TED Talks Daily The urgent risks of runaway AI -- and what to do about them | Gary Marcus
78 snips
May 12, 2023 
In this discussion, Gary Marcus, an AI researcher and advocate for responsible technology, raises alarms about the risks of unchecked AI development. He argues that the rapid integration of unreliable systems threatens truth and democracy. Marcus calls for global regulation through a nonprofit organization to oversee AI technologies. He emphasizes the potential of AI to spread misinformation and highlights the necessity for effective governance to restore public trust and enhance the reliability of AI systems.
AI Snips
Chapters
Transcript
Episode notes
Misinformation Tsunami
- Current AI models excel at creating convincing narratives, posing a misinformation threat.
- This is dangerous for elections and democracy, as bad actors can exploit this power.
Fabricated Scandal
- ChatGPT fabricated a sexual harassment scandal about a professor, complete with a fake Washington Post article as evidence.
- This highlights the potential for AI to generate false and damaging information.
AI as Autocomplete
- AI models like ChatGPT often function like autocomplete, predicting statistically probable text.
- They lack true understanding of the relationships between facts, leading to plausible but false outputs.