From Futurism: 2024-06-03 10:21:35

OpenAI CEO Sam Altman admits struggling to understand how the company’s AI models function, emphasizing lack of interpretability. A panel of experts highlighted limited knowledge among AI developers. Competitor Anthropic invests in mapping artificial neurons for better understanding. AI safety remains crucial amidst debates on rogue AGI risks. Altman dissolves “Superalignment” team, shifts focus to safety and security. AI interpretability key to making and verifying safety claims.



Read more at Futurism: Sam Altman Admits That OpenAI Doesn’t Actually Understand How Its AI Works