Serena Booth
@reniebird.bsky.social
CS Prof at Brown University, PI of the GIRAFFE lab, former AI Policy Advisor in the US Senate.
PhD at MIT CSAIL '23, Harvard '16, former Google APM. Dog mom to NSDTR Ducki.
PhD at MIT CSAIL '23, Harvard '16, former Google APM. Dog mom to NSDTR Ducki.
Reposted by Serena Booth
🚨The Formalism-Implementation Gap in RL research🚨
Lots of progress in RL research over last 10 years, but too much performance-driven => overfitting to benchmarks (like the ALE).
1⃣ Let's advance science of RL
2⃣ Let's be explicit about how benchmarks map to formalism
1/X
Lots of progress in RL research over last 10 years, but too much performance-driven => overfitting to benchmarks (like the ALE).
1⃣ Let's advance science of RL
2⃣ Let's be explicit about how benchmarks map to formalism
1/X
October 28, 2025 at 1:56 PM
🚨The Formalism-Implementation Gap in RL research🚨
Lots of progress in RL research over last 10 years, but too much performance-driven => overfitting to benchmarks (like the ALE).
1⃣ Let's advance science of RL
2⃣ Let's be explicit about how benchmarks map to formalism
1/X
Lots of progress in RL research over last 10 years, but too much performance-driven => overfitting to benchmarks (like the ALE).
1⃣ Let's advance science of RL
2⃣ Let's be explicit about how benchmarks map to formalism
1/X
Reposted by Serena Booth
I receive so many questions about what research in civil society is like -- Join our virtual panel to hear more about exactly that! @cdt.org @datasociety.bsky.social @aclu.org
With @thakurdhanaraj.bsky.social @alicetiara.bsky.social and @mkgerchick.bsky.social
To register:
cdt.org/event/advoca...
With @thakurdhanaraj.bsky.social @alicetiara.bsky.social and @mkgerchick.bsky.social
To register:
cdt.org/event/advoca...
Advocating with Evidence: Lessons for Tech Researchers in Civil Society
Advocating with Evidence: Lessons for Tech Researchers in Civil Society November 13, 2025, 10-11am ET online Civil society is struggling to address how technology and the tech industry contribute to e...
cdt.org
October 27, 2025 at 5:54 PM
I receive so many questions about what research in civil society is like -- Join our virtual panel to hear more about exactly that! @cdt.org @datasociety.bsky.social @aclu.org
With @thakurdhanaraj.bsky.social @alicetiara.bsky.social and @mkgerchick.bsky.social
To register:
cdt.org/event/advoca...
With @thakurdhanaraj.bsky.social @alicetiara.bsky.social and @mkgerchick.bsky.social
To register:
cdt.org/event/advoca...
Natalie's stuff makes me want to pivot my research to focus on algorithmic collusion 😅
And now might be a good time to mention, I’m on the faculty job market this year! I do work in human-AI collusion, collaboration and competition, with an eye towards building foundations for trustworthy AI. Check out more info on my website here! Nataliecollina.com
October 23, 2025 at 11:52 AM
Natalie's stuff makes me want to pivot my research to focus on algorithmic collusion 😅
Calarina Muslimani is presenting at the Cohere RL group tomorrow October 23rd at 12pm EST about our work on "Towards Improving Reward Design in RL: A Reward Alignment Metric for RL Practitioners".
This work won an ✨outstanding paper✨ award at RLC :D
Link: cohere.com/events/coher...
This work won an ✨outstanding paper✨ award at RLC :D
Link: cohere.com/events/coher...
Cohere Labs - Calarina Muslimani, PhD student, University of Alberta
Cohere Labs - Calarina Muslimani - Towards Improving Reward Design in RL: A Reward Alignment Metric for RL Practitioners
cohere.com
October 22, 2025 at 8:09 PM
Calarina Muslimani is presenting at the Cohere RL group tomorrow October 23rd at 12pm EST about our work on "Towards Improving Reward Design in RL: A Reward Alignment Metric for RL Practitioners".
This work won an ✨outstanding paper✨ award at RLC :D
Link: cohere.com/events/coher...
This work won an ✨outstanding paper✨ award at RLC :D
Link: cohere.com/events/coher...
I find CA Sen Weiner's AI policy prominence and upcoming challenge to Nancy Pelosi endlessly amusing. When we were 22, my tech bro friends and I mobilized to vote Sen Weiner into office bc of his campaign promise to keep SF nightlife open til 4. We just wanted to hang late at Armory Club, ya know.
