Comment by PrayagBhakar

But isn’t this also a step towards AGI? The model being able to find issues and self correct?