Musk backs claim that AI progress is not slowing, says recursive self-improvement is nearing

Elon Musk on July 30 endorsed comments from former OpenAI employee will depue, pushing back on a bearish view of frontier lab valuations and the idea that progress in leading AI models is starting to slow. Depue said that skepticism toward frontier lab valuations may make sense if AGI is excluded from the picture, but argued that AI capability gains are not decelerating. He added that recursive self-improvement, or RSI, is likely to arrive in less than 24 months. Musk echoed that view and framed it in stark terms, saying that AI from two years ago already feels old enough to belong in a museum, which in his view makes it obvious that AI two years from now will be “mind-blowing.” His remarks amount to a direct response to recent arguments that frontier-model development is losing momentum and that top labs may struggle to achieve durable profitability over the long run.