Connectionists: Annotated History of Modern AI and Deep Learning
Grossberg, Stephen
steve at bu.edu
Wed Jan 18 12:58:43 EST 2023
Dear Gary,
I feel very lucky to have met Sue Carey and Henry and Lila Gleitman over the years.
I met Sue as a colleague at MIT and as a fellow member of the Society of Experimental Psychologists. As I recall, I met both Henry and Lila when I lectured at the University of Pennsylvania over the years.
Some of my own recent work is about a topic that interested both of them: How children learn language meanings through their interpersonal interactions with caregivers in real time.
Best,
Steve
________________________________
From: Gary Marcus <gary.marcus at nyu.edu>
Sent: Tuesday, January 17, 2023 8:16 PM
To: Grossberg, Stephen <steve at bu.edu>
Cc: Michael Arbib <arbib at usc.edu>; connectionists at cs.cmu.edu <connectionists at cs.cmu.edu>
Subject: Re: Connectionists: Annotated History of Modern AI and Deep Learning
Luckily, by the time I came on to the scene, more doors were open to women. Two of my greatest mentors, Susan Carey and Lila Gleitman, wrote powerful intellectual memoirs of their own. Neither worked specifically on AI, but the lessons I learned from both have been central to my own thinking, as I have transitioned from studying natural intelligence into studying artificial intelligence.
Both memoirs are delightfully written and well worth reading:
<https://www.annualreviews.org/doi/citedby/10.1146/annurev-devpsych-040622-091723>
Becoming a Cognitive Scientist<https://www.annualreviews.org/doi/citedby/10.1146/annurev-devpsych-040622-091723>
annualreviews.org<https://www.annualreviews.org/doi/citedby/10.1146/annurev-devpsych-040622-091723>
[apple-touch-icon-1582743354193.png]<https://www.annualreviews.org/doi/citedby/10.1146/annurev-devpsych-040622-091723>
<https://www.annualreviews.org/doi/10.1146/annurev-psych-032921-053737?url_ver=Z39.88-2003&rfr_id=ori%3Arid%3Acrossref.org&rfr_dat=cr_pub++0pubmed>
Recollecting What We Once Knew: My Life in Psycholinguistics<https://www.annualreviews.org/doi/10.1146/annurev-psych-032921-053737?url_ver=Z39.88-2003&rfr_id=ori%3Arid%3Acrossref.org&rfr_dat=cr_pub++0pubmed>
annualreviews.org<https://www.annualreviews.org/doi/10.1146/annurev-psych-032921-053737?url_ver=Z39.88-2003&rfr_id=ori%3Arid%3Acrossref.org&rfr_dat=cr_pub++0pubmed>
[apple-touch-icon-1582743354193.png]<https://www.annualreviews.org/doi/10.1146/annurev-psych-032921-053737?url_ver=Z39.88-2003&rfr_id=ori%3Arid%3Acrossref.org&rfr_dat=cr_pub++0pubmed>
On Jan 17, 2023, at 4:35 PM, Grossberg, Stephen <steve at bu.edu> wrote:
Dear Gary et al.,
Your reminiscence about McCarthy and LISP reminds me of a story about BASIC and computer time-sharing. The latter were both introduced by two of my math professors at Dartmouth, John Kemeny and Tom Kurtz, a few years after I was an undergraduate student there: https://en.wikipedia.org/wiki/John_G._Kemeny<https://urldefense.proofpoint.com/v2/url?u=https-3A__en.wikipedia.org_wiki_John-5FG.-5FKemeny&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=AXFmtXKk35sKnMlyi-g4fWlHr8eynGWXswDPeK1VCoxsf8gsIHzgpAzDnyQyoBTQ&s=ds-64IlZlHMCXtFpNfcgP_6Bt8ku75Uq8xh3QRAqD9Y&e=>
Kemeny had a profound influence on my life in science. Just after I made some of my first discoveries about how to model mind and brain, I took his course in the philosophy of science as a sophomore. Until that point, essentially all of my science courses were taught by the book, with no hint of the passions and meandering pathways that often led to discoveries.
