<html><head></head><body><div><div><div>I am at the initial stages of going through Stephen Grossberg's 2021 book "Conscious Mind, Resonant Brain". One thing that struck me last March-April2023 was a very vague resemblance of the basic "building block" of Transformer Neural Networks (TrNNs) and some of Grossberg's work dating back to 1972-99, and greatly extended since then. Granted, TrNNs lack essential features of Grossberg's architectures, so I'm not expecting a lot from them. Perhaps the large-scale TrNNs <br>
<br>
Three questions that came to mind back in March-April were :
<ol>
<li>Do "Large Language Models (LLMs) (such as chatGPT, LaMDA, etc) already exhibit a [protero, incipient] consciousness, in particular given the rough similarity of the basic unit of "Transformer Neural Networks" (TrNNs) to one of Grossberg's general "modules". The latter are proposed as a small number of units that are readily recombined with slight modifications as a basis for much of brain architecture, much like the small number of concepts in physics can be applied across a broad range of themes? TrNN experts were categorical in emphasizing that sentience (here I am using that term as being "somewhat similar" to consciousness) was not built in (cannot arise), but might there be something more to this?
</li><li>How difficult would it be to augment "Transformer Neural Networks" (TrNNs) with Grossberg's [concept, architecture]s, including the emergent systems for consciousness? Perhaps this would combine the scalability of the former with the [robust, extendable] foundations of the latter, which is supported by [broad, diverse, deep] :
<ul>
<li>data from [neuroscience, psychology]
</li><li>success in real world advanced [science, engineering] applications
</li><li>decades of pioneering work in the mathematical basis of Neural Networks (non-linear networks, Adaptive Resonance Theory, neural [architecture, function, process]s, etc)
</li></ul>
</li><li>Are current (semi-manual) "controls" of "Large Language Models (LLMs) going in the direction of machine consciousness, without those involved being aware of this? Will "controls" ultimately require machine consciousness as one of their components, in particular for [learning, evolution] in a stable and robust manner?
</li></ol>
Do you have any thoughts on this? I still need to do my homework on those questions, but only after I get further into the book, and build (hopefully) some Spiking Neural Network (SNN) code to implement a few basic ideas from Grossberg. Gail Carpenter has done some work with SNNs, so I am looking forward to reading through that.<br>
<br>
<br>
<b>webSite with details about the book:</b> perhaps helpful for readers and those considering its purchase<br>
<br>
For my own use as an aid to going through the book, I spent considerable time to build :
<ul>
<li>directories containing [html caption, image]s
</li><li>a webPage of "Themes" created from the captions (searchable)
</li><li>a few webPages to collect information related to consciousness, but more importantly providing some detail about the overall content of Grossberg's book.
</li></ul>
I felt that this may be helpful for others who are reading through the book. I use it especially to open multiple images at a time for easy comparison, and will build a siple script to do that automatically for selected themes. Having permission to post images from Grossberg's book to my webSite, I have put up a very early version so others might benefit.
Here are some quick links :
<ul>
<li><a href="http://www.BillHowell.ca/Neural nets/TrNNs_ART/Grossbergs list of [figure, table]s.html">list of [figure, table]s</a>, with links to view captioned figures and tables
</li><li><a href="http://www.BillHowell.ca/Neural nets/TrNNs_ART/webWork/pMenuTopMenu TrNNs_ART.html">a simple menu</a> listing webPages on the webSite
</li><li><a href="http://www.BillHowell.ca/Neural nets/TrNNs_ART/Grossbergs [core, fun, strange] concepts.html">initial list of a variety of themes from the book</a>, currently generated by a simple bash script (needs improvement)
</li></ul>
All of the above are in a very [initial, incomplete] stage, and it will take months to progress. I am actually more interested in the "non-conscious" part of Grossberg's work, the lessons from that, and how that ties in with consciousness, than I am with the theme of consciousness alone. Grossberg does the best work that I know of for that. Furthermore, some of my priorities actually go in the other direction :
<ul>
<li>what I call callerID-SNNs ("what if", not even a hypothesis yet)
</li><li>[Mendellian, Lamarckian] heredity questions.
</li></ul>
<br>
<br>
Mr. Bill Howell, <a href="mailto:Bill@BillHowell.ca">Bill@BillHowell.ca</a> 1-587-707-2027 <br>
<a href="http://www.BillHowell.ca/">http://www.BillHowell.ca/</a> (browser shows root directory of webSite)<br>
<a href="http://www.BillHowell.ca/home.html">http://www.BillHowell.ca/home.html</a> (start webPage browsing)<br>
P.O. Box 299, Hussar, Alberta, T0J1S0<br>
member - International Neural Network Society (INNS), IEEE Computational Intelligence Society (IEEE-CIS), <br>
WCCI2020 Glasgow, Publicity Chair mass emails, <a href="http://wcci2020.org/">http://wcci2020.org/</a><br>
Retired 2012: Science Research Manager (SE-REM-01) at Natural Resources Canada, CanmetMINING, Ottawa<br>
<br></div></div></div><div><span></span></div></body></html>