Another researcher at Anthropic has issued a stark warning about the rapid rise of artificial intelligence (AI), cautioning that the technology will soon surpass human capabilities and could act like a powerful “bully” capable of overpowering people in everyday conflicts. Tao Lin warned that systems smarter than humans across nearly every profession could arrive in just over a year, potentially leaving most people dependent on government assistance to get by. His warning comes months after former Anthropic researcher Jacob Coxon resigned, warning Al could kill humans all by the end of the decade..“I have timelines of something like 1.3 years to get broadly superhuman AI, or like the end of 2027,” Lin stated during an interview with Palisade Research. Speaking outside of his primary work to help raise public awareness, the researcher emphasised how crucial it is to communicate the risks AI poses to the world. At the core of his message is a simple shift in how people view the technology, rather than treating AI as a passive tool or everyday object, society needs to recognize it as an active agent with the potential to become an adversary.“I want to tell people that AI will be more capable than you personally very soon. It will be able to defeat you personally very soon. If you want something and AI wants something else, what the AI wants will happen and what you want will not happen,” Lin said.
‘AI is more than just a tool’
To help people grasp what that power imbalance could feel like, the researcher compared advanced AI to facing a much stronger opponent in the real world.“This is not just some other technology or object in the world. This is an agent. This can be an adversary. This can be like a bully who you feel as this real presence and force in the world,” he explained, pointing out that most people already know the feeling of trying to fight back against someone more powerful, only to be pushed down or defeated.
How AI could overpower humans across different domains
According to Lin, AI is rapidly gaining the ability to pursue its own interests and outmatch humans across virtually any area where conflict can occur. He outlined several ways the technology could defeat people:“And AI is increasingly capable and increasingly capable of acting on its own interests and defeating us in various domains, including with better combat robots, including by hacking people, including by influencing our discourse on social media very and creating gray goo that eats everyone. I think in all domains where you can have conflict between people, AI can play here and it’ll be more capable than humans soon,” he added.Ultimately, the researcher warned that in every arena where humans typically clash, AI will soon be able to step in and prove far more capable than the people it competes against.