Matt DiGiuseppe
mdigiuseppe.bsky.social
Matt DiGiuseppe
@mdigiuseppe.bsky.social
Political Scientist @leiden.
PI of ERC MIDEBT project studying opinions on public debt (midebt.org), AI, and security.
Assc. Editor ISR.
Co-founder REALdegrees.ai
Reposted by Matt DiGiuseppe
A short summary of my argument that using ChatGPT isn't bad for the environment - Andy Masley andymasley.substack.com/p/a-short-summ… #AI #environment #energy
November 13, 2025 at 6:06 AM
Immigration is popular considering the alternatives - doi.org/10.1080/1350...
November 12, 2025 at 3:34 PM
Reposted by Matt DiGiuseppe
Your 'moment of doom' for Oct. 11, 2025 ~ Everything is fine.

"The invisible gas can be seen in streams of bubbles originating on the seafloor of Antarctica's Ross Sea ... describing the mechanism as 'seemingly widespread' throughout the region... "

abcnews.go.com/Internationa...
Researchers find methane leaking out of cracks in Antarctic seabed
One of the most potent greenhouse gas emissions has been discovered seeping out of cracks of the Antarctic sea floor, researchers have found.
abcnews.go.com
October 11, 2025 at 1:26 PM
Reposted by Matt DiGiuseppe
Anthropic AI tries to cheer us up a bit www.anthropic.com/research/sma...
A small number of samples can poison LLMs of any size
Anthropic research on data-poisoning attacks in large language models
www.anthropic.com
October 9, 2025 at 8:02 PM
Reposted by Matt DiGiuseppe
Criticizing Charlie Kirk is a fireable offense that incites domestic terrorism. But calling political opponents "the party of hate, evil, and Satan" is proper and good. Got it.
October 3, 2025 at 3:19 PM
Very grateful for the invite to this conference. I learned a lot.
Last Friday (Sept. 19), @andduer.bsky.social, @mcortinaescudero.bsky.social, @yavuzmehmet.bsky.social and I had the great pleasure of hosting an excellent workshop at @unisalzburg.bsky.social on how to use Large Language Models #LLMs to boost International Relations research. A short 🧵1/4
September 23, 2025 at 4:26 PM
Yesterday I posted about how I used comet browser to take a Qualtrics survey almost undetected. Last night, I ran a pilot (N=400) on @joinprolific.bsky.social . I found that almost 10% of "respondents" identified as AI when directly asked.
I added a question in the survey to identify itself as an AI if instructed to be a human. That seemed to work. Though, I imagine I could jailbreak it eventually.
September 23, 2025 at 8:38 AM
🚨🚨🚨 Survey Researchers! WE HAVE A HUGE PROBLEM!🚨🚨🚨

I have access to the Comet browser by Perplexity.

I asked it to pretend to be a human, take the survey, and give me feedback. IT DID!
September 22, 2025 at 9:51 AM
Reposted by Matt DiGiuseppe
What's the carbon footprint of using ChatGPT or Gemini? [August 2025 update] - Hannah Ritchie https://www.sustainabilitybynumbers.com/p/ai-footprint-august-2025 #AI #energy
August 22, 2025 at 3:08 AM
Reposted by Matt DiGiuseppe
Earlier this year, over 230 Venezuelan men were sent to a maximum-security prison in El Salvador by the Trump administration.

Our @propublica.org team spent months looking into who they were to find the truth behind their stories. projects.propublica.org/venezuelan-i...
August 11, 2025 at 7:19 PM
Reposted by Matt DiGiuseppe
I'm hiring another postdoc (research-focused, for almost 5 years) for my @erc.europa.eu project on the educational cleavage!
I'm looking for someone with strong quantitative text-analysis skills (e.g. #NLP, #LLM, etc.) to study the role of political actors in cleavage formation.
July 25, 2025 at 12:37 PM
Reposted by Matt DiGiuseppe
Do sanction threats work? We find that threats—even if credibly backed—do not increase support for concessions in the target public (Turkiye case). Plus, positive inducements provoke backlash in some groups, highlighting coercive diplomacy’s limits

Check out our research note: shrturl.app/3l4wWJ
July 21, 2025 at 11:45 AM
Reposted by Matt DiGiuseppe
I think we really ought to consider the dramatic rise in popularity of podcasts and video essays as conclusive disproof of the 'everyone hates lectures' idea in pedagogy. Turns out that, even when faced with many alternatives, many people just like to be told something at length.
July 12, 2025 at 2:14 PM
Reposted by Matt DiGiuseppe
UB's new Department of AI and Society is hiring faculty across ranks (Assistant, Associate, Full Professor). We’re looking for transdisciplinary scholars interested in building AI by society, for society. Start dates begin Fall 2025.

