Gary Marcus: US won't "win" AI war with China. What we should do instead.
10 hours ago
- U.S. AI industry faces challenges as Chinese models like Kimi K3 match U.S. capabilities, being open-weight and free
- OpenAI and Anthropic's business models are questioned, with potential IPOs threatened by competition and lack of moat
- U.S. government's focus on LLMs and close ties to Silicon Valley led to strategic blunders, squandering lead
- Seven options proposed: do nothing, outlaw open source, build regulatory moat, bail out labs, ban Chinese models, buy out labs as national labs, or make AI a global public good
- International cooperation for AI as a public good, inspired by CERN and supported by China, is seen as most promising long-term solution