Understand

Tech and AI: 10. Can We Control AI?

10 snips
Jul 7, 2023
Ask episode
AI Snips
Chapters
Transcript
Episode notes
INSIGHT

AI Amplifies Convincing Misinformation

  • AI's most plausible near-term harms are misinformation and convincingly written fake content, not an immediate robot takeover.
  • Dr Michael Pound cites smarter phishing emails and auto-generated text that people can't reliably fact-check as a current major risk.
INSIGHT

Pausing Development Is Practically Fragile

  • Regulatory pauses sound appealing but risk being ineffective because some actors will continue development secretly.
  • Dr Michael Pound warns that competition and national ambitions make universal pauses unlikely to succeed.
INSIGHT

Government Competition Fuels AI Progress

  • There is an AI arms race element as governments seek superiority for economic and strategic advantage.
  • Dr Michael Pound notes many governments want better AI for varied uses, implying competitive pressure to accelerate progress.
Get the Snipd Podcast app to discover more snips from this episode
Get the app