Artificial General Intelligence (AGI), a theoretical AI that could reason on par with humans, remains shrouded in uncertainty. However, leading industry researchers suggest we’re approaching a form of AGI within the next few years.
Miles Brundage, OpenAI’s former head of policy research and AGI readiness, recently shared on the Hard Fork podcast that in the coming years, the industry might see “systems that can basically do anything a person can do remotely on a computer.” This could include actions like navigating a mouse, typing on a keyboard, or even appearing convincingly human in video calls. Brundage stressed that governments should start preparing for the economic and educational shifts that such advancements would demand.
The question of when AGI will arrive is an intense debate within tech circles. Key figures like John Schulman, an OpenAI co-founder and research scientist until recently, believe AGI could emerge in a few years. Dario Amodei, CEO of Anthropic (an OpenAI competitor), speculates that some version of AGI could be realized by 2026.
Brundage, who announced his departure from OpenAI last month after six years, has valuable insights into the company’s AGI trajectory. He clarified that safety concerns didn’t drive his decision to leave. “I’m pretty confident that no other lab is completely ‘on top of things,’” he remarked, suggesting that the field as a whole is navigating uncharted territory.
Explaining his choice to step away, Brundage, now interested in the nonprofit sector, emphasized a desire to broaden his impact on cross-industry issues, like AGI regulation. “I wanted to focus on not just what we do internally at OpenAI but also on broader regulations,” he said. Additionally, he wanted independence from corporate influence: “I didn’t want my views dismissed as just corporate hype,” Brundage added, underlining his commitment to a balanced perspective on AGI’s future.
