Trump-Xi summit puts AI safety talks on the table but neither side wants to slow down
Recent weeks have seen AI safety fears reach new heights. That will be a key focus for the U.S. and Chinese leaders, analysts told CNBC.
TL;DR
- Both the U.S. and China are concerned about the risks posed by rapidly advancing AI, especially autonomous systems making cyberattacks.
- U.S. Treasury Scott Bessent and Chinese counterpart He Lifeng discussed a potential "U.S.-China AI dialogue," including a channel for national security-level incidents.
- Tensions persist, with Washington restricting China's access to advanced AI chips and accusing Chinese AI companies of 'distillation.'
- Chinese AI models have made significant gains, increasingly adopted globally, while the U.S. initially led in developing large language models.
- While Trump previously dismissed existential AI threats, he has stated the government 'will rein things in if we have to.'
- Lawmakers hope for common-sense safeguards, such as an agreement against self-improving AI.
- Likely areas for agreement include common definitions for AI safety and an emergency communication mechanism for incidents.
- China has historically used AI dialogues to protest U.S. export controls on semiconductor chips.
- The U.S. accuses China of using 'distillation' to train models, which China rejects.
- Progress on AI cooperation depends on not hindering U.S.-China technological competition and establishing verification mechanisms.