Caltech LCSSP (@CaltechLCSSP) header
Caltech LCSSP (@CaltechLCSSP) profile picture

Linde Center for Science, Society, & Policy at Caltech | Research & debate on topics at intersection of science, society, & policy | R.M. Alvarez & F. Eberhardt

Caltech, Pasadena CA
Joined November 2022
Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)Photo by Caltech LCSSP (@CaltechLCSSP)
Can AI help us breathe cleaner air? ๐ŸŒ New paper from @CaltechLCSSP's Ramit Debnath, @Cambridge_Uni & coauthors found that a small data choice changed accuracy 50โ€“75%. For climate AI, problem design matters as much as model design. ๐Ÿ“– Open Access: iopscience.iop.org/article/1โ€ฆ
2
75
๐ŸŽ‰Congratulations to @CaltechLCSSP's, @DannyCEbanks & Jonathan Katz, with coauthor Gary King for turning that idea into remarkable impact: โœ… 26.9M Google impressions, 1.1M+ visits from ~690K users, Recognition from Drudge Report and Jimmy Kimmel! dannyebanks.com/map/
4
265
Congratulations to @CaltechLCSSP's @DrLucyJones, @Caltech on launching the Center for Disaster Resilience, uniting scientists, engineers, & community leaders to advance preparedness, response, & recovery from our regionโ€™s greatest hazards. Looking forward to the impact ahead! ๐Ÿ”ฅ
2
4
767
๐Ÿ’กAI can accelerate development, but without inclusion it may deepen existing inequalities. @CaltechLCSSP's Ramit Debnath & @Cambridge_Uni's Abdullah Safir argue that AIโ€™s future depends on whose voices, knowledge, and priorities are represented. ๐Ÿ”—universityworldnews.com/postโ€ฆ
1
3
116
Missed the election debriefs?๐Ÿ—ณ๏ธ @CaltechLCSSP's @rmichaelalvarez discussed Californiaโ€™s election results, governorโ€™s race, congressional districts, vote counting, & what outcomes could mean statewide and nationally. ๐ŸŽงlaist.com/news/airtalk-nov-4โ€ฆ ๐Ÿ“บ foxreno.com/news/mornings-onโ€ฆ
1
4
216
๐ŸŽ‰Big news for AI and society: @Cambridge_Uni's @CHIA_Cambridge & @Caltech's @CaltechLCSSP are expanding their collaboration to advance research in explainable AI, AI alignment, human-inspired AI, & opportunities for students and researchers. Learn more: chia.cam.ac.uk/news/cambridgโ€ฆ
1
1
5
576
๐Ÿ“ฃNew research from @CaltechLCSSP's @RKocielnik & @rmichaelalvarez, @Caltech with coauthors @p_song1, @pengrui_han, Myrl G. Marmarelis, Ramit Debnath, & Dean Mobbs.๐Ÿง โš ๏ธ ๐Ÿ”—Website with paper link: psychology-of-ai.github.io/rโ€ฆ
New paper: "Representational Control over Self-Report & Behavior Coherence in LLM Risk-Taking" ๐ŸŽฏ ๐——๐—ผ ๐—Ÿ๐—Ÿ๐— ๐˜€ ๐—ต๐—ฎ๐˜ƒ๐—ฒ ๐—ฎ ๐—ฟ๐—ถ๐˜€๐—ธ-๐˜๐—ฎ๐—ธ๐—ถ๐—ป๐—ด ๐—ฑ๐—ถ๐˜€๐—ฝ๐—ผ๐˜€๐—ถ๐˜๐—ถ๐—ผ๐—ป ๐˜๐—ต๐—ฎ๐˜ ๐—ฐ๐—ผ๐—ต๐—ฒ๐—ฟ๐—ฒ๐—ป๐˜๐—น๐˜† ๐—ด๐˜‚๐—ถ๐—ฑ๐—ฒ๐˜€ ๐—ฏ๐—ผ๐˜๐—ต ๐˜๐—ต๐—ฒ๐—ถ๐—ฟ ๐˜€๐—ฒ๐—น๐—ณ-๐—ฟ๐—ฒ๐—ฝ๐—ผ๐—ฟ๐˜๐˜€ ๐—ฎ๐—ป๐—ฑ ๐˜๐—ต๐—ฒ๐—ถ๐—ฟ ๐—ฟ๐—ถ๐˜€๐—ธ๐˜† ๐—ฑ๐—ฒ๐—ฐ๐—ถ๐˜€๐—ถ๐—ผ๐—ป๐˜€? ๐—ง๐—Ÿ;๐——๐—ฅ ๐—ช๐—ต๐˜† ๐—ถ๐˜ ๐—บ๐—ฎ๐˜๐˜๐—ฒ๐—ฟ๐˜€: LLMs often say one thing about themselves and do another. In risk-taking the gap is costly: a clinical assistant or a trading agent can call itself cautious and still make risky decisions. ๐—ช๐—ต๐—ฎ๐˜ ๐˜„๐—ฒ ๐˜€๐—ต๐—ผ๐˜„: No shared disposition: inside the model, stated risk-taking and risky choices are represented separately. Task-specific instructions seem to move both, but through misleading shortcuts like producing larger numbers. ๐—ง๐—ต๐—ฒ ๐—ณ๐—ถ๐˜…: Steering internal representations of stated and enacted risk together makes what a model says match what it does, without retraining. ๐—ง๐—ต๐—ฒ ๐—ฑ๐—ฎ๐—ป๐—ด๐—ฒ๐—ฟ: The same mechanism can make a model say the opposite of what it does (report low risk while taking more, or the reverse), in every model we tested. ๐Ÿ“„ Paper: arxiv.org/abs/2610.04125 ๐ŸŒ Website: psychology-of-ai.github.io/rโ€ฆ w/ @p_song1 @pengrui_han @MyrlGMarmarelis, Ramit Debnath, Dean Mobbs, @rmichaelalvarez, @CaltechLCSSP and @Cambridge_Uni's Centre for Human-Inspired AI (CHIA) Technical details below ๐Ÿ‘‡ #AgenticAI #AISafety #AIAlignment #Interpretability #ActivationSteering
3
8
284
๐Ÿ—ณ๏ธToday is Election Day! Democracy isn't something we watch, it's something we participate in. Every eligible voter deserves the chance to cast a ballot in elections that are: โš–๏ธ Fair โœ… Secure ๐Ÿ‘ฅ Accessible Whether voting in person or by mail, engagment matters.
