Séb Krier, AGI Policy Development Lead at Google DeepMind, argues frontier AI will likely become a low-margin modular component

This is why it's vital that AI remains democratized. Right now, competition among frontier labs + open weight models means that most value accrues to customers. If AI becomes monopolized, through network effects or regulation, we end up in a more unequal and dangerous world.

@sebkrier @AtlantisPleb 💯% that’s exactly the current fork in the road. My sense is that relevant knowledge is still distributed enough (for now) to make your second scenario possible.

Indeed and we might get both. We might be on our way to the bifurcation we saw with high performance computing and everything else.

HPC was sort of hidden in national labs with very restricted access to everyone. Whereas normal computers sort of diffused slowly with very different form factors for each use case. From transistor radios to calculators to mainframes in banks and finally to personal computers. Only in this case we have sort of personal AIs first and AI native commercial products are still slowly coming (glasses?, ai email?)

@sebkrier High margin? How?