[javawizard]

I'll say as an ex-FAANG that recently released a similar stat that it was absolutely real, but only because doing anything required an atrocious amount of boilerplate that AI was really good at figuring out.

Compared to how productive I am at my current startup? Not a chance.

[namaria]

Well put. LLMs seem great in environments full of the cruft LLMs are trained on. Seems obvious but I hadn't put it together. Thanks for the insight.