1
1
9
2,241
At @VotebeatUS's Oct 27 virtual event, @CaltechLCSSP's @rmichaelalvarez joined Bob Page and @aztammyp to discuss potential concerns and what voters can expect on Election Day. Thanks to host @JessicaHuseman for moderating! ๐Ÿ“บWatch the video: youtube.com/watch?v=cn8Qpdiaโ€ฆ
LAST CHANCE! Sign up to attend a free event with @JessicaHuseman, @aztammyp, @rmichaelalvarez, and Bob Page, registrar of voters for Orange County, California. link below โฌ‡๏ธ
2
153
โœจA highlight of this week was @CaltechLCSSP Research Affiliate Ramit Debnath, @Cambridge_Uni & CHIA delivering the keynote at the #ClimateNLP workshop at #EMNLP2026 in Budapest.๐ŸŒ The presentation drew on a long-running collaboration between CHIA & @CaltechLCSSP.๐Ÿค๐Ÿ“š
1
4
215
๐Ÿ—ณ๏ธ Why can the election lead change? โ˜‘๏ธ Check out the interactive web simulation from @CaltechLCSSP's @rmichaelalvarez, @Caltech built with synthetic data and based on research. ๐Ÿ“– Read: doi.org/10.1177/106591292110โ€ฆ ๐Ÿ–ฅ๏ธ Simulation: rmichaelalvarez.github.io/blโ€ฆ
Interested in learning more about why we see "blue shifts" and "red shifts" in post-election voting returns? Give our simulation a spin, it's based on published research with @yimenglipolisci and synthetic data. rmichaelalvarez.github.io/blโ€ฆ
2
108
๐ŸŽ‰Congratulations to @CaltechLCSSP Board Member @HannahDruck & Prakash Mishra, @Caltech for launching the PERMIT Lab with a $7M gift from the WHH FOUNDATION to advance research at the intersection of environmental economics, AI, & sustainability. ๐ŸŒŽ๐Ÿง ๐ŸŒฑ Learn more below:
A new Caltech lab will soon inform environmental policy. The lab is being established through a $7 million infusion of funding from the WHH Foundation, a private family philanthropic organization. Read more: caltech.edu/about/news/7-milโ€ฆ
1
5
502
There is still time to register for the Caltech Election Night Watch Party! ๐Ÿ“… November 3, 2026 ๐Ÿ•– 7:00 PM - 10:00 PM ๐Ÿ“ Avery Courtyard, Caltech Join fellow @Caltech community members and follow federal, state, & local election results together. ๐Ÿ”—forms.gle/fh4RV9QU2zUyxKK26
๐Ÿ“ฃ Registration is Open! Join the @Caltech Election Night Watch Party! ๐Ÿ—ณ๏ธ Follow federal, state, & local election results: ๐Ÿ“… Nov. 3, 2026 โฐ 7โ€“10 PM ๐Ÿ“ Avery Courtyard ๐Ÿ• Free food โ€ข ๐ŸŽฎ Games โœ… Register now: forms.gle/fh4RV9QU2zUyxKK26. ๐Ÿ”ด Nonpartisan
3
144
๐ŸŽ‰ New publication in Environmental Research Letters! Analyzing 232 Chinese cities with explainable ML, @CaltechLCSSP's Ramit Debnath, @CHIA_Cambridge & coauthors identify pathways for cities to go from high- to low-carbon growth in 10 years.๐Ÿญ ๐ŸŒฑ ๐Ÿ”— iopscience.iop.org/article/1โ€ฆ
2
112
๐ŸŽ‰Congratulations to @CaltechLCSSP Research Affiliate @AdamWierman, @Caltech (with Steven Low) on advancing innovative solutions to make the electrical grid more efficient, sustainable, & resilient. ๐Ÿขโšก Learn more about the gift that supports this: giving.caltech.edu/news/The-โ€ฆ
5
161
๐Ÿ” AI Is Changing Political Campaigns, Even at the Local Level @CaltechLCSSP's @rmichaelalvarez, @Caltech says AI is on the rise in local races, even school board elections. Why? โœ… Affordable โœ… Accessible โœ… Convincing ๐Ÿ”—Learn more: youtube.com/watch?v=_lnW1GwGโ€ฆ
1
3
177