A proposed AGI marker centers on problems solvable by humans but not current AI systems, with gap closure indicating AGI arrival
Greg Kamradt quoted the criterion on X.
#1182Greg Kamradt@GREGKAMRADTWhen this gap* closes it means we have AGI
* a class of problems humans can solve but AI cannot
Dwarkesh Patel@dwarkesh_spCurrently it is shocking and newsworthy when AIs solve an important open problem that humans couldn't Before AI totally surpass us intellectually, there will be an interesting era, where it will be just as shocking (but not impossible) for a human to solve a problem AI couldn't
#1182Greg Kamradt@GREGKAMRADT“But Greg, we can RL on any problem”
Ok but who is gathering the training data, making the harness, picking the problem and judging outcomes?
Greg Kamradt@GregKamradtWhen this gap* closes it means we have AGI * a class of problems humans can solve but AI cannot
And to note, ofc these limitations will go away too
Greg Kamradt@GregKamradt“But Greg, we can RL on any problem” Ok but who is gathering the training data, making the harness, picking the problem and judging outcomes?