Sam Altman, OpenAI CEO, believes that a whole generation of researchers held AI back by overestimating what scaling couldn’t do. He’s pushing back against skeptics like Yann LeCun, who claim LLMs are a dead end. Altman argues that LLMs have already surpassed human intelligence in some areas, disproving mathematical conjectures and gaining new knowledge. However, for long-horizon tasks requiring high judgment, LLMs seem to fall short. Despite critics, Altman remains confident in the potential of scaling large language models. Why it matters: Altman’s stance highlights the ongoing debate about the capabilities of LLMs and the importance of considering the potential risks and benefits of continued scaling.