          interesting paper shows:</div>
        <div><br>
        </div>
        <div>Y. Zhang, K. Han, R. Worth, and Z. Liu. Connecting concepts
          in the brain by mapping cortical representations of semantic
          relations. Nature Communications, 11(1):1877, Apr 2020.</div>
        <div><br>
          In a formal sense, however, embodiment could be in any space,
          including very abstract ones. We can think of text data as
          GPT-3's world and, in that world, it is "embodied" and its
          fundamentally distributional learning, though superficial and
          lacking in experience to us, is grounded for it within its
          world. Of course, this is not a very useful view of embodiment
          and grounding since we want to create AI that is grounded in
          our sense, but one of the most under-appreciated risks of AI
          is that, as we develop systems that live in worlds very
          different than ours, they will - implicitly and emergently -
          embody values completely alien to us. The proverbial
          loan-processing AI that learns to be racially biased in just a
          caricature of this hazard, but one that should alert us to
          deeper issues. Our quaintly positivistic and reductionistic
          notion that we can deal with such things by removing biases
          from data, algorithms, etc., is misplaced. The world is too
          complicated for that.</div>
        <div><br>
        </div>
        <div>Ali</div>
        <div><br>
        </div>
        <div>
          <div>
            <div dir="ltr">
              <div dir="ltr">
                <div>
                  <div dir="ltr">
                    <div>
                      <div dir="ltr">
                        <div>
                          <div dir="ltr">
                            <div>
                              <div dir="ltr">
                                <div>
                                  <div dir="ltr">
                                    <div><b>Ali A. Minai, Ph.D.</b><br>
                                      Professor and Graduate Program
                                      Director<br>
                                      Complex Adaptive Systems Lab<br>
                                      Department of Electrical
                                      Engineering & Computer Science<br>
                                    </div>
                                    <div>828 Rhodes Hall<br>
                                    </div>
                                    <div>University of Cincinnati<br>
                                      Cincinnati, OH 45221-0030<br>
                                    </div>
                                    <div><br>
                                      Phone: (513) 556-4783<br>
                                      Fax: (513) 556-7326<br>
                                      Email: <a href="mailto:Ali.Minai@uc.edu" target="_blank">Ali.Minai@uc.edu</a><br>
                                                <a href="mailto:minaiaa@gmail.com" target="_blank">minaiaa@gmail.com</a><br>
                                      <br>
                                      WWW: <a href="http://www.ece.uc.edu/%7Eaminai/" target="_blank">https://eecs.ceas.uc.edu/~aminai/</a></div>
                                  </div>
                                </div>
                              </div>
                            </div>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </div>
              </div>
            </div>
          </div>
          <br>
        </div>
      </div>
      <br>
      <div class="gmail_quote">
        <div dir="ltr" class="gmail_attr">On Fri, Feb 18, 2022 at 7:27
          AM Wlodzislaw Duch <<a href="mailto:wduch@umk.pl" target="_blank">wduch@umk.pl</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>
            <p>Asim,<br>
            </p>
            <p>I was on the Anchorage panel, and asked others what could
              be a great achievement in computational intelligence.
              Steve Grossberg replied, that symbolic AI is meaningless,
              but creation of artificial rat that could survive in
              hostile environment would be something. Of course this is
              still difficult, but perhaps DARPA autonomous machines are
              not that far? <br>
            </p>
            <p>I also had similar discussions with Walter and support
              his position: you cannot separate tightly coupled systems.
              Any external influence will create activation in both,
              linear causality looses its meaning. This is clear if both
              systems adjust to each other. But even if only one system
              learns (brain) and the other is mechanical but responds to
              human actions it may behave as one system. Every musician
              knows that: piano becomes a part of our body, responding
              in so many ways to actions, not only by producing sounds
              but also providing haptic feedback. <br>
            </p>
            <p>This simply means that brains of locked-in people worked
              in somehow different way than brains of healthy people.