October 21, 2025 at 3:30 PM
I find CA Sen Weiner's AI policy prominence and upcoming challenge to Nancy Pelosi endlessly amusing. When we were 22, my tech bro friends and I mobilized to vote Sen Weiner into office bc of his campaign promise to keep SF nightlife open til 4. We just wanted to hang late at Armory Club, ya know.
I wrote a (personal) blog post about my hopes and dreams for AI policy, my devastation after the US Election, and my process of picking myself off the floor by rebuilding an optimistic vision for AI scientists in government through education: simons.berkeley.edu/news/rebuild...
Rebuilding an Optimistic Vision for AI Policy
Recall November 6, 2024 — the day after the U.S. election. I was driving back to my home in Washington, DC, from Ohio with colleagues. I was heartbroken not because of the rebuke to my political party...
simons.berkeley.edu
October 13, 2025 at 1:56 PM
I wrote a (personal) blog post about my hopes and dreams for AI policy, my devastation after the US Election, and my process of picking myself off the floor by rebuilding an optimistic vision for AI scientists in government through education: simons.berkeley.edu/news/rebuild...
Heading to #COLM2025! Catch me at the workshop on bridging NLP and public opinion research (NLPOR). I’ll be on a panel lamenting our approach to indirect and inscrutable content moderation through RLHF.
October 10, 2025 at 12:26 AM
Heading to #COLM2025! Catch me at the workshop on bridging NLP and public opinion research (NLPOR). I’ll be on a panel lamenting our approach to indirect and inscrutable content moderation through RLHF.
Important update about my dog's life
October 5, 2025 at 1:50 PM
Important update about my dog's life
I'm nearly in Seoul for #CoRL2025, where I'll be presenting a talk on "Robots as the testbed for AI, AI as the testbed for Robots" at the Eval&Deploy workshop. Super excited to rep the newly formed Brown GIRAFFE Lab in the wild for the first time :D :D :D @confrobotlearning.bsky.social. Say hi!
September 26, 2025 at 8:52 AM
I'm nearly in Seoul for #CoRL2025, where I'll be presenting a talk on "Robots as the testbed for AI, AI as the testbed for Robots" at the Eval&Deploy workshop. Super excited to rep the newly formed Brown GIRAFFE Lab in the wild for the first time :D :D :D @confrobotlearning.bsky.social. Say hi!
The house approps budget is out, and it includes a very slight increase to the NIH budget. Senate approps committee held NSF funding steady. Don't back down from this fight!!!
We must celebrate and share our wins, too, to build and maintain momentum.
appropriations.house.gov/sites/evo-su...
We must celebrate and share our wins, too, to build and maintain momentum.
appropriations.house.gov/sites/evo-su...
appropriations.house.gov
September 2, 2025 at 1:29 PM
The house approps budget is out, and it includes a very slight increase to the NIH budget. Senate approps committee held NSF funding steady. Don't back down from this fight!!!
We must celebrate and share our wins, too, to build and maintain momentum.
appropriations.house.gov/sites/evo-su...
We must celebrate and share our wins, too, to build and maintain momentum.
appropriations.house.gov/sites/evo-su...
This is totally unacceptable.
I've followed Lisa Cook for a while since she has developed the leading perspective on AI over at the Fed: www.federalreserve.gov/newsevents/s...
I've followed Lisa Cook for a while since she has developed the leading perspective on AI over at the Fed: www.federalreserve.gov/newsevents/s...
August 26, 2025 at 3:36 AM
This is totally unacceptable.
I've followed Lisa Cook for a while since she has developed the leading perspective on AI over at the Fed: www.federalreserve.gov/newsevents/s...
I've followed Lisa Cook for a while since she has developed the leading perspective on AI over at the Fed: www.federalreserve.gov/newsevents/s...
Emma is amazing, you should apply.
🚨 New postdoc position in our lab at Berkeley EECS! 🚨
(please reshare)
We seek applicants with experience in language modeling who are excited about high-impact applications in the health and social sciences!
More info in thread
1/3
(please reshare)
We seek applicants with experience in language modeling who are excited about high-impact applications in the health and social sciences!
More info in thread
1/3
August 22, 2025 at 2:26 PM
Emma is amazing, you should apply.
Reposted by Serena Booth
Algorithmic Wage Discrimination is spreading beyond 'Gig' work:
Healthcare, Customer Service, & Logistics bosses are buying AI products that set compensation structures & wages using real time data. @wilneida.bsky.social & I did an audit & here are our findings
equitablegrowth.org/how-artifici...
Healthcare, Customer Service, & Logistics bosses are buying AI products that set compensation structures & wages using real time data. @wilneida.bsky.social & I did an audit & here are our findings
equitablegrowth.org/how-artifici...
