Post by @rowancheung

The AI Futures Project, the same forecasters behind AI 2027, just put out a new scenario report called AI 2040. The question they're trying to answer is how humanity gets to superintelligence without ending in extinction or a permanent power grab. Their recommended path is Plan A. The US and China pause frontier AI training in 2029, then scale slowly together under total research transparency, pushing superintelligence out to 2040. The enforcement mechanism is the strange part. Each country builds its datacenters on rival-adjacent soil, America in Mongolia and China in Canada, so the chips can be destroyed quickly if the deal falls apart. They call it mutually assured compute destruction. Worth noting it's a recommendation, not a prediction. The team puts the odds of AI-driven extinction somewhere between 10 and 30 percent. Not exactly reassuring… but it’s a fascinating read and worth exploring more at 👉 ai-2040.com/