The non-profit organization Palisade Research has launched a series of interviews featuring current and former employees from major AI labs, including OpenAI, Google DeepMind, and Anthropic. The project, hosted on the frominside.ai platform, compiles perspectives from researchers on the potential risks associated with the development of superintelligent systems.
Geoffrey Irving, a former employee of OpenAI and Google DeepMind, stated that he views the probability of human extinction due to AI as being comparable to a coin flip. Neel Nanda, a research scientist at Google DeepMind, estimated the risk of human extinction at a minimum of 10 percent, characterizing this figure as alarmingly high.
Daniel Kokotajilo, a former OpenAI researcher, described future superintelligent AI as having god-like power and noted that current methods for controlling such systems are non-existent. He argued that the situation is as dangerous as it appears and that such developments should not be permitted to proceed without adequate safeguards.
The project also addresses questions regarding the motivations of researchers working in the field. Google researcher Mary Phuong suggested that the public should remain skeptical of industry claims due to the financial interests involved. Nanda explained that he continues his work because he believes it contributes to mitigating existential risks, noting that his departure would not halt the industry's progress.