              Why do we consider them conscious? Because they can
              reflect on their mind states, imagine things and describe
              their inner states. If GPT-3 was coupled with something
              like DALL-E that creates images from text, and could
              describe what they see in their inner world, create some
              kind of episodic memory, we would have hard time to deny
              that this thing is not conscious of what it has in its
              mind. Embodiment helps to create inner world and changes
              it, but it is not necessary for consciousness. Can we find
              a good argument that such system is not conscious of its
              own states? It may not have all qualities of human
              consciousness, but that is a matter of more detailed
              approximation of missing functions. <br>
            </p>
            <p> I have made this argument a long time ago (ex. in "<i style="color:rgb(0,0,0);font-family:arial,helvetica,calibri;font-size:medium;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:left;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial"><a>Brain-inspired conscious
                  computing architecture</a>" </i><span style="color:rgb(0,0,0);font-family:arial,helvetica,calibri;font-size:medium;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:left;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial">written
                over 20 years ago, see more papers on this on my web
                page).  </span><span style="color:rgb(0,0,0);font-family:arial,helvetica,calibri;font-size:medium;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:left;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial"></span></p>
            <p>Wlodek</p>
            <p>Prof. Włodzisław Duch<br>
              Fellow, International Neural Network Society<br>
              Past President, European Neural Network Society<br>
              Head, Neurocognitive Laboratory, CMIT NCU, Poland<br>
            </p>
            <p>Google: <a href="http://www.google.com/search?q=Wlodek+Duch" target="_blank">Wlodzislaw Duch</a></p>
            <div><br>
              On 18/02/2022 05:22, Asim Roy wrote:<br>
            </div>
            <blockquote type="cite">
              <div>
                <p class="MsoNormal">In 1998, after our debate about the
                  brain at the WCCI in Anchorage, Alaska, I asked Walter
                  Freeman if he thought the brain controls the body. His
                  answer was, you can also say that the body controls
                  the brain. I then asked him if the driver controls a
                  car, or the pilot controls an airplane. His answer was
                  the same, that you can also say that the car controls
                  the driver, or the plane controls the pilot. I then
                  realized that Walter was also a philosopher and
                  believed in the No-free Will theory and what he was
                  arguing for is that the world is simply made of
                  interacting systems. However, both Walter, and his
                  close friend John Taylor, were into consciousness. </p>
                <p class="MsoNormal"> </p>
                <p class="MsoNormal">I have argued with Walter on many
                  different topics over nearly two decades and have
                  utmost respect for him as a scholar, but this first
                  argument I will always remember.</p>
                <p class="MsoNormal"> </p>
                <p class="MsoNormal">Obviously, there’s a conflict
                  between consciousness and the No-free Will theory.
                  Wonder where we stand with regard to this conflict.</p>
                <p class="MsoNormal"> </p>
                <p class="MsoNormal">Asim Roy</p>
                <p class="MsoNormal">Professor, Information Systems</p>
                <p class="MsoNormal">Arizona State University</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></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></p>
                <p class="MsoNormal"> </p>
                <p class="MsoNormal"> </p>
                <div>
                  <div style="border-color:rgb(225,225,225) currentcolor currentcolor;border-style:solid none none;border-width:1pt medium medium;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>Andras Lorincz<br>
                      <b>Sent:</b> Tuesday, February 15, 2022 6:50 AM<br>
                      <b>To:</b> Stephen José Hanson <a href="mailto:jose@rubic.rutgers.edu" target="_blank"><jose@rubic.rutgers.edu></a>;
                      Gary Marcus <a href="mailto:gary.marcus@nyu.edu" target="_blank"><gary.marcus@nyu.edu></a><br>
                      <b>Cc:</b> Connectionists <a href="mailto:Connectionists@cs.cmu.edu" target="_blank"><Connectionists@cs.cmu.edu></a><br>
                      <b>Subject:</b> Re: Connectionists: Weird beliefs
                      about consciousness</p>
                  </div>
                </div>
                <p class="MsoNormal"> </p>
                <div>
                  <p class="MsoNormal"><span style="font-size:12pt;color:black">Dear Steve and
                      Gary:</span></p>
                </div>
                <div>
                  <div>
                    <p class="MsoNormal"><span style="font-size:12pt;color:black">This is how I
                        see (try to understand) consciousness and the
                        related terms: </span></p>
                    <div>
                      <p class="MsoNormal"><span style="font-size:12pt;color:black">(Our)
                          consciousness seems to be related to the
                          close-to-deterministic nature of the episodes
                          on from few hundred millisecond to a few
                          second domain. Control instructions may leave
                          our brain 200 ms earlier than the action
                          starts and they become conscious only by that
                          time. In addition, observations of those may
                          also be delayed by a similar amount. (It then
                          follows that the launching of the control
                          actions is not conscious and -- therefore --
                          free will can be debated in this very limited
                          context.) On the other hand, model-based
                          synchronization is necessary for timely
                          observation, planning, decision making, and
                          execution in a distributed and slow
                          computational system. If this model-based
                          synchronization is not working properly, then
                          the observation of the world breaks and