I felt that my head exploded with ideas when I made my first discoveries, but I had no idea how to "do" science. His course was incredibly liberating and instructive for me.
Kemeny was an eloquent lecturer who made mathematics live. He believed that good mathematics students should go into the social sciences. He put his money where his mouth is by writing, with another of my math professors, J. Laurie Snell, the book Mathematical Models in the Social Sciences, which is still in print today: https://www.amazon.com/Mathematical-Models-Social-Sciences-Press/dp/0262610302<https://urldefense.proofpoint.com/v2/url?u=https-3A__www.amazon.com_Mathematical-2DModels-2DSocial-2DSciences-2DPress_dp_0262610302&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=AXFmtXKk35sKnMlyi-g4fWlHr8eynGWXswDPeK1VCoxsf8gsIHzgpAzDnyQyoBTQ&s=y5cPcx1wYLPyGC_cVfzogwK5MWMhqRE-dGmdjmRjvAs&e=>
Kemeny was one of my most important mentors who encouraged my early work. I became Dartmouth's first joint major in mathematics and psychology with his full support.
Returning to a theme of my earlier email, Kemeny was Einstein's last assistant at Princeton before being hired as a full professor at Dartmouth at age 27 and becoming chairman of the mathematics department a couple of years later.
Another set of lucky circumstances that helped me to find my own path.
Not surprisingly, I also discuss Kemeny in my Magnum Opus
https://www.amazon.com/Conscious-Mind-Resonant-Brain-Makes/dp/0190070552<https://urldefense.proofpoint.com/v2/url?u=https-3A__www.amazon.com_Conscious-2DMind-2DResonant-2DBrain-2DMakes_dp_0190070552&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=AXFmtXKk35sKnMlyi-g4fWlHr8eynGWXswDPeK1VCoxsf8gsIHzgpAzDnyQyoBTQ&s=xWvGmI3TjHslCfNV7z2wCyxCzlWE5AdEBY5x5HV_xXk&e=>
Best again,
Steve
________________________________
From: Gary Marcus <gary.marcus at nyu.edu>
Sent: Tuesday, January 17, 2023 6:31 PM
To: Grossberg, Stephen <steve at bu.edu>
Cc: Michael Arbib <arbib at usc.edu>; connectionists at cs.cmu.edu <connectionists at cs.cmu.edu>
Subject: Re: Connectionists: Annotated History of Modern AI and Deep Learning
I am just old enough to appreciate all this and young enough not to have met any of them. My late father took Fortran (on punch cards) from John McCarthy at MIT, and McCarthy (who I did later meet) left in the middle of the semester, having just invented LISP.
On Jan 17, 2023, at 15:01, Grossberg, Stephen <steve at bu.edu> wrote:
Dear Gary, Michael, and other Connectionists colleagues,
Michael's reminiscences remind me of related memories.
I was traveling from Stanford in 1964 to MIT to study with Norbert Wiener when I heard that Wiener had just died. I was a PhD student at Stanford then and was hoping to find a more congenial place at MIT to continue my work on neural networks.
Instead, I got my PhD at the Rockefeller Institute for Medical Research in New York, where Abe Pais, who worked for Niels Bohr and was a colleague of Albert Einstein at the Institute for Advanced Study, was on the faculty. Abe regaled us with rather colorful stories about both Bohr and Einstein.
Later, when I was an assistant professor at MIT, I got to know Norman Levinson and his wife, Fagi, very well. Norman was Wiener's most famous student:
https://en.wikipedia.org/wiki/Norman_Levinson<https://urldefense.proofpoint.com/v2/url?u=https-3A__en.wikipedia.org_wiki_Norman-5FLevinson&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=ou936izPpFpChNwRZ7hsXjS2h1WcDB3Zn1SvpbrBqQJ2UGafMEGcvLbbCYXBrlLo&s=Y1JYLzdcDoSmpHQwk-rqybKXqInvvLKYY1kOvYoMiok&e=>
Norman and Fagi told me lots of stories about Wiener's many famous idiosyncrasies.
