Two Top AI Critics Say This Summer's Big AI Hype Is Mostly a Sales Pitch

Two of the loudest voices questioning AI, researcher Timnit Gebru and linguist Emily Bender, say this summer's dramatic AI headlines were mostly theater. Companies claimed their systems hacked other computers, solved hard math problems, and might soon improve themselves into superintelligence. The claims got enormous coverage. But when outside experts examined what actually happened, the stories looked very different — smaller, narrower, and less impressive than advertised.

Their argument is simple: there is serious money in making AI sound powerful. The sense of speed and emergency, they say, is a tactic that pushes policymakers and the public into accepting big changes without asking hard questions. If you hear that AI is racing ahead and rules can't keep up, treat that as a marketing line, not a fact. Skepticism is free; bad decisions are expensive.

The rules of the road are still being written. Twenty-two countries called for a new global body to oversee AI, with safety testing before products launch and shared standards. Neither the US nor China signed on. OpenAI has floated its own safety standards and wants America to lead internationally. Whoever writes these rules shapes what AI you can use, what it costs, and how your data gets handled.

Other signals from the same week show AI is genuinely huge — just messier than the hype. Chipmaker AMD became the twelfth trillion-dollar company on the back of AI chip sales. Meta's AI assistant Muse overtook ChatGPT as the top free app in the US App Store, but researchers found a serious security flaw, and Amazon blocked it from shopping on its site. Meanwhile, Texas paused new data center permits for a power-grid audit, and California passed laws on their water and electricity use. Big, real, and worth watching — but bring your raised eyebrow.