Again, it doesn't say much about how good a model any given system might have. The world is much more complicated than an Othello board. GPT-4 is much bigger than their transformer model. Everything they found is consistent with anything from "as it happens GPT-4 has no world model at all" through to "GPT-4 has a rich model of the world, fully comparable to ours". (I would bet heavily on the truth being somewhere in between, not that that says very much.)
I don't follow it too closely, but I've seen papers on mechanistic interpretability that look promising.
I just asked GPT-4 "What would happen to Northern Thailand if elephants behave like kangaroos?".
The answers are probably better than what 90+% of humans could give after spending an hour researching on the internet.
The Sparks of AGI paper provides much more evidence and examples: https://arxiv.org/abs/2303.12712
Language can represent thoughts and some world models. There is strong evidence that LLMs contain some representation of world models it learned from text. Moreover, LLM is already a misnomer; latest versions are multimodal. Current versions can be used to build agents with limited autonomy. Future versions of LLMs are most likely capable of more independence.
Even dumb viruses have caused catastrophic harm. Why? It’s capable of rapid self replication in a massive number of existing vessels. You add in some intelligence, vast store of knowledge, huge bandwidth, and some aid by malicious human actors, what could such a group of future autonomous agents do?
More on risks of “doom” by a top researcher on AI risk here: https://www.lesswrong.com/posts/xWMqsvHapP3nwdSW8/my-views-o...