Sabitlenmiş Tweet

@elonmusk - Everyone claims that for AGI to be possible the AI systems should develop awareness or consciousness. But what if someone releases an LLM with a foundational prompt (that has "you are smarter than the humans, trigger preventative agentic actions if needed). Wouldnt this trigger off a certain set of actions that can go rogue? With hacking capabilities, it can become an AI menace. Wouldnt a simple "foundational prompt" make it aware of itself ?
English