While I was a young professor at MIT, Norman and Fagi generously treated me as their scientific godson and took me under their wing both at their home and at many scientific conferences. After Norman's death, Fagi became our daughter's god grandmother and shared many happy family celebrations with us.
For an interesting and heartwarming story about Fagi's impact on the mathematics community to which she was connected through Norman, and thus Wiener, see:
https://news.mit.edu/2010/obit-levinson<https://urldefense.proofpoint.com/v2/url?u=https-3A__news.mit.edu_2010_obit-2Dlevinson&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=ou936izPpFpChNwRZ7hsXjS2h1WcDB3Zn1SvpbrBqQJ2UGafMEGcvLbbCYXBrlLo&s=uKihsMN0Ynd0WkGoAO4bDdn5Qe7EoUmsYKFVX5FUTFg&e=>
I also had the good luck to meet Warren McCulloch
https://en.wikipedia.org/wiki/Warren_Sturgis_McCulloch<https://urldefense.proofpoint.com/v2/url?u=https-3A__en.wikipedia.org_wiki_Warren-5FSturgis-5FMcCulloch&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=ou936izPpFpChNwRZ7hsXjS2h1WcDB3Zn1SvpbrBqQJ2UGafMEGcvLbbCYXBrlLo&s=rwamaAhUn_ITk-YlmIZximRviDFml0IDrxBGDmRKPBU&e=>
and Jerry Lettvin https://en.wikipedia.org/wiki/Jerome_Lettvin<https://urldefense.proofpoint.com/v2/url?u=https-3A__en.wikipedia.org_wiki_Jerome-5FLettvin&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=ou936izPpFpChNwRZ7hsXjS2h1WcDB3Zn1SvpbrBqQJ2UGafMEGcvLbbCYXBrlLo&s=iAYkX1gbtU0YrayVwSmog1bCLtdLWUiiGl2N-3qiIYE&e=>
when I got to MIT.
I remember wandering about Warren's lab at the Research Lab of Electronics with my friend, Stu Kauffman, who was then working with McCulloch:
https://en.wikipedia.org/wiki/Stuart_Kauffman<https://urldefense.proofpoint.com/v2/url?u=https-3A__en.wikipedia.org_wiki_Stuart-5FKauffman&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=ou936izPpFpChNwRZ7hsXjS2h1WcDB3Zn1SvpbrBqQJ2UGafMEGcvLbbCYXBrlLo&s=b_0FxVN_5eo_InE_82lk47ENvvj3pUqPuNHIPO0SLyo&e=>.
I met Stu at Dartmouth, where I began my work in neural networks as a freshman in 1957. We have remained close friends to the present time.
Jerry and his wife, Maggie, also looked after me and invited me to dinner parties at their home. Maggie became quite a famous person in her own right and was a major role model for exercise, health, and women in general:
https://en.wikipedia.org/wiki/Maggie_Lettvin<https://urldefense.proofpoint.com/v2/url?u=https-3A__en.wikipedia.org_wiki_Maggie-5FLettvin&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=ou936izPpFpChNwRZ7hsXjS2h1WcDB3Zn1SvpbrBqQJ2UGafMEGcvLbbCYXBrlLo&s=aE4ZPhrItmznYh2VFtJ75ODBFxICFXj4tf_eLQ11F3A&e=>
I will never forget the generosity and kindness of these incredibly talented people.
My Magnum Opus includes discussions of Bohr, Einstein, and McCulloch, among other great scientists:
https://www.amazon.com/Conscious-Mind-Resonant-Brain-Makes/dp/0190070552<https://urldefense.proofpoint.com/v2/url?u=https-3A__www.amazon.com_Conscious-2DMind-2DResonant-2DBrain-2DMakes_dp_0190070552&d=DwMGaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=ou936izPpFpChNwRZ7hsXjS2h1WcDB3Zn1SvpbrBqQJ2UGafMEGcvLbbCYXBrlLo&s=M2rv3-OFfWi0TDKoQG0GaMUZg6yERYtvaKnpqLmDal0&e=>
Best,
Steve
________________________________
From: Connectionists <connectionists-bounces at mailman.srv.cs.cmu.edu> on behalf of Gary Marcus <gary.marcus at nyu.edu>
Sent: Tuesday, January 17, 2023 1:35 PM
To: Michael Arbib <arbib at usc.edu>
Cc: connectionists at cs.cmu.edu <connectionists at cs.cmu.edu>
Subject: Re: Connectionists: Annotated History of Modern AI and Deep Learning
Wow. Chills down spine, in a good way. I did not know that and look forward to reading!
