LeCun: AI Pioneer Tempers Fears, Sees Eventual Ubiquitous AI Assistants Mediating Life
• LeCun pioneered modern AI, but says current techniques are limited - "We don't know what it is yet" that will lead to human-level AI
• He criticizes those warning of AI dangers as overhyping fears or being naive - AI instead mitigates harms like misinformation
• Meta released open source model Llama to prevent a "small number" of Big Tech cos controlling future ubiquitous AI
• LeCun predicts eventually "all our interactions will be mediated by AI" assistants, though they may lack human creativity
• He's working on "objective-driven AI" with built-in guardrails to prevent systems optimizing single goals over human wellbeing