[lucasay]

This feels less like automated research and more like structured trial and error with a decent feedback loop. Still useful, but I think the real bottleneck is how good your eval metric is. If that’s weak, the whole loop just optimizes for the wrong thing faster.

[Almondsetat]

Designing a good fitness function, a tale as old as time...

[kridsdale1]

I mean, isn’t that “the scientific method”?

[lucasay]

Partially—but science also questions the hypothesis and the metric. This mostly assumes both are correct and just optimizes within that box.

[svnt]

Only if the model is actually a human or equivalent, otherwise we don’t know what it is.