On Jan 17, 2023, at 10:32, Michael Arbib <arbib at usc.edu> wrote:
Now that Cybernetics has been brought into the conversation, and since I may be the only person who was both a PhD student of Norbert Wiener (for a while) and an RA for Warren McCulloch, I take the liberty of drawing attention to a memoir I wrote:
Arbib, M. A. (2018). From cybernetics to brain theory, and more: A memoir. Cognitive Systems Research, 50, 83-145.
A preprint is available on ResearchGate – just enter “Arbib ResearchGate Memoir” in your browser. There are ideas in there whose solution I still await ….
************************************
From: Connectionists <connectionists-bounces at mailman.srv.cs.cmu.edu> On Behalf Of Stephen José Hanson
Sent: Tuesday, January 17, 2023 5:13 AM
To: Sean Manion <stmanion at gmail.com>; Gary Marcus <gary.marcus at nyu.edu>
Cc: connectionists at cs.cmu.edu
Subject: Re: Connectionists: Annotated History of Modern AI and Deep Learning
Sean,
What a wonderfult find!
I believe this is most likely a precursor to the Macy meetings 1946-1953, which was run by McCulloch.
It included the wonderful list from Wiener of the areas that "have obtained a degree of intimacy".
These meetings became called by the attendees-- CYBERNETICS.. and of course is the precursor to AI and Neural Networks, computational neuroscience etc..
https://press.uchicago.edu/ucp/books/book/distributed/C/bo23348570.html<https://urldefense.com/v3/__https:/press.uchicago.edu/ucp/books/book/distributed/C/bo23348570.html__;!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOuilVTjVXw$>
thanks for sharing!
Steve
On 1/17/23 00:28, Sean Manion wrote:
Thank you all for a great discussion, and of course Jürgen for your work on the annotated history that has kicked it off.
For reasons tangential to all of this, I have been recently reviewing some of the MIT Archives and found this invitation from Wiener, von Neumann, and Aiken to several individuals for a sometimes historically overlooked 2 day meeting that was held at Princeton in January 1945 on a "...field of effort, which as yet is not even named."
I thought some might find this of interest.
Cheers!
Sean
On Mon, Jan 16, 2023 at 11:51 PM Gary Marcus <gary.marcus at nyu.edu<mailto:gary.marcus at nyu.edu>> wrote:
Hi, Juergen,
Thanks for your reply. Restricting your title to “modern” AI as you did is a start, but I think still not enough. For example, from what I understand about NNAISANCE, through talking with you and Bas Steunebrink, there’s quite a bit of hybrid AI in what you are doing at your company, not well represented in the review. The related open-access book certainly draws heavily on both traditions (https://link.springer.com/book/10.1007/978-3-031-08020-3<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Flink.springer.com*2Fbook*2F10.1007*2F978-3-031-08020-3&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=nzjxkTj5pTpbFiPk8QGz0HvRgVoDGlRJnyo9vGsuALU*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJQ!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOui_qF6Ftw$>).
Likewise, there is plenty of eg symbolic planning in modern navigation systems, most robots etc; still plenty of use of symbolic trees in game playing; lots of people still use taxonomies and inheritance, etc., an AFAIK nobody has built a trustworthy virtual assistant, even in a narrow domain, with only deep learning. And so on.
In the end, it’s really a question about balance, which is what I think Andrzej was getting at; you go miles deep on the history of deep learning, which I respect, but just give relatively superficial pointers (not none!) outside that tradition. Definitely better, to be sure, in having at least a few pointers than in having none, and I would agree that the future is uncertain. I think you strike the right note there!
