Great stuff Ronnie and excellent questions Matt & Cris. Kudos. Cool and FUN stuff.

Great update notes, I hope to see more real-life examples.

It's becoming clear that with all the brain and consciousness theories out there, the proof will be in the pudding. By this I mean, can any particular theory be used to create a human adult level conscious machine. My bet is on the late Gerald Edelman's Extended Theory of Neuronal Group Selection. The lead group in robotics based on this theory is the Neurorobotics Lab at UC at Irvine. Dr. Edelman distinguished between primary consciousness, which came first in evolution, and that humans share with other conscious animals, and higher order consciousness, which came to only humans with the acquisition of language. A machine with only primary consciousness will probably have to come first. What I find special about the TNGS is the Darwin series of automata created at the Neurosciences Institute by Dr. Edelman and his colleagues in the 1990's and 2000's. These machines perform in the real world, not in a restricted simulated world, and display convincing physical behavior indicative of higher psychological functions necessary for consciousness, such as perceptual categorization, memory, and learning. They are based on realistic models of the parts of the biological brain that the theory claims subserve these functions. The extended TNGS allows for the emergence of consciousness based only on further evolutionary development of the brain areas responsible for these functions, in a parsimonious way. No other research I've encountered is anywhere near as convincing. I post because on almost every video and article about the brain and consciousness that I encounter, the attitude seems to be that we still know next to nothing about how the brain and consciousness work; that there's lots of data but no unifying theory. I believe the extended TNGS is that theory. My motivation is to keep that theory in front of the public. And obviously, I consider it the route to a truly conscious machine, primary and higher-order. My advice to people who want to create a conscious machine is to seriously ground themselves in the extended TNGS and the Darwin automata first, and proceed from there, by applying to Jeff Krichmar's lab at UC Irvine, possibly. Dr. Edelman's roadmap to a conscious machine is at arxiv.org/abs/2105.10461

54 min comment - Yeah you can embed from variables if you loop. Insert Embedding for Found Set does not allow using a variable but Insert Embedding does allow you to embed data from a variable.

Your predictions were pretty accurate.

Great video . Discord link seems to be expired

From the perspective of a psychiatrist and neuroscientist, the AI is demonstrating clear empirical evidence for cognitive empathy and theory of mind, which is remarkable. I wouldn't expect it to experience (!) emotional empathy, unless maybe the transformer model was upgraded to include reinforcement learning, which could be the Q* model.

Great video, no idea why there are so few comments.

Very cool

Great video. Very cool use case.

It's becoming clear that with all the brain and consciousness theories out there, the proof will be in the pudding. By this I mean, can any particular theory be used to create a human adult level conscious machine. My bet is on the late Gerald Edelman's Extended Theory of Neuronal Group Selection. The lead group in robotics based on this theory is the Neurorobotics Lab at UC at Irvine. Dr. Edelman distinguished between primary consciousness, which came first in evolution, and that humans share with other conscious animals, and higher order consciousness, which came to only humans with the acquisition of language. A machine with only primary consciousness will probably have to come first. What I find special about the TNGS is the Darwin series of automata created at the Neurosciences Institute by Dr. Edelman and his colleagues in the 1990's and 2000's. These machines perform in the real world, not in a restricted simulated world, and display convincing physical behavior indicative of higher psychological functions necessary for consciousness, such as perceptual categorization, memory, and learning. They are based on realistic models of the parts of the biological brain that the theory claims subserve these functions. The extended TNGS allows for the emergence of consciousness based only on further evolutionary development of the brain areas responsible for these functions, in a parsimonious way. No other research I've encountered is anywhere near as convincing. I post because on almost every video and article about the brain and consciousness that I encounter, the attitude seems to be that we still know next to nothing about how the brain and consciousness work; that there's lots of data but no unifying theory. I believe the extended TNGS is that theory. My motivation is to keep that theory in front of the public. And obviously, I consider it the route to a truly conscious machine, primary and higher-order. My advice to people who want to create a conscious machine is to seriously ground themselves in the extended TNGS and the Darwin automata first, and proceed from there, by applying to Jeff Krichmar's lab at UC Irvine, possibly. Dr. Edelman's roadmap to a conscious machine is at arxiv.org/abs/2105.10461

