Part 5/9:
Sam Altman discusses how each advancement accelerates the journey toward full superintelligence. For example, GPT-4 implemented reasoning capabilities that were previously absent, and the introduction of O1 represents a quantum leap in reasoning power. Altman believes that, based on current trends, the leap from Level 2 (Reasoning) to Level 3 (Agency) could happen relatively soon, possibly within just a few years.
He characterizes this progression as a steep improvement curve, where models rapidly overcome current limitations, leading to new, unforeseen applications. Altman sees this as a pivotal era—the “GPT two stage”—and anticipates that in the near future, AI models will evolve into tools that act autonomously, managing complex tasks across many domains.