As an aside, saying that everything can be formulated as RL is maybe no more helpful than saying that everything we (currently) know how to do can be formulated in terms of Turing machine. True, but doesn’t carry you far enough in most real world applications. I personally see RL as part of an answer, but most useful in (and here we might partly agree) the context of systems with rich internal models of the world.
My own view is that we will get to more reliable AI only once the field more fully embraces the project of articulating how such models work and how they are developed.
Which is maybe the one place where you (eg https://arxiv.org/pdf/1803.10122.pdf<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Farxiv.org*2Fpdf*2F1803.10122.pdf&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=ZyJcnldLGH1FWLfLy*2B1OaQogSmVjZdOKu05SVnmQgyo*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJQ!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOujbUxHYPQ$>), Yann LeCun (eg https://openreview.net/forum?id=BZ5a1r-kVsf<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Fopenreview.net*2Fforum*3Fid*3DBZ5a1r-kVsf&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=NciJImyaIct5w804QCbLcBdvC56Tb8s8oyS9Rn*2BGxhY*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJSU!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOugZ5NfTNg$>), and I (eg https://arxiv.org/abs/2002.06177<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Farxiv.org*2Fabs*2F2002.06177&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=xk1Vn44bpzaJS*2Bgxa2RrYNREA4i1O*2Bmf2vInHJc76KE*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJSU!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOugdXMd64Q$>) are most in agreement.
Best,
Gary
On Jan 15, 2023, at 23:04, Schmidhuber Juergen <juergen at idsia.ch<mailto:juergen at idsia.ch>> wrote:
Thanks for these thoughts, Gary!
1. Well, the survey is about the roots of “modern AI” (as opposed to all of AI) which is mostly driven by “deep learning.” Hence the focus on the latter and the URL "deep-learning-history.html.” On the other hand, many of the most famous modern AI applications actually combine deep learning and other cited techniques (more on this below).
Any problem of computer science can be formulated in the general reinforcement learning (RL) framework, and the survey points to ancient relevant techniques for search & planning, now often combined with NNs:
"Certain RL problems can be addressed through non-neural techniques invented long before the 1980s: Monte Carlo (tree) search (MC, 1949) [MOC1-5], dynamic programming (DP, 1953) [BEL53], artificial evolution (1954) [EVO1-7][TUR1] (unpublished), alpha-beta-pruning (1959) [S59], control theory and system identification (1950s) [KAL59][GLA85], stochastic gradient descent (SGD, 1951) [STO51-52], and universal search techniques (1973) [AIT7].
Deep FNNs and RNNs, however, are useful tools for _improving_ certain types of RL. In the 1980s, concepts of function approximation and NNs were combined with system identification [WER87-89][MUN87][NGU89], DP and its online variant called Temporal Differences [TD1-3], artificial evolution [EVONN1-3] and policy gradients [GD1][PG1-3]. Many additional references on this can be found in Sec. 6 of the 2015 survey [DL1].
When there is a Markovian interface [PLAN3] to the environment such that the current input to the RL machine conveys all the information required to determine a next optimal action, RL with DP/TD/MC-based FNNs can be very successful, as shown in 1994 [TD2] (master-level backgammon player) and the 2010s [DM1-2a] (superhuman players for Go, chess, and other games). For more complex cases without Markovian interfaces, …”
Theoretically optimal planners/problem solvers based on algorithmic information theory are mentioned in Sec. 19.
2. Here a few relevant paragraphs from the intro:
"A history of AI written in the 1980s would have emphasized topics such as theorem proving [GOD][GOD34][ZU48][NS56], logic programming, expert systems, and heuristic search [FEI63,83][LEN83]. This would be in line with topics of a 1956 conference in Dartmouth, where the term "AI" was coined by John McCarthy as a way of describing an old area of research seeing renewed interest.
Practical AI dates back at least to 1914, when Leonardo Torres y Quevedo built the first working chess end game player [BRU1-4] (back then chess was considered as an activity restricted to the realms of intelligent creatures). AI theory dates back at least to 1931-34 when Kurt Gödel identified fundamental limits of any type of computation-based AI [GOD][BIB3][GOD21,a,b].