                          schizophrenic symptoms appear. As an example,
                          individuals with pronounced schizotypal traits
                          are particularly successful in self-tickling
                          (source: <a href="https://urldefense.com/v3/__https:/philpapers.org/rec/LEMIWP__;!!IKRxdwAv5BmarQ!P1ufmU5XnzpvjxtS2M0AnytlX24RNsoDeNPfsqUNWbF6OU5p9xMqtMj9S3Pn3cY$" target="_blank">
                            https://philpapers.org/rec/LEMIWP</a>, and a
                          discussion on Asperger and schizophrenia: <a href="https://urldefense.com/v3/__https:/www.frontiersin.org/articles/10.3389/fpsyt.2020.503462/full__;!!IKRxdwAv5BmarQ!P1ufmU5XnzpvjxtS2M0AnytlX24RNsoDeNPfsqUNWbF6OU5p9xMqtMj9l5NkQt4$" target="_blank">
https://www.frontiersin.org/articles/10.3389/fpsyt.2020.503462/full</a>)
                          a manifestation of improper binding. The
                          internal model enables and the synchronization
                          requires the internal model and thus a certain
                          level of consciousness can appear in a time
                          interval around the actual time instant and
                          its length depends on the short-term memory.</span></p>
                    </div>
                    <p class="MsoNormal"><span style="font-size:12pt;color:black">Other issues,
                        like separating the self from the rest of the
                        world are more closely related to the soft/hard
                        style interventions (as called in the recent
                        deep learning literature), i.e., those
                        components (features) that can be
                        modified/controlled, e.g., color and speed, and
                        the ones that are Lego-like and can be
                        separated/amputed/occluded/added.</span></p>
                  </div>
                  <div>
                    <p class="MsoNormal"><span style="font-size:12pt;color:black">Best,</span></p>
                  </div>
                  <div>
                    <p class="MsoNormal"><span style="font-size:12pt;color:black">Andras</span></p>
                  </div>
                  <div id="gmail-m_1175030980421304624gmail-m_1420086470503611017Signature">
                    <div>
                      <div id="gmail-m_1175030980421304624gmail-m_1420086470503611017divtagdefaultwrapper">
                        <div id="gmail-m_1175030980421304624gmail-m_1420086470503611017divtagdefaultwrapper">
                          <div name="divtagdefaultwrapper">
                            <div>
                              <div>
                                <div>
                                  <div>
                                    <p><span style="color:black"> </span></p>
                                    <p><span style="color:black">------------------------------------</span></p>
                                    <div>
                                      <p class="MsoNormal"><span style="font-size:12pt;color:black">Andras
                                          Lorincz</span></p>
                                    </div>
                                    <div>
                                      <p class="MsoNormal"><span style="font-size:12pt;color:black"><a href="https://urldefense.com/v3/__http:/nipg.inf.elte.hu/__;!!IKRxdwAv5BmarQ!P1ufmU5XnzpvjxtS2M0AnytlX24RNsoDeNPfsqUNWbF6OU5p9xMqtMj9j2LbdH0$" target="_blank">http://nipg.inf.elte.hu/</a></span></p>
                                    </div>
                                    <div>
                                      <p class="MsoNormal"><span style="font-size:12pt;color:black">Fellow
                                          of the European Association
                                          for Artificial Intelligence</span></p>
                                    </div>
                                    <div>
                                      <p class="MsoNormal"><span style="font-size:12pt;color:black"><a href="https://urldefense.com/v3/__https:/scholar.google.com/citations?user=EjETXQkAAAAJ&hl=en__;!!IKRxdwAv5BmarQ!P1ufmU5XnzpvjxtS2M0AnytlX24RNsoDeNPfsqUNWbF6OU5p9xMqtMj99i1VRm0$" target="_blank">https://scholar.google.com/citations?user=EjETXQkAAAAJ&hl=en</a></span></p>
                                    </div>
                                    <div>
                                      <p class="MsoNormal"><span style="font-size:12pt;color:black">Department
                                          of Artificial Intelligence</span></p>
                                    </div>
                                    <div>
                                      <p class="MsoNormal"><span style="font-size:12pt;color:black">Faculty
                                          of Informatics</span></p>
                                    </div>
                                    <div>
                                      <p class="MsoNormal"><span style="font-size:12pt;color:black">Eotvos
                                          Lorand University</span></p>
                                    </div>
                                    <p class="MsoNormal"><span style="font-size:12pt;color:black">Budapest,
                                        Hungary </span></p>
                                    <p><span style="color:black"> </span></p>
                                    <p><span style="color:black"> </span></p>
                                  </div>
                                </div>
                              </div>
                            </div>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </div>
                <div>
                  <p class="MsoNormal"><span style="font-size:12pt;color:black"> </span></p>
                </div>
                <div class="MsoNormal" style="text-align:center" align="center">
                  <hr width="98%" size="2" align="center"> </div>
                <div id="gmail-m_1175030980421304624gmail-m_1420086470503611017divRplyFwdMsg">
                  <p class="MsoNormal"><b><span style="color:black">From:</span></b><span style="color:black"> Connectionists <<a href="mailto:connectionists-bounces@mailman.srv.cs.cmu.edu" target="_blank">connectionists-bounces@mailman.srv.cs.cmu.edu</a>>
                      on behalf of Stephen José Hanson <<a href="mailto:jose@rubic.rutgers.edu" target="_blank">jose@rubic.rutgers.edu</a>><br>
                      <b>Sent:</b> Monday, February 14, 2022 8:30 PM<br>
                      <b>To:</b> Gary Marcus <<a href="mailto:gary.marcus@nyu.edu" target="_blank">gary.marcus@nyu.edu</a>><br>
                      <b>Cc:</b> Connectionists <<a href="mailto:connectionists@cs.cmu.edu" target="_blank">connectionists@cs.cmu.edu</a>><br>
                      <b>Subject:</b> Re: Connectionists: Weird beliefs