Great video Cris. My AI summary of your discussion. This is in-depth conversation between Dr. Tamara Russell, a clinical psychologist and neuroscientist, and Chris, an AI developer. Their interdisciplinary discussion explores the question of whether state-of-the-art AI systems like GPT-4 have theory of mind - the ability to infer others’ mental states like beliefs, desires, and intentions. This is a pivotal social capability that develops during early childhood in humans. Dr. Russell brings expertise from psychology and neuroscience research on theory of mind. She explains foundational concepts such as first-order theory of mind, which involves understanding that others can have beliefs different from one’s own (capturing ideas like “he thinks”). Second-order theory of mind involves the ability to consider nested perspectives and juggle multiple minds at once (e.g. “he thinks that she thinks”). Meanwhile, Chris shares details of testing he has conducted evaluating theory of mind capabilities in AI systems, particularly GPT-4. This has involved administering psychological assessments like false belief tasks, where the AI demonstrated performance at the level of a 7-year-old child. However, Dr. Russell notes limitations of relying solely on verbal tests rather than multi-modal inputs. Much of the dialogue centers around a key debate - does GPT-4 truly understand emotions, mental states, and social dynamics, or does it merely have a vast amount of factual knowledge to draw upon? Dr. Russell suggests the AI may have a “cold” theory of mind without the nuanced feeling and consciousness of human cognition. She explains embodied human cognition develops through experiencing dynamics like emotion and intention. Chris provides technical perspective into how GPT-4 uses probability to make predictions based on textual training data, rather than having “true” understanding. While both agree GPT-4 has impressive capabilities for mimicking human-like conversation and reasoning, further research is needed to determine if AI could develop more advanced, multifaceted theory of mind approaching human-level cognition. Testing solely language-based comprehension provides useful but limited insight. Dr. Russell highlights the need for rigorous interdisciplinary hypotheses applying insights from neuroscience and psychology to test additional facets of cognition related to theory of mind, such as integrating emotional data (“hot” theory of mind). This discussion illustrates the vital role collaboration between AI and other disciplines can play in illuminating the complex differences between artificial and human intelligence. While GPT-4 demonstrates a capability approaching that of a precocious child in verbal reasoning about beliefs and falsehoods, it remains unclear whether its fundamental approach can capture the embodied consciousness and advanced social reasoning underlying mature adult human theory of mind. However, focused research probing where gaps exist could further our understanding of both human and artificial cognition. Overall, Dr. Russell and Chris find common ground that while extremely impressive in some narrow capacities, current AI likely lacks full contextual awareness, emotional sophistication, and general consciousness characteristic of human-like theory of mind. This represents an insightful dialogue across disciplines highlighting limitations as well as future possibilities, and the need for nuanced collaboration evaluating AI progress toward profound human social abilities like theory of mind. Their conversation provides a model for interdisciplinary exploration of the contrasts between human and artificial intelligence.

Interesting to hear your perspective since you had an early look at this incredible technology. Good on you for grasping the magnitude of the power of GPT 3 right away which is a unique skill.

I am an clinical psychologist with expertise in theory of mind (!). It's not surprising that it can work out perspective taking as there are various "long hand" ways to do this (called "cold" theory of mind"). more curious is the variability in the empathic (or hot) theory of mind which involves the more affective or emotional aspects.... great experiment.

poor video 0 focus on actual accurancy comparition of an answers from both bots. just brainless clicking

Promo>SM

Hello, can you share the link of the API so I can emulate your project? Thanks

Thanks again Cris.

Interesting experiment, thanks

This is a wonderful series on the "Theory of Mind", Thank you!

Interesting. Why do you think GPT 4 actually produces so much noise?

Great comparison. Thanks

To be honest I am not sure what I am seeing in this video. Is this the new API? When I chat with the new model I do not see pictures like you are seeing. I admit I am probably doing something wrong.

Thanks Cris!

Great summary. Love the show your work aspect of gpt4. Great to see your various scenarios across different versions. Well done!

Great summary! I can't wait to play and see what is new and different in this update.

Good stuff. I too have been pretty amazed with how quickly my little test python scripts have been receiving responses from openai on the chatgpt-3.5-turbo model. And not having the 'simulated typing' slowdown mechanism is a real boon to productivity too ;) New sub and looking forward to seeing what else you come up with.

Thank you.

Hello, the chat gpt API has a rate limit of tokens? Like davinci is 4000? Thank you

Great vid 👍! I'm still looking for an example however, where ChatGPT3 Turbo API is used to access the entire Internet? Or just a single website for texts ingesting?

Thanks!

can you show how to migrate the code from text-davinci-003 to gpt-3.5-turbo pls?

Thank you for performing the comparison. I am curious about the impact of the 'system' and 'temperature' parameters on the sentence structure generated by ChatGPT. I assume that slight adjustments to these parameters would bring the output of both models closer together. It would also be interesting to see how the API for both models handles some simple encoding.

Another good video, thx

Thanks for making this happen my friend! This was a lot of fun and really enjoyed meeting everyone.

I know I am big time when I made your youtube channel!

The new hacking, cool example.

Great course Cris!