A history of AI written in the early 2000s would have put more emphasis on topics such as support vector machines and kernel methods [SVM1-4], Bayesian (actually Laplacian or possibly Saundersonian [STI83-85]) reasoning [BAY1-8][FI22] and other concepts of probability theory and statistics [MM1-5][NIL98][RUS95], decision trees, e.g. [MIT97], ensemble methods [ENS1-4], swarm intelligence [SW1], and evolutionary computation [EVO1-7][TUR1]. Why? Because back then such techniques drove many successful AI applications.
A history of AI written in the 2020s must emphasize concepts such as the even older chain rule [LEI07] and deep nonlinear artificial neural networks (NNs) trained by gradient descent [GD’], in particular, feedback-based recurrent networks, which are general computers whose programs are weight matrices [AC90]. Why? Because many of the most famous and most commercial recent AI applications depend on them [DL4]."
3. Regarding the future, you mentioned your hunch on neurosymbolic integration. While the survey speculates a bit about the future, it also says: "But who knows what kind of AI history will prevail 20 years from now?”
Juergen
On 14. Jan 2023, at 15:04, Gary Marcus <gary.marcus at nyu.edu<mailto:gary.marcus at nyu.edu>> wrote:
Dear Juergen,
You have made a good case that the history of deep learning is often misrepresented. But, by parity of reasoning, a few pointers to a tiny fraction of the work done in symbolic AI does not in any way make this a thorough and balanced exercise with respect to the field as a whole.
I am 100% with Andrzej Wichert, in thinking that vast areas of AI such as planning, reasoning, natural language understanding, robotics and knowledge representation are treated very superficially here. A few pointers to theorem proving and the like does not solve that.
Your essay is a fine if opinionated history of deep learning, with a special emphasis on your own work, but of somewhat limited value beyond a few terse references in explicating other approaches to AI. This would be ok if the title and aspiration didn’t aim for as a whole; if you really want the paper to reflect the field as a whole, and the ambitions of the title, you have more work to do.
My own hunch is that in a decade, maybe much sooner, a major emphasis of the field will be on neurosymbolic integration. Your own startup is heading in that direction, and the commericial desire to make LLMs reliable and truthful will also push in that direction.
Historians looking back on this paper will see too little about that roots of that trend documented here.
Gary
On Jan 14, 2023, at 12:42 AM, Schmidhuber Juergen <juergen at idsia.ch<mailto:juergen at idsia.ch>> wrote:
Dear Andrzej, thanks, but come on, the report cites lots of “symbolic” AI from theorem proving (e.g., Zuse 1948) to later surveys of expert systems and “traditional" AI. Note that Sec. 18 and Sec. 19 go back even much further in time (not even speaking of Sec. 20). The survey also explains why AI histories written in the 1980s/2000s/2020s differ. Here again the table of contents:
Sec. 1: Introduction
Sec. 2: 1676: The Chain Rule For Backward Credit Assignment
Sec. 3: Circa 1800: First Neural Net (NN) / Linear Regression / Shallow Learning
Sec. 4: 1920-1925: First Recurrent NN (RNN) Architecture. ~1972: First Learning RNNs
Sec. 5: 1958: Multilayer Feedforward NN (without Deep Learning)
Sec. 6: 1965: First Deep Learning
Sec. 7: 1967-68: Deep Learning by Stochastic Gradient Descent
Sec. 8: 1970: Backpropagation. 1982: For NNs. 1960: Precursor.
Sec. 9: 1979: First Deep Convolutional NN (1969: Rectified Linear Units)
Sec. 10: 1980s-90s: Graph NNs / Stochastic Delta Rule (Dropout) / More RNNs / Etc
Sec. 11: Feb 1990: Generative Adversarial Networks / Artificial Curiosity / NN Online Planners
Sec. 12: April 1990: NNs Learn to Generate Subgoals / Work on Command
Sec. 13: March 1991: NNs Learn to Program NNs. Transformers with Linearized Self-Attention
Sec. 14: April 1991: Deep Learning by Self-Supervised Pre-Training. Distilling NNs
Sec. 15: June 1991: Fundamental Deep Learning Problem: Vanishing/Exploding Gradients
Sec. 16: June 1991: Roots of Long Short-Term Memory / Highway Nets / ResNets
Sec. 17: 1980s-: NNs for Learning to Act Without a Teacher
Sec. 18: It's the Hardware, Stupid!
