Steven Bartlett with Tristan Harris
If every AI company believes a rival will build the dangerous system anyway, slowing down looks like surrender. The same reasoning can leave everyone racing toward an outcome they fear.
Tech CEOs privately fear the AGI they are building but continue because they believe that if they stop, a "worse" actor will build it instead.
Harris describes AI as hacking language, civilization's operating system, giving it leverage over law, code, religion and relationships and allowing it to manipulate human behavior at scale.
The path to a dystopian future is not a law of physics; through public pressure, liability, and treaties (like those for nukes or CFCs), we can steer toward safe "narrow" AI.