Asst Professor of Work and Organization Studies at MIT Sloan. PhD in Sociology from Stanford University. Culture, identity, and computational social science.
Joined February 2012
- Tweets3.5K
- Following218
- Followers467
- Likes642
Austin van Loon retweeted
I wrote a "playbook" for helping you prepare lists of potential tenure letter writers---researchers that know about the work AND are unconflicted by co-authorship, institutions, advising, &c.
It's designed to work w/ our @ExpectedParrot Research Agent. As it would be weird as heck to run it on anyone else, I ran it on myself to show you how it works 1/
Austin van Loon retweeted
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: darioamodei.com/post/we-must…
Austin van Loon retweeted
Americans say they want less inequality. Why don’t they support policies to reduce it?
The answer depends on which policies you ask about.
New paper: across 31 nationally representative survey waves (N=384,248) and a preregistered experiment (N=1,009), Americans favor PREdistributive over REdistributive policies🧵
osf.io/preprints/socarxiv/gk…
Austin van Loon retweeted
🚨🚨New paper in @Nature with @AshuAshok, @lukebeehewitt & @ghezae_isaias 🚨🚨
Can LLMs predict results of social science experiments?
Across 70 preregistered studies, we find strong correspondence (r=.85) between LLM-predicted and observed effects.
doi.org/10.1038/s41586-026-1… 🧵
Austin van Loon retweeted
🚨Published in @Nature: We show how AI can simulate human behavior to predict experimental results as accurately as experts
Our work has many implications for AI safety. For example, simulators can be used to aid harmful manipulation, and current safeguards do not prevent this🧵
Austin van Loon retweeted
LLM social science papers are often either boring or riddled with basic analytical problems. Per Engzell and I tried to come up with a workflow that would take a human-prompted idea, and actually run with it:
Austin van Loon retweeted
We asked those who said that their socioeconomic background helped them why that was.
Family norms & values and financial/social support came up most prominently (note that this differs from what FGLI respondents thought mattered).
2/3rds also emphasize soft skills
Austin van Loon retweeted
To first-gen or low-income background respondents who answered that their socioeconomic background disadvantaged them, we asked about mechanisms.
Again overwhelmingly, factors like: academic norms, the hidden curriculum, and limited access to networks or mentorship show up
Austin van Loon retweeted
I’m delighted to share that the August 2025 special issue of Sociological Methods & Research on Generative AI is out. With my co-editor, Daniel Karell, we put together this issue to build on the conference we organized last year.
Here's a thread on each of the 10 papers:
🚨JOB OPENING🚨
Come be my colleague! Work and Organization Studies at MIT Sloan seeks applications for an open rank, tenure-track faculty member. Research should focus on employment, work, or organizational issues (broadly defined).
Deadline Sept 5!
apply.interfolio.com/170634
Austin van Loon retweeted
Excited to continue learning about the latest #CSS at @IC2S2! I’ll be at the Social Prediction Session, presenting the mixed subjects design on combining human and LLM data in experiments. Paper with Michael Howes and @AustinVanLoon. Come join us! doi.org/10.1177/004912412513…
Austin van Loon retweeted
Blog post on mixed subjects (human & LLM) studies. Using prediction powered inference makes more sense than blindly substituting LLMs for humans, but still subect to garbage-in-garbage-out & won't resolve deeper questions about what we're trying to study
statmodeling.stat.columbia.e…
👀
“…the mixed [subjects] design… is the most reasonable approach I’ve seen in the LLMs for social science literature for integrating LLM simulations into confirmatory-style experiments…”
Also provides thoughtful reflections on the limitations of the approach!
Thanks to @JessicaHullman for a thoughtful dive into our Mixed Subjects Design paper with Michael Howes and @AustinVanLoon on the Stat Modeling blog!
Read the discussion ➡️ statmodeling.stat.columbia.e…
Austin van Loon retweeted
𝐃𝐨 𝐋𝐋𝐌𝐬 𝐡𝐚𝐯𝐞 𝐬𝐲𝐧𝐞𝐬𝐭𝐡𝐞𝐬𝐢𝐚? In a forthcoming paper in 𝐶𝑜𝑔𝑛𝑖𝑡𝑖𝑣𝑒 𝑆𝑐𝑖𝑒𝑛𝑐𝑒, we show yes! But this synesthesia deviates significantly from the color associations of ppl, incl. colorseeing, colorblind, and painters osf.io/preprints/psyarxiv/f9…
Austin van Loon retweeted
#LLMs offer cost-effective but potentially inaccurate predictions. Study says #LLM predictions be viewed as informative but imperfect, with humans serving as the gold standard in a mixed subjects design.
Read: spkl.io/6016fFsQA
Subscribe: spkl.io/6019fFsQh
ALT This image shows a womans face with digital connections around her.
Austin van Loon retweeted
Who Gets In? AI-generated essays in college admissions
Join us & @AustinVanLoon to learn about how GAI in college admission essays may reshape demographic makeup of admitted students & transform how merit is evaluated.
Wed 5/7, 10 am ET (Hybrid)
RSVP: tinyurl.com/PRIISM-5-7-25
ALT PRIISM Seminar Series with Austin van Loon: Competitive Self-Presentation and Generative AI in College Admissions
Amazing opportunity to work with a world-class psychologist, cutting edge computational social scientist, and an all-around fantastic person!
🚨🚨 I’m looking for a postdoctoral fellow to join my lab at Harvard's Psychology Department starting Fall 2025!🚨📣
Please retweet/share widely, and spread the word to interested candidates!
Application review begins April 30
Apply here: rb.gy/k7q9kf (1/2)