TapTechNews May 22nd news, reported by Reuters, 16 technology companies including Microsoft, Google, OpenAI, etc. promised to safely develop AI models at the Artificial Intelligence (AI) Seoul Summit held on the 21st and 22nd, shutting down their cutting-edge systems in case the most extreme risks cannot be controlled.
The participating companies also include domestic enterprise Zhipu AI (TapTechNews note: Tencent, Ali, Meituan, and Xiaomi have invested in it), IBM, Amazon, Meta, and Samsung.
The safety commitment includes publishing a framework to clarify how the company will measure the risks of its models. In extremely serious and unacceptable extreme cases, if the risks cannot be mitigated, the artificial intelligence companies will have to hit the emergency stop switch and stop developing or deploying their models and systems.
This is not the first time that artificial intelligence companies have made lofty-sounding but non-binding safety commitments. Amazon, Google, Meta, and Microsoft signed voluntary safeguards brokered by the US government last year to ensure that their products are safe when released.
The computer scientist Joshua Bengio, known as the godfather of artificial intelligence, welcomes these commitments, but he points out that voluntary commitments must be accompanied by supervision.