Users with the hardest problems will use the smartest model available. So if you release your best model and your competitor only releases their 2nd smartest, then you will get better training data from users. Not just code, but vision and robotics too.
Among those affected by AI job loss is AI/ML devs/researchers themselves. The higher the hiring bar for AI devs, (A) the more applicants will put effort into learning stuff deeply and doing interesting useful novel work to get a job and (B) the easier it is to spot/snag unusual talent.
FWIW, my experience was that the utility of user data was always much higher in promise than in actual outcomes. This might have changed over time though.
There’s a lot of “they used user data to shoot themselves in the foot” and not nearly enough “they used user data to improve performance” happening in the industry.
Maybe frontier labs will finally crack applying user feedback once the training data bottleneck begins to bite? I imagine that getting good utility out of user data is hard, both from the standpoint of the engineering required, and the computation required.
Factor 1 is true only to the degree that the models cannot effectively generate hard problems for themselves to solve. If they can generate problems with verifiable rewards just at the edge of their capabilities, similar to AlphaZero, I expect these to be more useful than human generated problems. E.g. there are a lot of incredibly difficult unsolved conjectures in math which provide no RL gradients for labs to train on.
Two AI accelerants I had not noticed before:
Users with the hardest problems will use the smartest model available. So if you release your best model and your competitor only releases their 2nd smartest, then you will get better training data from users. Not just code, but vision and robotics too.
Among those affected by AI job loss is AI/ML devs/researchers themselves. The higher the hiring bar for AI devs, (A) the more applicants will put effort into learning stuff deeply and doing interesting useful novel work to get a job and (B) the easier it is to spot/snag unusual talent.
FWIW, my experience was that the utility of user data was always much higher in promise than in actual outcomes. This might have changed over time though.
There’s a lot of “they used user data to shoot themselves in the foot” and not nearly enough “they used user data to improve performance” happening in the industry.
Maybe frontier labs will finally crack applying user feedback once the training data bottleneck begins to bite? I imagine that getting good utility out of user data is hard, both from the standpoint of the engineering required, and the computation required.
Factor 1 is true only to the degree that the models cannot effectively generate hard problems for themselves to solve. If they can generate problems with verifiable rewards just at the edge of their capabilities, similar to AlphaZero, I expect these to be more useful than human generated problems. E.g. there are a lot of incredibly difficult unsolved conjectures in math which provide no RL gradients for labs to train on.