A.I. Giants Unite for Safety Standards as Language Models Falter

A standards body is being developed by OpenAI, Anthropic, and Google to address safety concerns in artificial intelligence. This comes after Dario Amodei, Sam Altman, Elon Musk, and Demis Hassabis called for a slowdown in the development of frontier AI.

The language model's ability to translate different languages has been found to be lacking, with OpenAI's models being particularly bad at translating Hindi, Thai, and Estonian. According to research by Smartling, OpenAI and Google have closed the performance gap between their older and newer models for non-English languages.

Mustafa Suleyman, Microsoft AI CEO, has written an essay opposing the approach taken by Anthropic, which he claims is training its AI system Claude that it may be conscious and deserving of rights. He warns that this could have a disastrous impact on humanity.