Jaron Lanier warns chatbot fluff inflates US AI costs

Western artificial intelligence labs are burning billions on digital parlor tricks.

On September 24, 2026, host Adam Curry highlighted computer scientist Jaron Lanier’s critique of American model design on No Agenda. Lanier argued at Steve Bannon’s Pro-Human Assembly that Western tech companies stack massive compute overhead onto models to maintain the illusion of an empathetic, living agent. Voice clones, chatty rapport, and persistent memory tricks exist to convince users they are speaking with a deity rather than software.

That anthropomorphic theater carries a staggering financial bill. Lanier noted that Chinese AI developers ship lean, utility-focused tools that achieve comparable performance at a fraction of the operating expense. Because the Chinese Communist Party enforces an explicit ideological ceiling, Chinese engineering skips personality tests and focuses on raw execution.

The financial vulnerability is already showing up in real-world software usage. Curry and guest Dana Brunetti tested that dynamic against consumer behavior on the show, pointing out that utility-focused tools win market share when they bypass conversational fluff.

Users want instant results - whether pricing consignment dresses or scraping travel deals - not a talkative companion. Western frontier models burn capital on brand positioning and synthetic empathy while basic task execution gets buried under compute bloat.

To shield those massive operating budgets, Western AI labs routinely demand regulatory pauses and safety carve-outs under the guise of existential risk.

Lanier’s analysis strips away that safety narrative, framing anthropomorphic design as a self-inflicted cost disadvantage. Unadorned software from abroad can undercut American models while delivering identical work.

The compute bill is paying for a personality test.