2023: RL agent trained for multi-task learning solves majority of perfect information games. It's a scaled up decision transformer. Scaling laws for RL agents are discovered, similar to language models.
2024: Large scale RL agents are combined with frozen vision and language models via cross-attention, can be prompted one-shot with language/vision tokens to solve novel tasks.
2025: RL agents enter the real world - first pre-trained in diverse synthetic environments, then via imitation learning from youtube videos, and finally in an online fashion via realtime human interaction.
timeline might be optimistic, but one can hope!
I'm interested to see how the field advances, but it won't lead to AGI, it will lead to cool tricks that the ignorant think are sufficient to replace a real person. That will suck
Still waiting for a followup to https://arxiv.org/abs/2104.03113 ...
Of course, it's not actually performing introspection, and it's just lucky that it guessed the right answer here. Perhaps it's just learned that when conversations discuss a general case (how do humans perform) and then turn to a specific case (how about you?), there is typically some difference between the two that should be noted. But it still gives an illusion of an unbelievable capability.
