But where is mention of world models in AI 2027 and 2040 or in your summary? Don’t all the robots and AIs driving them need to understand the physical world before taking it over?
What exactly do you mean by world models, and why do you think they’re necessary? I have some initial thoughts on what I think you mean (e.g. I can imagine a future in which an AI-assisted coder or AI coder develops a program that can model external environments, which can then be used by LLMs that themselves don’t have world models) but I want to make sure we’re talking about the same thing first.
My thought is that LLMs can probably program physical “muscle memory” for themselves with basic multimodal inputs + tool calling capability then control them just fine. The muscle memory themselves would perhaps need a world model to execute akin to a brain’s motor cortex but those models probably don’t need any capability beyond functioning as a motor cortex, i.e. “follow simple instructions and forward important surprises to the LLM”, so there’s not really much doubt that they can be developed.
Thank you for the summary,
But where is mention of world models in AI 2027 and 2040 or in your summary? Don’t all the robots and AIs driving them need to understand the physical world before taking it over?
Thanks in advance!
What exactly do you mean by world models, and why do you think they’re necessary? I have some initial thoughts on what I think you mean (e.g. I can imagine a future in which an AI-assisted coder or AI coder develops a program that can model external environments, which can then be used by LLMs that themselves don’t have world models) but I want to make sure we’re talking about the same thing first.
My thought is that LLMs can probably program physical “muscle memory” for themselves with basic multimodal inputs + tool calling capability then control them just fine. The muscle memory themselves would perhaps need a world model to execute akin to a brain’s motor cortex but those models probably don’t need any capability beyond functioning as a motor cortex, i.e. “follow simple instructions and forward important surprises to the LLM”, so there’s not really much doubt that they can be developed.