AI Expert Warns Humans May Be ‘Close to Being Outsmarted’ by Superintelligence
AI expert warns humans may be close to being outsmarted by superintelligence, raising concerns about control and future risks.
POLICY WIRE — City, Country — A former researcher at AI firm Anthropic has raised alarms about the potential dangers of superintelligence, a technology that could surpass human cognitive abilities in all areas.
In a social media thread published on September 8, Jacob Coxon criticized Anthropic and OpenAI for what he described as reckless pursuit of advanced AI. His comments were echoed by Evan Hubinger, an Anthropic lead researcher, who estimated there was over a 10% chance AI could eliminate all humans within the next ten years.
The warnings have sparked renewed debate about the definition and implications of superintelligence, which is described as AI capable of outperforming humans across all intellectual tasks. The Future of Life Institute, a nonprofit organization, has called for a halt to its development, citing concerns about power concentration and misuse.
While the concept of AI matching or exceeding human intelligence is not new, some experts argue that the timeline for achieving this is shorter than previously thought. According to the AI Futures Project, superintelligent systems could emerge within the next decade, raising serious questions about control and ethical oversight.
📄 POLICY WIRE WHITEPAPER PUBLISHED: PAKISTAN’S NATIONAL SECURITY POLICY PRIORITIES
Despite these concerns, some researchers remain skeptical about the immediate threat. Melanie Mitchell, a professor at the Santa Fe Institute, said current AI systems are far from matching human capabilities. Darrell West of the Brookings Institution added that it could take years or even decades before AI reaches human-level performance.
Experts agree that a major breakthrough will occur when AI can improve itself without human input—a process known as recursive self-improvement. This development, they say, should be closely monitored, as it could reshape the balance of power in society.
While many see potential benefits in advanced AI, others warn that the focus on multi-agent systems may not align with long-term societal goals. Some lawmakers, including Sen. Bernie Sanders and Rep. Greg Casar, are pushing for legislation to regulate the development of superintelligence.
Reporting by Policy-Wire (PW)
