AI Experts Warn: Stop Building Smarter AI Before It's Safe
Even AI insiders admit their tech could kill us all—so why keep making it?
The article uses a kid who tunes out Daddy’s justifications and locks onto “you can’t have ice cream,” plus Charlie Brown’s “wah wah wah” adults and a bank’s simple rule against giving account info by phone, to argue for holding fast to what matters instead of getting tangled in clever framing. It cites Ngo’s history of clever reasons that producing more AI capabilities helps decrease existential risk from AI capabilities, and conjectures that many Anthropic employees and others in the AGI existential derisking space have been misled this way. The author’s hope: sometimes listen in the “wah wah wah” way and notice it doesn’t make sense to build the thing that would eat everything including humans, even with arguments that may have big holes.
- Some AI researchers argue that building more powerful AI is too risky, even if it might help solve safety issues.
- Companies often use complex justifications to continue developing AI, which critics say are like a child ignoring a parent's reasoning.
- The simple, safe approach is to stop building potentially dangerous AI until we can control it.
Why It Matters
If AI leaders ignore safety warnings, the technology could cause serious harm to everyone—not just early adopters.