These models are static snapshots of an arbitrary level of intelligence. Are there any small LLMs which are not static and can keep updating in real time with external or internal feedback?
That talk[1] (also paper[2]) covering some of this by the Microsoft team involved with the unfettered GPT-4 model certainly shifted my views last year at a time when until then it was more common to see downplaying about LLMs based on prior gen models. Recently saw an AI safety specialist who also become more concerned about the rapidity of progress cite their research.
Proving theory of mind in LLMs is certainly one step toward recognizing that they are well on the path of acquiring "human-like" intelligence - if there are still people around who doubt that (proof of which would be if the ensuing discussion centers around if this is still just guessing the next token or whether hallucinations are proof of permanent unintelligence, etc).
Theory of mind is obviously coupled with the recognition that we can be outsmarted - so this type of research is crucial in order to gauge what stage we are on.
I liked this discussion from a while back with Tim Lee and Robert Wright talking about ToM: https://youtu.be/yAJkmwn8jDo?si=3_9cSdqQBriM58SA&t=2488 (particularly the bit about how the model correctly called a complex case of schadenfreude)
5 comments
[ 3.0 ms ] story [ 23.6 ms ] thread[1] https://www.youtube.com/watch?v=qbIk7-JPB2c
[2] https://arxiv.org/abs/2303.12712
Theory of mind is obviously coupled with the recognition that we can be outsmarted - so this type of research is crucial in order to gauge what stage we are on.