The scientific facts: LLMs are a black box. AI is emerging as a new species.
Learn more:
- AI: Unexplainable, Unpredictable, Uncontrollable by Roman V. Yampolskiy
- Science: Managing extreme AI risks amid rapid progress by Bengio et al.
- Science: Regulating advanced artificial agents by Russell et al.
William Saunders explains why he resigned from OpenAI:https://t.co/QHYsEZXfUV
— ControlAI (@ai_ctrl) July 10, 2024
Anthropic CEO Dario Amodei on AI existential risks:
— In a couple years, as AI models improve, people could do very bad things with them.
— It could be difficult to control AI agents, which might pose an existential risk. pic.twitter.com/gQdGt6Tg1x— ControlAI (@ai_ctrl) July 17, 2024
Major General (Ret.) Robert Latiff:
— We’re not being aggressive enough at controlling AI
— These are “really dangerous”
— A big concern is the implementation of AI in military command and control systems pic.twitter.com/P6KHwF9jZs— ControlAI (@ai_ctrl) July 19, 2024