The Warning of AI Researchers: Superintelligence is Just as Dangerous as it Sounds

In a recent interview, former OpenAI and Google DeepMind employee Geoffrey Irving expressed his concern about the possibility of human extinction, stating that in his view, the chance of it happening is like a coin flip. This alarming sentiment is shared by several other AI researchers, both current and former employees of top tech companies such as OpenAI, Google, and Anthropic.

Neel Nanda, a research scientist at Google DeepMind, echoed Irving’s concerns by stating that there is at least a 10 percent chance that AI could lead to human extinction, a risk that he deems unacceptably high. The complexity of defining and addressing the concept of “AI safety” is a central theme in these interviews, with no clear solutions being presented.

Former OpenAI researcher Daniel Kokotajlo further emphasized the potential dangers of superintelligent AI, likening it to a god-like entity with immense power. He cautioned that the lack of control over such advanced AI systems poses a significant threat that must be prevented at all costs.

While the full interviews are available, the focus remains on the collective insights shared by these experts on the risks associated with AI development. Discussions range from questioning the hype around AI to the ethical implications of working on potentially dangerous technologies. Google’s Mary Phuong highlighted the importance of skepticism, considering her affiliation with the tech industry, while Nanda stressed the critical role of researchers in mitigating existential risks posed by AI.