Andrej Karpathy just admitted OpenAI’s own researchers feel the same career anxiety we do — his actual reasoning is more useful than the doom headlines

Watching a former Tesla AI Director shrug and say "I can't tell if that's temporary, I'm not sure how I feel about it yet" did something to me.

Usually it's the junior guy admitting that.

Not the guy who helped build the thing.

The people who end up fine here aren't the loudest about how safe their job is.

They're just already standing close enough to the mechanism to redirect it, instead of getting redirected by it.

Karpathy's actual point isn't doom.

It's the Jevons paradox — code gets cheaper, so total demand for it goes up.

Just not for the same kind of engineer who got hired in 2019.

I watched a version of this play out years ago, before any of this AI stuff existed. I was the technical guy in a construction tender department. Rule-based work — you follow A, you get B. A sub-contractor came in to pitch his quotation. On his way out, in the corridor, we locked eyes and instantly recognized each other. I knew him — my senior once told me how this guy forced his way into building an illegal bungalow, moving the boundary survey line onto his neighbour's land. I caught a flicker of panic on his face. He wasn't expecting to see me there.

I couldn't keep it to myself. I walked straight to my contract department and told them. They wrote him off after their own investigation.

Our technical and contractual work was rule-based — AI eats that easily. What I did with that information wasn't. Insider judgment, only humans have.

Actually — this is the same mechanism as a former SpaceX CIO's take on headcount compression, just proven with the actual numbers.

Clip credit: No Priors — full video on their channel. DM for credit or removal requests.

What would you have done in that corridor? Drop your take. 👇