The Center of AI Has Moved, Academia Has Not
AI in academia is not bottlenecked by compute alone.
It is also bottlenecked by a publication culture that turns too much research into signaling, hedging, and procedural theater.
Too many smart people are spending their best hours optimizing for reviewer taste, rebuttal survival, benchmark cosmetics, and rejection avoidance instead of asking bigger questions or building deeper ideas.
A field slows down when novelty becomes harder to publish than to kill.
The problem is not just scarce resources. It is a low-trust, gatekeeping equilibrium that taxes ambition and rewards defensibility.
A lot of talent is not being used to create knowledge. It is being used to survive the venue.