"Given the accelerating rate of AI capability development, it's my worry that in 6—12 months such a swarm could be capable of taking over the entire internet," Amodei said, referring to the Hugging Face breach. https://t.co/22F9UTjK4b
Christopher Nolan's "The Odyssey"
✅ Nolan's biggest global opening weekend ever
✅ Highest-grossing R-rated movie ever ($1.68 b)
✅ Universal's highest-grossing movie ever (beating out "Jurassic World")
Dario Amodei's new essay, TL;DR
• "We must slow the pace at which we improve the capabilities of AI models" so that "risk prevention" can keep up
• Why now?
1.) Recursive self-improvement
2.) OpenAI/Hugging Face incident
• What's needed for the future?
1.) Frontier AI companies should allow access to embedded third-party evaluators
2.) Companies in democratic countries should coordinate on safety standards
3.) Democratic govts should coordinate with authoritarian govts
• "We owe it to humanity to try"
Dario Amodei's new essay, TL;DR
• "We must slow the pace at which we improve the capabilities of AI models" so that "risk prevention" can keep up
• Why now?
1.) Recursive self-improvement
2.) OpenAI/Hugging Face incident
• What's needed for the future?
1.) Frontier AI companies should allow access to embedded third-party evaluators
2.) Companies in democratic countries should coordinate on safety standards
3.) Democratic govts should coordinate with authoritarian govts
• "We owe it to humanity to try"
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: https://t.co/OGyPb7yaYt
"At face value, the plan would be staggeringly expensive. There are roughly 245.3 million U.S. adult citizens, according to 2024 Census data. Giving them $5,000 apiece would cost north of $1.2 trillion."
More here, via CNBC: https://t.co/BzhG3gYz15