More info: www.ubjobs.buffalo.edu/postings/57734
Assistant, Associate or Full Professor, AI & Society
The Department of AI and Society (AIS) at the University at Buffalo (UB) invites candidates to apply for multiple positions as Assistant Professor, Associate Professor, or Full Professor. The new AIS ...
www.ubjobs.buffalo.edu
July 17, 2025 at 4:11 PM
Reposted by Matt DiGiuseppe
oh this is interesting. i pretty firmly believe, just as a matter of observation and experience, that complaints about affirmative action are a psychological salve for people who aren’t confident of their ability to succeed
July 13, 2025 at 2:49 PM
Reposted by Matt DiGiuseppe
Shreyas Meher and Patrick T. Brandt demonstrate how state-of-the-art language models can be efficiently adapted for political event classification, achieving up to 1400% improvement in accuracy while remaining computationally accessible.

Read more here: journals.sagepub.com/doi/full/10....
ConflLlama: Domain-specific adaptation of large language models for conflict event classification - Shreyas Meher, Patrick T. Brandt, 2025
We present ConflLlama, demonstrating how efficient fine-tuning of large language models can advance automated classification tasks in political science research...
journals.sagepub.com
July 9, 2025 at 7:47 PM
🤖 New Pre-Print: Ad Machina: Partisanship and Support for Delegating Government Decisions to Autonomous Algorithms

osf.io/preprints/so...
OSF
osf.io
July 4, 2025 at 9:01 AM
Reposted by Matt DiGiuseppe
🚨 @citap.bsky.social is hiring 🚨 tenure-track assistant prof w/ faculty appt @uncsils.bksy.social - looking for a scholar whose research examines the complex interconnections between AI technologies & social, organizational, and/or political systems. unc.peopleadmin.com/postings/303...
Assistant Professor
The School of Information and Library Science (SILS) and Center for Information, Technology, and Public Life (CITAP) at the University of North Carolina at Chapel Hill invite applications for a tenure...
unc.peopleadmin.com
July 3, 2025 at 3:17 PM
Reposted by Matt DiGiuseppe
Indiana Jones and Are You Fucking Kidding Me Why Do I Have to Put Learning Outcomes on My Syllabus The Outcome I Want is For the Students to Learn What's Assigned on the Parts of the Syllabus They Can Barely Find For All the Dumb Boilerplate Nonsense I'm Supposed to Put on the Syllabus
July 2, 2025 at 6:44 PM
Reposted by Matt DiGiuseppe
Only a few days left to register to the Amsterdam Replication Games on July 19. Virtual participation is possible and coauthorship to a meta paper is granted.

Register: www.surveymonkey.ca/r/Replicatio...
July 1, 2025 at 11:21 AM
Reposted by Matt DiGiuseppe
Did you know that @epssnet.bsky.social will honor your #EPSA membership? Sign up today!

epssnet.org/sign-up/#join

That said, your membership dues are staying with EPSA & EPSS is starting with zero financial funds. So if you can, consider paying for an EPSS membership to help us get off the ground.
June 28, 2025 at 1:25 PM
This THE way to use LLMs for scaling text data.
@mdigiuseppe.bsky.social uploaded an updated copy of our paper "Scaling Open-ended Survey Responses Using LLM-Paired Comparisons" to SocArXiv and it's available for viewing/download. doi.org/10.31235/osf...
OSF
doi.org
June 27, 2025 at 5:49 PM
Reposted by Matt DiGiuseppe
Europe needs to understand that the presence of AC is way more important for climate adaptation than the absence of AC is for preventing climate change.
June 22, 2025 at 5:24 PM
Reposted by Matt DiGiuseppe
CRS has a really good overview of US uses of force and the authorities presidents have cited from 2001 through 2017. They all tend to lean on a mixture, but it’s not unusual to cite inherent constitutional powers alone or alongside the AUMFs.
June 22, 2025 at 4:54 AM