Paper argues that diminishing returns to AI scale are an illusion Economic value comes from completing long projects, not single questions. And accuracy drives how long a project AI does: small gains compound exponentially! Reasoners are much more accurate, with big impacts.

→ View original post on X — @emollick,