Current and former researchers from OpenAI, Google DeepMind, and Anthropic have released a series of interviews warning that superintelligent AI systems could pose an existential threat to humanity.
Key Points
- Palisade Research launched the "frominside.ai" platform featuring interviews with industry experts regarding AI safety and existential risks.
- Former OpenAI researcher Daniel Kokotajlo stated that superintelligent AI would be god-like and currently lacks any effective control mechanisms.
- Google DeepMind scientist Neel Nanda estimated a 10 percent probability that advanced AI could lead to human extinction.
- Geoffrey Irving, formerly of OpenAI and Google DeepMind, suggested the likelihood of human extinction from AI development is approximately 50 percent.
- Contributors acknowledged the conflict of interest in their warnings, noting they are employed by the very labs developing these powerful technologies.