Sam Altman says a generation of researchers held AI back by underestimating the power of scaling, according to The Decoder. The comment revisits one of the central debates in modern AI: whether bigger models and more compute were an obvious path or a surprising break from older assumptions.
The argument matters because scaling remains the economic engine behind frontier AI. Belief in continued returns supports enormous infrastructure spending, while skepticism pushes researchers toward efficiency, architecture, and data-quality alternatives.
The lesson for the industry may be less about declaring one side right and more about understanding how research priors shape investment decisions long before benchmarks settle the question.