AI safety, Econ, new liberalism, math, and a bit of art history (as a treat)
Behavioral evaluations @TransluceAI. Prev Astra, MATS & Walmart's Econ Team
Problem: AIs can detect when they are being tested and fake good behavior.
Can we suppress the “I’m being tested” concept & make them act normally?
Yes! In a new paper, we show that subtracting this concept vector can elicit real-world behavior even when normal prompting fails.
e.g., I know I'm going to london but I might still be coordinating with others on specifics, pretty sure one flight would be good but fare lock just in case
Although sometimes I'm just anxious that I might fuck it up somehow and get the fare lock to reduce the friction of buying
My fifty page neel nanda mats application will fr be one in the history records
(I did not have time to format some super simple graphs and tables and just stacked them one after another. There's 16 pages of appendix that's just simple graphs and screenshots)
Wait I thought it's actually less than 0.5% globally but's more! 3% of the world earn more than 100k a year! That's 240 million rich people
pretty good
If you’re making 100k at 27, you’re:
> doing better than the median Ivy League graduate
> In the top 10% of income nationwide
> In the top 0.5% of income globally
When so many ask, “How is this economics?”, it’s nice when the QJE answers, “yes this is economics.”
Economics, like early hominids, thrives by aggressively expanding its territory.
My fifty page neel nanda mats application will fr be one in the history records
(I did not have time to format some super simple graphs and tables and just stacked them one after another. There's 16 pages of appendix that's just simple graphs and screenshots)
Nate Soares and I are publishing a traditional book: _If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All_. Coming in Sep 2025.
You should probably read it! Given that, we'd like you to preorder it! Nowish!