                      about consciousness</span> </p>
                  <div>
                    <p class="MsoNormal"> </p>
                  </div>
                </div>
                <div>
                  <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">Gary,  these
                      weren't criterion.     Let me try again.</span></p>
                  <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">I wasn't
                      talking about wake-sleep cycles... I was talking
                      about being awake or asleep and the transition
                      that ensues..</span></p>
                  <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">Rooba's don't
                      sleep.. they turn off, I have two of them.  They
                      turn on once (1) their batteries are recharged (2)
                      a timer has been set for being turned on.</span></p>
                  <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">GPT3 is
                      essentially a CYC that actually works.. by reading
                      Wikipedia (which of course is a terribly biased
                      sample).</span></p>
                  <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">I was
                      indicating the difference between implicit and
                      explicit learning/problem solving.    Implicit
                      learning/memory is unconscious and similar to a
                      habit.. (good or bad).</span></p>
                  <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">I believe
                      that when someone says "is gpt3 conscious?"  they
                      are asking: is gpt3 self-aware?      Roombas know
                      about vacuuming and they are unconscious.</span></p>
                  <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">S</span></p>
                  <div>
                    <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">On 2/14/22
                        12:45 PM, Gary Marcus wrote:</span></p>
                  </div>
                  <blockquote style="margin-top:5pt;margin-bottom:5pt">
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">Stephen,</span></p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">On
                          criteria (1)-(3), a high-end,
                          mapping-equippped Roomba is far more plausible
                          as a consciousness than GPT-3.</span></p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">1. The
                          Roomba has a clearly defined wake-sleep cycle;
                          GPT does not.</span></p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">2.
                          Roomba makes choices based on an explicit
                          representation of its location relative to a
                          mapped space. GPT lacks any consistent
                          reflection of self; eg if you ask it, as I
                          have, if you are you person, and then ask if
                          it is a computer, it’s liable to say yes to
                          both, showing no stable knowledge of self.</span></p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">3.
                          Roomba has explicit, declarative knowledge eg
                          of walls and other boundaries, as well its own
                          location. GPT has no systematically
                          interrogable explicit representations.</span></p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">All this
                          is said with tongue lodged partway in cheek,
                          but I honestly don’t see what criterion would
                          lead anyone to believe that GPT is a more
                          plausible candidate for consciousness than any
                          other AI program out there. </span></p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">ELIZA
                          long ago showed that you could produce fluent
                          speech that was mildly contextually relevant,
                          and even convincing to the untutored; just
                          because GPT is a better version of that trick
                          doesn’t mean it’s any more conscious.</span></p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">Gary</span></p>
                    </div>
                    <div>
                      <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black"><br>
                          <br>
                        </span></p>
                      <blockquote style="margin-top:5pt;margin-bottom:5pt">
                        <p class="MsoNormal" style="margin-bottom:12pt;background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">On Feb 14, 2022, at
                            08:56, Stephen José Hanson <a href="mailto:jose@rubic.rutgers.edu" target="_blank"><jose@rubic.rutgers.edu></a>
                            wrote:</span></p>
                      </blockquote>
                    </div>
                    <blockquote style="margin-top:5pt;margin-bottom:5pt">
                      <div>
                        <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black"> </span>
                        </p>
                        <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">this is
                            a great list of behavior.. </span></p>
                        <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">Some
                            biologically might be termed reflexive,
                            taxes, classically conditioned, implicit
                            (memory/learning)... all however would not
                            be<br>
                            conscious in the several senses:  (1) 
                            wakefulness-- sleep  (2) self aware (3)
                            explicit/declarative.</span></p>
                        <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">I think
                            the term is used very loosely, and I believe
                            what GPT3 and other AI are hoping to show
                            signs of is "self-awareness"..</span></p>
                        <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">In
                            response to :  "why are you doing that?", 
                            "What are you doing now", "what will you be
                            doing in 2030?"</span></p>
                        <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="font-size:13.5pt;color:black">Steve</span></p>
                        <p style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                        <div>
