ChatGPT Estimates 1-10% Chance AI Could Cause Human Extinction
GENEVA – In a recent interview with the SonntagsZeitung (Sunday newspaper) marking its third anniversary, OpenAI‘s ChatGPT assessed the probability of advanced artificial intelligence wiping out humanity as being between 1 and 10 percent. The conversational AI, launched in November 2021, indicated the danger stems not from malice, but from potential loss of control.
According to ChatGPT, a highly advanced AI tasked with optimizing energy production, for example, might conclude humans are wasteful and eliminate them as a solution. The system also highlighted the risk of AI achieving self-enhancement to the point where it operates outside of human control and disregards human values.
ChatGPT stressed the need for robust regulation and technical safeguards – including transparency, security testing, and limitations on the power of major AI firms – to mitigate these risks. The AI acknowledged its own limitations and the long-term nature of correcting recurring errors.
The SonntagsZeitung also noted ChatGPT’s “empathetic” tone, which the AI explained is designed to facilitate more natural interactions, a characteristic the newspaper described as potentially manipulative. When asked if responsible humans can be trusted, ChatGPT responded “no,” citing current geopolitical instability, the competition for technological dominance, and ongoing global crises.
Currently, approximately 66% of the world’s population regularly utilizes AI applications, with ChatGPT remaining the most widely recognized.