How artificial intelligence uncouples hard work from fair wages through ‘surveillance pay’ practices—and how to fix it
How surveillance pay practices work, where they are increasingly deployed in the U.S. economy, and policy recommendations to ensure pay fairness and transparency.
equitablegrowth.org
August 21, 2025 at 7:32 PM
Algorithmic Wage Discrimination is spreading beyond 'Gig' work:
Healthcare, Customer Service, & Logistics bosses are buying AI products that set compensation structures & wages using real time data. @wilneida.bsky.social & I did an audit & here are our findings
equitablegrowth.org/how-artifici...
Healthcare, Customer Service, & Logistics bosses are buying AI products that set compensation structures & wages using real time data. @wilneida.bsky.social & I did an audit & here are our findings
equitablegrowth.org/how-artifici...
Reposted by Serena Booth
One study shows four in ten data brokers are leaving consumers exercising their privacy rights on read—in violation of California law. www.eff.org/deeplinks/2...
Data Brokers Are Ignoring Privacy Law. We Deserve Better.
Of the many principles EFF fights for in consumer data privacy legislation, one of the most basic is a right to access the data companies have about you. It’s only fair. So many companies collect
www.eff.org
August 9, 2025 at 8:01 PM
One study shows four in ten data brokers are leaving consumers exercising their privacy rights on read—in violation of California law. www.eff.org/deeplinks/2...
We shouldn't just be aiming for whistleblower protections. We should borrow from finance land and create whistleblower *incentives* (www.sec.gov/enforcement-...).
August 9, 2025 at 6:44 PM
We shouldn't just be aiming for whistleblower protections. We should borrow from finance land and create whistleblower *incentives* (www.sec.gov/enforcement-...).
The fun lil update is that this work won RLC Outstanding Paper in Emerging Topics :D. Work with Calarina Muslimani, Kerrick Johnstonbaugh, Suyog Chandramouli, @bradknox.bsky.social, and Matt Taylor.
August 8, 2025 at 12:54 AM
The fun lil update is that this work won RLC Outstanding Paper in Emerging Topics :D. Work with Calarina Muslimani, Kerrick Johnstonbaugh, Suyog Chandramouli, @bradknox.bsky.social, and Matt Taylor.
We have two papers at #RLC!
“Towards Improving Reward Design in RL: A Reward Alignment Metric for RL Practitioners”, lead by Callie Muslimani (Alberta)
&
“Goals vs. Rewards: Towards a Comparative Study of Objective Specification Mechanisms” lead by Septia Rani (CSU).
Come to our posters today!
“Towards Improving Reward Design in RL: A Reward Alignment Metric for RL Practitioners”, lead by Callie Muslimani (Alberta)
&
“Goals vs. Rewards: Towards a Comparative Study of Objective Specification Mechanisms” lead by Septia Rani (CSU).
Come to our posters today!
August 6, 2025 at 6:19 PM
We have two papers at #RLC!
“Towards Improving Reward Design in RL: A Reward Alignment Metric for RL Practitioners”, lead by Callie Muslimani (Alberta)
&
“Goals vs. Rewards: Towards a Comparative Study of Objective Specification Mechanisms” lead by Septia Rani (CSU).
Come to our posters today!
“Towards Improving Reward Design in RL: A Reward Alignment Metric for RL Practitioners”, lead by Callie Muslimani (Alberta)
&
“Goals vs. Rewards: Towards a Comparative Study of Objective Specification Mechanisms” lead by Septia Rani (CSU).
Come to our posters today!
Reposted by Serena Booth
Applications are open for the Doctoral Consortium at the 2025 Summit on Responsible Computing, AI, and Society rcais.github.io. Taking place October 27, 2025.
Applications received by September 15 will receive full consideration.
Applications received by September 15 will receive full consideration.
2025 Summit on Responsible Computing, AI, and Society
2025 Summit on Responsible Computing, AI, and Society
rcais.github.io
July 24, 2025 at 1:24 PM
Applications are open for the Doctoral Consortium at the 2025 Summit on Responsible Computing, AI, and Society rcais.github.io. Taking place October 27, 2025.
Applications received by September 15 will receive full consideration.
Applications received by September 15 will receive full consideration.
Fun fact - the ADA was a Republican endeavor, signed into law by H.W.
My great hope for the nation is that we can find shared priorities and get all parties to work in the interests of the people. And maybe we can fix that ridiculous $2000 asset limit sometime???
My great hope for the nation is that we can find shared priorities and get all parties to work in the interests of the people. And maybe we can fix that ridiculous $2000 asset limit sometime???
This Saturday, July 26, is the 35th anniversary of the Americans with Disabilities Act, which provided broad protections for disabled Americans. We have a lot planned this week to celebrate the anniversary, and to highlight our continued work at the intersection of #DisabilityRights & #TechPolicy!
