Zach Freitas-Groff

Link: ai-2027.com

Most compelling rogue-AI story I’ve encountered.

Goes faster at various points than I can readily believe, but perhaps that's to be expected. A useful illustration of some of the many things that could happen.

Comments welcome! Paper: www.zachfreitasgroff.com/CooperatingP...

This suggests that, at least in our experimental environment, generational reciprocity alone could boost future-oriented investments, without complex monitoring or detailed historical transparency.

Thus, even the expectation that future generations might reciprocate—without direct observation of your own generosity—motivates investing forward.

Remarkably, observing past actions doesn't matter. Players contribute more to future generations simply when the next generation can potentially give back—even if past generosity is hidden.

To examine this, we randomly vary the info sets:

Some see nothing about previous players.

Some see past investments forward.

Some see how much was passed back and forward.

There are other treatments described in the paper and appendix.

We vary treatments by changing the ability of future players to give back and the information available about what others did. Theory predicts a clear link: knowing past generosity should incentivize giving forward.

Each player represents a "generation" and decides how much of their resources to pass forward to the next player and—crucially—how much to pass backward to the previous one.

📢 Working paper w/ @benleoecon.bsky.social, Johannes Lohse, and Oliver Hauser!

Some have proposed that you can get current generations to invest in the future by enabling future generations to give back.

We test this in an experiment with sequential "generations" of players.

🧵 #EconTwitter

Co-authored with the wonderful @zdgroff.bsky.social, Oliver Hauser, and Johannes Lohse.

Talking to people in the Bay Area last week, I heard of more partial patches for AI alignment than I'd encountered before, such that I'd feel pretty optimistic if many of them were implemented. How much is the issue now *making sure the things we know are good actually happen*?

Dean was a dear mentor to me as I approached economics PhD programs—deeply committed to evidence and efficiency in helping others.

What a shame to no longer have him applying that lens to USAID.

@evavivalt.bsky.social posted -> i texted it to the gc -> gc emails to tyler -> marginal revolution post!

Here’s the full paper: evavivalt.com/wp-content/u...

In short, cultured meat may have large direct effects on animals, but in our experiment, it does not prompt a social turn against animal agriculture.

The backlash effect even extends to a stronger preference for meat relative to plant-based meat when people are exposed to cultured meat.

These attitudes depend a lot on the social environment-exposure to strongly negative attitudes toward cultured meat can enhance this effect (while some positive messages we study do not succeed in counteracting this).

Why? It looks like people react to a fear of cultured meat—perhaps thinking it will replace animal-based meat—by expressing protest views in a backlash or, in psych, "reactance" effect.

Working paper out!

Does cultured meat make people more open to animal welfare concerns with factory farming? Models of motivated beliefs (or cognitive dissonance) would generally predict yes (substitute = less attachment to meat). We find no!

@zdgroff.bsky.social, Bobbie Macdonald and I consider the impacts of a different new technology: "cultured" meat, i.e., meat that is grown in a lab from cells. 3/

After 11 months of work, we proudly announce Third Opinion: A free of charge expert consultation service for frontier AI professionals. To help you clarify if what you're seeing is cause for concern. Anonymous, without sharing confidential information 🧵
Launch announcement: tinyurl.com/2-9unc19193

See the Twitter thread here for a social-media-style summary: x.com/zdgroff/stat...

📈Job market paper📉

Excited to finally share my JMP! I study whether and why policy choices are stubbornly persistent.

Example: Oregon has an income tax, but not Washington—seemingly because of nearly century-old choices. Is this typical?

Read the paper: www.zachfreitasgroff.com/FreitasGroff...

Saw this on the list for the terrific Economics of LGBTQ+ Individuals Virtual Seminar Series:

www.aeaweb.org/about-aea/co...

Neat. Evidence that legal same-sex marriage decreased enrollment in priestly studies (because fewer gay men avoid coming out):

digitalcommons.chapman.edu/cgi/viewcont...

Bonkers. (See updated chart in comment thread.)

AGI Catastrophe and Takeover: Some Reference Class-Based Priors — EA Forum This is a linkpost for https://docs.google.com/document/d/11YKTKRumtlheK_9Dv9ECKwwoTeSG3RNcs6qUSajzqDw/edit?usp=sharing …

Here is my post for those interested: forum.effectivealtruism.org/posts/MDNcML...

Excited to share that I won a second prize in Open Philanthropy's 2023 AI Worldviews contest!

www.openphilanthropy.org/research/ann...

Anyone have an NBER invite code?