Beautiful graphs, except I think the Pareto frontier works differently. You have the y-axis intercept for humans starting around $8. Even if the task is trivial, you need to pay some fixed cost to get a human to start working on it. That’s a fair assumption.
As you from Haiku->Sonnet->Opus the y-axis intercept should also be increasing. For cheap tasks it is more economical to run Haiku than Opus. You want Opus to start out “to the right” of Haiku, but increase faster. This implies of course the curves will intersect at some point (which makes sense).
Note that I say “as of the release of various frontier AIs”. The pareto frontier includes the possibility of using cheaper AIs. I maybe should have mentioned this in the post in a footnote. From this perspective the AI pareto frontier only strictly improves (putting aside AIs being de-deployed etc).
Beautiful graphs, except I think the Pareto frontier works differently. You have the y-axis intercept for humans starting around $8. Even if the task is trivial, you need to pay some fixed cost to get a human to start working on it. That’s a fair assumption.
As you from Haiku->Sonnet->Opus the y-axis intercept should also be increasing. For cheap tasks it is more economical to run Haiku than Opus. You want Opus to start out “to the right” of Haiku, but increase faster. This implies of course the curves will intersect at some point (which makes sense).
Note that I say “as of the release of various frontier AIs”. The pareto frontier includes the possibility of using cheaper AIs. I maybe should have mentioned this in the post in a footnote. From this perspective the AI pareto frontier only strictly improves (putting aside AIs being de-deployed etc).