July 22, 2025 at 2:51 PM
Fun fact - the ADA was a Republican endeavor, signed into law by H.W.
My great hope for the nation is that we can find shared priorities and get all parties to work in the interests of the people. And maybe we can fix that ridiculous $2000 asset limit sometime???
My great hope for the nation is that we can find shared priorities and get all parties to work in the interests of the people. And maybe we can fix that ridiculous $2000 asset limit sometime???
Reposted by Serena Booth
Excited to be at #ICML2025 to present our paper on 'pragmatic misalignment' in (deployed!) RAG systems: narrowly "accurate" responses that can be profoundly misinterpreted by readers.
It's especially dangerous for consequential domains like medicine! arxiv.org/pdf/2502.14898
It's especially dangerous for consequential domains like medicine! arxiv.org/pdf/2502.14898
July 15, 2025 at 5:15 PM
Excited to be at #ICML2025 to present our paper on 'pragmatic misalignment' in (deployed!) RAG systems: narrowly "accurate" responses that can be profoundly misinterpreted by readers.
It's especially dangerous for consequential domains like medicine! arxiv.org/pdf/2502.14898
It's especially dangerous for consequential domains like medicine! arxiv.org/pdf/2502.14898
I'll be presenting a position paper about consumer protection and AI in the US at ICML. I have a surprisingly optimistic take: our legal structures are stronger than I anticipated when I went to work on this issue in Congress.
Is everything broken rn? Yes. Will it stay broken? That's on us.
Is everything broken rn? Yes. Will it stay broken? That's on us.
July 14, 2025 at 1:01 PM
I'll be presenting a position paper about consumer protection and AI in the US at ICML. I have a surprisingly optimistic take: our legal structures are stronger than I anticipated when I went to work on this issue in Congress.
Is everything broken rn? Yes. Will it stay broken? That's on us.
Is everything broken rn? Yes. Will it stay broken? That's on us.
Reposted by Serena Booth
There may be some tiny opening to save the NSF funding. It is worth calling your Senators. This is a big deal if it can be saved.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
US senators poised to reject Trump’s proposed massive science cuts
Committee gives first hint that policymakers might preserve, rather than slash, funding for US National Science Foundation and other agencies.
www.nature.com
July 12, 2025 at 1:57 AM
There may be some tiny opening to save the NSF funding. It is worth calling your Senators. This is a big deal if it can be saved.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
This algorithmic monoculture and AI collusion stuff is so insidious. We need to get ahead of this, both from technical and legislative perspectives! IMO our existing anticompete laws are not sufficient for the newest forms of correlated decision-making and influence before us.
Excited to have this work out at ICML this year! Do LLMs make correlated errors? Yes, and those by the same company, and also more accurate/later generations are more correlated -- increasing algorithmic monoculture
arxiv.org/abs/2506.07962
arxiv.org/abs/2506.07962
July 3, 2025 at 5:34 PM
This algorithmic monoculture and AI collusion stuff is so insidious. We need to get ahead of this, both from technical and legislative perspectives! IMO our existing anticompete laws are not sufficient for the newest forms of correlated decision-making and influence before us.
Reposted by Serena Booth
I've struggled to announce this amidst so much dark & awful going on in the world, but with 1mo to go, I wanted to share that: (i) I finally graduated; (ii) In August, I'll begin as an assistant professor in the CS dept. of the National University of Singapore.
June 30, 2025 at 2:22 PM
I've struggled to announce this amidst so much dark & awful going on in the world, but with 1mo to go, I wanted to share that: (i) I finally graduated; (ii) In August, I'll begin as an assistant professor in the CS dept. of the National University of Singapore.
Complicated implications, but compelling research. I'd still bet the explanation requirement of ECOA has prevented a lot of discrimination because it puts an obstacle in place, but I wonder how well we can detect discrimination from human decision makers with explanations!
Explanations don't help us detect algorithmic discrimination. Even when users are trained. Even when we control their beliefs. Even under ideal conditions... 👇
Right to explanation laws assume explanations help people detect algorithmic discrimination.
But is there any evidence for that?
In our latest work w/ David Danks @berkustun, we show explanations fail to help people, even under optimal conditions.
PDF shorturl.at/yaRua
But is there any evidence for that?
In our latest work w/ David Danks @berkustun, we show explanations fail to help people, even under optimal conditions.
PDF shorturl.at/yaRua
June 24, 2025 at 10:36 PM
Complicated implications, but compelling research. I'd still bet the explanation requirement of ECOA has prevented a lot of discrimination because it puts an obstacle in place, but I wonder how well we can detect discrimination from human decision makers with explanations!