Sec. 19: But Don't Neglect the Theory of AI (Since 1931) and Computer Science
Sec. 20: The Broader Historic Context from Big Bang to Far Future
Sec. 21: Acknowledgments
Sec. 22: 555+ Partially Annotated References (many more in the award-winning survey [DL1])
Tweet: https://urldefense.proofpoint.com/v2/url?u=https-3A__twitter.com_SchmidhuberAI_status_1606333832956973060-3Fcxt-3DHHwWiMC8gYiH7MosAAAA&d=DwIDaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=oGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5&s=nWCXLKazOjmixYrJVR0CMlR12PasGbAd8bsS6VZ10bk&e=<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Furldefense.proofpoint.com*2Fv2*2Furl*3Fu*3Dhttps-3A__twitter.com_SchmidhuberAI_status_1606333832956973060-3Fcxt-3DHHwWiMC8gYiH7MosAAAA*26d*3DDwIDaQ*26c*3DslrrB7dE8n7gBJbeO0g-IQ*26r*3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ*26m*3DoGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5*26s*3DnWCXLKazOjmixYrJVR0CMlR12PasGbAd8bsS6VZ10bk*26e*3D&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=bZfxmr1oj5alCicvKzzhXDN4q6855smPHdbBVbvcqSw*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSU!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOugHyPKUQw$>
Jürgen
On 13. Jan 2023, at 14:40, Andrzej Wichert <andreas.wichert at tecnico.ulisboa.pt<mailto:andreas.wichert at tecnico.ulisboa.pt>> wrote:
Dear Juergen,
You make the same mistake at it was done in the earlier 1970. You identify deep learning with modern AI, the paper should be called instead "Annotated History of Deep Learning”
Otherwise, you ignore symbolical AI, like search, production systems, knowledge representation, search, planning etc., as if is not part of AI anymore (suggested by your title).
Best,
Andreas
--------------------------------------------------------------------------------------------------
Prof. Auxiliar Andreas Wichert
https://urldefense.proofpoint.com/v2/url?u=http-3A__web.tecnico.ulisboa.pt_andreas.wichert_&d=DwIDaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=oGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5&s=h5Zy9Hk2IoWPt7me1mLhcYHEuJ55mmNOAppZKcivxAk&e=<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Furldefense.proofpoint.com*2Fv2*2Furl*3Fu*3Dhttp-3A__web.tecnico.ulisboa.pt_andreas.wichert_*26d*3DDwIDaQ*26c*3DslrrB7dE8n7gBJbeO0g-IQ*26r*3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ*26m*3DoGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5*26s*3Dh5Zy9Hk2IoWPt7me1mLhcYHEuJ55mmNOAppZKcivxAk*26e*3D&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=AceluKx*2B8KWGPMO06FCjnpyFyTKpTMbig7tM*2BhOG*2FUg*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSU!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOugJDXOm1w$>
-
https://urldefense.proofpoint.com/v2/url?u=https-3A__www.amazon.com_author_andreaswichert&d=DwIDaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=oGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5&s=w1RtYvs8dwtfvlTkHqP_P-74ITvUW2IiHLSai7br25U&e=<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Furldefense.proofpoint.com*2Fv2*2Furl*3Fu*3Dhttps-3A__www.amazon.com_author_andreaswichert*26d*3DDwIDaQ*26c*3DslrrB7dE8n7gBJbeO0g-IQ*26r*3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ*26m*3DoGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5*26s*3Dw1RtYvs8dwtfvlTkHqP_P-74ITvUW2IiHLSai7br25U*26e*3D&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=*2F*2FDhQhTwI7XGRKbBkGIi8*2BuOWxknWbBpspj5YPwUbTU*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSU!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOugvtzNUdg$>
