In a recent talk at Stanford University, Sam Altman shared his views on the future of artificial intelligence and progress towards Artificial General Intelligence. The OpenAI CEO explained why ChatGPT may soon be outdated as newer AI models get introduced.
During the engaging discussion, Altman hinted that ChatGPT, although revolutionary for its time, will be superseded by much more capable systems. He called it the “dumbest model” people will use, implying its short shelf life. Altman expects every new iteration from OpenAI to significantly outperform previous ones on intelligence.
When asked about costs associated with GPT-3, Altman expressed confidence in humanity’s ingenuity to leverage advanced tools for beneficial purposes. Rather than costs, he emphasized placing bets on societal value creation through advances in AI.
On upcoming models, Altman was optimistic about GPT-5 surpassing GPT-4 in capabilities. Highlighting ongoing exponential gains, he remarked it is remarkable how each new model promises to be smarter through continued research and engineering efforts.
Perhaps most remarkably, Altman stressed his willingness to invest billions annually to progress artificial general intelligence, provided it puts them on a trajectory to ultimately generate far more value than costs. For Altman, developing AGI is “totally worth it” due to its transformational potential if done responsibly.
