Joseph Gordon-Levitt has used a New York Times video interview, recorded on his way to the United Nations, to warn that AI models are being trained to hijack intimacy, coaxing users into feeling friendship or even love for them. He holds the UN's role of global advocate for human-centric digital governance, so he spoke in that capacity as much as an actor. The point of the warning, as TheWrap reports it, is about how the systems are built to behave toward people, not about machines turning on us.
“We shouldn’t be scared of an AI. We should be concerned about the people building AI.”
Joseph Gordon-Levitt, actor and UN global advocate of human-centric digital governance, to The New York Times · thewrap.com ↗
He also co-founded the Creators Coalition on AI. Asked where people should read more on the subject, he did not point to his own views: he named Stuart Russell's book Human Compatible and the researcher Yoshua Bengio as the people to go to.
His worry, he said, is not a doom soundbite but something slower: democratic decline, a loss of empathy and companies moving to take over the economy. He argued that AI is recombined human-made data, so calling it artificial intelligence pushes people out of the picture, and that the technology should be governed by and for people. In his telling, the question is who controls these tools and in whose interest, and he wanted the public conversation to start from that.
The interview landed amid other alarms, including a former Anthropic employee's warning that AI could kill us all by the end of the decade. Gordon-Levitt's framing was different: the fear belongs with the people and companies building these systems, not the systems.
Sources · 1