MIT Professor Outlines Path to True AI, While Urging Caution on Bias and Legal Issues
• Current AI systems like ChatGPT, while impressive, lack true understanding of the world and can make harmful mistakes.
• MIT professor Melanie Mitchell believes the "big leap" in AI will come when systems actively experience the world like humans/children do, likely via robot form.
• Achieving human-level AI is possible but more complex than we realize; our intuitions about our own intelligence are often wrong.
• While speculative threats like AI getting out of control are concerning, more pressing issues like algorithmic bias and legal considerations need addressing now.
• Applications of AI to science problems, like protein folding and drug discovery, are an exciting development and potentially as important as when computers were first invented.