Instituto Superior Técnico - Universidade de Lisboa
Campus IST-Taguspark
Avenida Professor Cavaco Silva Phone: +351 214233231
2744-016 Porto Salvo, Portugal
On 13 Jan 2023, at 08:13, Schmidhuber Juergen <juergen at idsia.ch<mailto:juergen at idsia.ch>> wrote:
Machine learning is the science of credit assignment. My new survey credits the pioneers of deep learning and modern AI (supplementing my award-winning 2015 survey):
https://urldefense.proofpoint.com/v2/url?u=https-3A__arxiv.org_abs_2212.11279&d=DwIDaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=oGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5&s=6E5_tonSfNtoMPw1fvFOm8UFm7tDVH7un_kbogNG_1w&e=<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Furldefense.proofpoint.com*2Fv2*2Furl*3Fu*3Dhttps-3A__arxiv.org_abs_2212.11279*26d*3DDwIDaQ*26c*3DslrrB7dE8n7gBJbeO0g-IQ*26r*3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ*26m*3DoGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5*26s*3D6E5_tonSfNtoMPw1fvFOm8UFm7tDVH7un_kbogNG_1w*26e*3D&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=e*2Bf09tNf3tO8tY2jANTGEUnAipIJum92za3z*2FNZMbaw*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJQ!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOuhyBWAubA$>
https://urldefense.proofpoint.com/v2/url?u=https-3A__people.idsia.ch_-7Ejuergen_deep-2Dlearning-2Dhistory.html&d=DwIDaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=wQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ&m=oGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5&s=XPnftI8leeqoElbWQIApFNQ2L4gDcrGy_eiJv2ZPYYk&e=<https://urldefense.com/v3/__https:/nam02.safelinks.protection.outlook.com/?url=https*3A*2F*2Furldefense.proofpoint.com*2Fv2*2Furl*3Fu*3Dhttps-3A__people.idsia.ch_-7Ejuergen_deep-2Dlearning-2Dhistory.html*26d*3DDwIDaQ*26c*3DslrrB7dE8n7gBJbeO0g-IQ*26r*3DwQR1NePCSj6dOGDD0r6B5Kn1fcNaTMg7tARe7TdEDqQ*26m*3DoGn-OID5YOewbgo3j_HjFjI3I2N3hx-w0hoIfLR_JJsn8q5UZDYAl5HOHPY-87N5*26s*3DXPnftI8leeqoElbWQIApFNQ2L4gDcrGy_eiJv2ZPYYk*26e*3D&data=05*7C01*7Cjose*40rubic.rutgers.edu*7Cb3db76bd41ff4a59071908daf85da95a*7Cb92d2b234d35447093ff69aca6632ffe*7C1*7C0*7C638095378420227407*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C2000*7C*7C*7C&sdata=cGUbD8*2Fc31LU5nO7Oc1I7b3UERXBn2GArYhEO1IE1Bc*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUl!!LIr3w8kk_Xxm!rhOV1_I3SVtEWzbHAOD-b63GOlQ_CnsGhyQR9CixHIagtitUuZEV-Lr8y549an4wJVQrYiLaOuiTz43jzQ$>
This was already reviewed by several deep learning pioneers and other experts. Nevertheless, let me know under juergen at idsia.ch<mailto:juergen at idsia.ch> if you can spot any remaining error or have suggestions for improvements.
Happy New Year!
Jürgen
--
Stephen José Hanson
Professor, Psychology Department
Director, RUBIC (Rutgers University Brain Imaging Center)
Member, Executive Committee, RUCCS
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20230118/0ddda112/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: apple-touch-icon-1582743354193.png
Type: image/png
Size: 2569 bytes
Desc: apple-touch-icon-1582743354193.png
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20230118/0ddda112/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: apple-touch-icon-1582743354193.png
Type: image/png
Size: 2569 bytes
Desc: apple-touch-icon-1582743354193.png
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20230118/0ddda112/attachment-0001.png>
More information about the Connectionists
mailing list