Many have asked for the LLM Survey paper. The release was bungled a bit by PNAS, but it is live now: www.pnas.org/doi/10.1073/...

AI presents a fundamental threat to our ability to use polls to assess public opinion. Bad actors who are able to infiltrate panels can flip close election polls for less than the cost of a Starbucks coffee. Models will also infer and confirm hypotheses in experiments. Current quality checks fail

Careful polling shows that Americans overwhelmingly reject political violence - see e.g. brightlinewatch.org/tempered-exp..., brightlinewatch.org/accelerated-..., and www.pnas.org/doi/abs/10.1....

Unfortunately, it only takes one person with a gun to create a tragedy like what've seen today.

As scary as this moment is, remember that support for political violence went DOWN after the July 2024 assassination attempt against Trump, not up www.pnas.org/doi/abs/10.1...

We need to make the same thing happen this time.

I am looking for a post-bac (Dartmouth language for a pre-doc) researcher to join our team studying elections and democracy. If you (or someone you know) have expertise in data science or political science, please apply or share this opportunity!

polarizationresearchlab.org/hiring/

Also, to be clear: I have no interest in a specific estimate of support for partisan violence. I just think that we have to use high-quality measures and deploy best practices to ensure we are informing policy debate with credible evidence.

Assessing support for political violence based on social media vibes toward Luigi isn't all that informative. First, it conditions on the narrow silo of views someone is exposed to. Second, there is a lot of expressive responding to Luigi. We should update based on evidence not feeling.

Hot take: we should more worried about society's refusal to condemn political violence against corporate targets than the violence itself (very, very rare).

How did American attitudes toward democracy shift with the election? Our final Path to 2024 report shows substantive changes in perceptions of accuracy, trust, and resignation toward democratic backsliding.

Read the full report: prlpublic.s3.us-east-1.amazonaws.com/reports/Elec...

Job alert! Interested in working with PRL as a postdoc next year? Apply by January 15! We offer the opportunity to collaborate on and lead projects related to democratic attitudes, elite behavior, and more! polarizationresearchlab.org/hiring/

Funded conference! Join PRL (Yphtach Lelkes, Sean Westwood, and Shanto Iyengar) in SLC on May 1-2 to discuss new work on democratic attitudes. Grad students/postdocs welcome. This is a small, funded meeting with great papers and discussion. Apply by 12/31
polarizationresearchlab.org/annual-meeti...

My first post on this platform is sharing this great op-ed from Dartmouth's President.

www.wsj.com/opinion/dart...