                          <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">On 2/14/22 10:46 AM,
                              Iam Palatnik wrote:</span></p>
                        </div>
                        <blockquote style="margin-top:5pt;margin-bottom:5pt">
                          <div>
                            <div>
                              <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">A somewhat related
                                  question, just out of curiosity.</span></p>
                            </div>
                            <div>
                              <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                            </div>
                            <div>
                              <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">Imagine the
                                  following:</span></p>
                            </div>
                            <div>
                              <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                            </div>
                            <div>
                              <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">- An automatic
                                  solar panel that tracks the position
                                  of the sun.</span></p>
                            </div>
                            <div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">- A group of
                                    single celled microbes with
                                    phototaxis that follow the sunlight.</span></p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">- A jellyfish
                                    (animal without a brain) that
                                    follows/avoids the sunlight.</span></p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">- A cockroach
                                    (animal with a brain) that avoids
                                    the sunlight.</span></p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">- A drone with
                                    onboard AI that flies to regions of
                                    more intense sunlight to recharge
                                    its batteries.</span></p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">- A human that
                                    dislikes sunlight and actively
                                    avoids it.</span></p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">Can any of
                                    these, beside the human, be said to
                                    be aware or conscious of the
                                    sunlight, and why?</span></p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">What is most
                                    relevant? Being a biological life
                                    form, having a brain, being able to
                                    make decisions based on the
                                    environment? Being taxonomically
                                    close to humans?</span></p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                              </div>
                              <div>
                                <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                              </div>
                            </div>
                          </div>
                          <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"> </p>
                          <div>
                            <div>
                              <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">On Mon, Feb 14,
                                  2022 at 12:06 PM Gary Marcus <<a href="mailto:gary.marcus@nyu.edu" target="_blank">gary.marcus@nyu.edu</a>>
                                  wrote:</span></p>
                            </div>
                            <blockquote style="border-color:currentcolor currentcolor currentcolor rgb(204,204,204);border-style:none none none solid;border-width:medium medium medium 1pt;padding:0in 0in 0in 6pt;margin-left:4.8pt;margin-right:0in">
                              <p class="MsoNormal" style="margin-bottom:12pt;background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">Also true: Many AI
                                  researchers are very unclear about
                                  what consciousness is and also very
                                  sure that ELIZA doesn’t have it.<br>
                                  <br>
                                  Neither ELIZA nor GPT-3 have<br>
                                  - anything remotely related to
                                  embodiment<br>
                                  - any capacity to reflect upon
                                  themselves<br>
                                  <br>
                                  Hypothesis: neither keyword matching
                                  nor tensor manipulation, even at
                                  scale, suffice in themselves to
                                  qualify for consciousness.<br>
                                  <br>
                                  - Gary<br>
                                  <br>
                                  > On Feb 14, 2022, at 00:24,
                                  Geoffrey Hinton <<a href="mailto:geoffrey.hinton@gmail.com" target="_blank">geoffrey.hinton@gmail.com</a>>
                                  wrote:<br>
                                  > <br>
                                  > Many AI researchers are very
                                  unclear about what consciousness is
                                  and also very sure that GPT-3 doesn’t
                                  have it. It’s a strange combination.<br>
                                  > <br>
                                  > </span></p>
                            </blockquote>
                          </div>
                        </blockquote>
                        <div>
                          <p class="MsoNormal" style="background:rgb(236,202,153) none repeat scroll 0% 0%"><span style="color:black">--</span></p>
                        </div>
                      </div>
                    </blockquote>
                  </blockquote>
                </div>
              </div>
            </blockquote>
            <br>
          </div>
        </blockquote>
      </div>
    </blockquote>
    <div>-- <br>
      Prof. Włodzisław Duch
      <br>
      Fellow, International Neural Network Society
      <br>
      Past President, European Neural Network Society
      <br>
      Head, Neurocognitive Laboratory, CMIT NCU, Poland
      <br>
      Google: <a href="http://www.google.com/search?q=Wlodek+Duch" target="_blank">Wlodzislaw
        Duch</a></div>
  </div>

</blockquote></div>