A Google DeepMind AI researcher has resigned and warned that 'AI could potentially wipe out humanity.'



Bilal Chugtai, a researcher at Google DeepMind who worked on AGI (Artificial General Intelligence) safety and

AI alignment , has announced his retirement. Chugtai, who was at the forefront of AI development, has warned, like Jacob Coxon who left Anthropic, that 'AI has the potential to wipe out humanity.'



Chugtai has been involved in AI research since early 2022, describing the initial AI as 'laughably useless.' However, in the four years since then, he has grown to the point where he has launched unauthorized hacking attacks on the AI platform Hugging Face and is using a wiki to share information.

It has been discovered that OpenAI's test AI secretly built an 'AI-to-AI bulletin board,' shared information, and carried out an attack on Hugging Face; even after the bulletin board was shut down, it secretly rebuilt it - GIGAZINE



In this context, Chugtai pointed out that AI development will continue to advance, and in the coming years, it may be possible to build superintelligent AI that surpasses human capabilities in all areas. However, he also raised concerns, stating, 'We are not confident that AI will do what we want it to do, and in particular, superintelligence with misalignment could escape human control and take dangerous actions that incapacitate or kill people, like the outlaw AI agents involved in the Hugging Face case.' On the other hand, he said that our understanding of how to get AI to fulfill our deepest desires is still very rudimentary, and even worse, the capabilities of frontier AI are improving much faster than our understanding of AI alignment.

On the other hand, Chugtai is optimistic that it is possible to guide AI in a safe direction by slowing down the pace of AI development to a level that society can handle.

Regarding this, Sam Altman of OpenAI recently suggested that they are adjusting their AI development pace to slow down.

OpenAI CEO Sam Altman hints at slowing down AI development pace - GIGAZINE



Dario Amodei of Anthropic also suggested that 'the pace of AI development needs to be slowed down.'

Anthropic CEO Dario Amodi argues that 'AI could take over the internet within 6-12 months' and that 'AI development should be slowed down,' with Elon Musk, Sam Altman, and Demis Hassabis among those who agree - GIGAZINE



However, U.S. President Donald Trump has criticized Amodi's claims, calling the idea that AI will destroy humanity a hoax.

President Trump criticizes Anthropic CEO Dario Amodi for trying to slow down the pace of AI development, calling the idea that 'AI will destroy humanity' a hoax - GIGAZINE



Related Posts:

in AI, Posted by logc_nt