PhD student @UW | algorithms & AI for pro-social objectives Ex- @AIatMeta @MSFTResearch, @umsi, @ashokauniv Also on https://nitter.cf/t.co/7DkUJLjFgm
Seattle, WA
Joined June 2020
- Tweets663
- Following1.3K
- Followers998
- Likes4.2K
Pinned Tweet
Major personal update:
After a bit of hard work and a whole lot of extreme good luck, I'm happy to share that I'll be joining @UW (@uw_ischool) to start my PhD this Fall!
It has been an extremely rewarding journey so far and I look forward to the next phase!
Soham De retweeted
Could social media make us less polarized instead of more?
We tested 5 algorithms on 3 platforms with 10,000 people for 6 months during the 2024 election, and found that the answer is yes.
🧵
Soham De retweeted
🚨🎉Excited to announce that our paper “Grok in the Wild: Characterizing the Roles and Uses of Large Language Models on Social Media” is accepted at @icwsm 2026! In this paper, we investigate how, when, and to what effect Grok is used on X.
🧵1/n thread
Our auditing tools are integrated into a deliberation platform @StanfordDDL used in 50+ countries, allowing moderators to measure and improve representation in real time (thanks, Harshvardhan Agarwal and others!) (6/7)
All thanks to my amazing collaborators @Effeietsanders @ashishgoel @SmithaMilli @ArielProcaccia and Alice Siu!!!
This was some work I did @AIatMeta this summer (more on the way!)
Read more here: arxiv.org/abs/2511.04588 (7/7)
Soham De retweeted
(please reshare) I'm recruiting multiple PhD students and Postdocs @uwcse @uwnlp
(bdata.uw.edu). Focus areas incl. psychosocial AI simulation and safety, Human-AI collaboration.
PhD: cs.washington.edu/academics/…
Postdocs: docs.google.com/document/d/1…
Soham De retweeted
Today we're releasing Community Alignment - the largest open-source dataset of human preferences for LLMs, containing ~200k comparisons from >3000 annotators in 5 countries / languages!
There was a lot of research that went into this... 🧵
interesting day on the internet for @actuallysoham 🙃
Here’s our articulation of what the future of Community Notes in an LLM era could look like 👀 :
Very interested to hear your thoughts!
Brand new paper on Arxiv with @Li_Haiwen_ @actuallysoham Manon Revel @andreas_h0wpt @__bradmiller__ @kcoleman @msaveski: Scaling Human Judgment in Community Notes with LLMs
Very bullish on the LLM-assisted future of Community Notes 📈
Currently traveling to Sydney, to present this paper at @TheWebConf 🚀If you're around and interested in Community Notes and/or social media algorithms in general, I would love to chat!
Full paper: dl.acm.org/doi/10.1145/36964…
#www2025 #thewebconf2025
🚨🚨 New preprint alert! 🚨🚨
Our new preprint w/ @jaybaxter, @bakkermichiel @msaveski (dream team!) proposes a framework to generate ✨Supernotes✨ - AI-generated notes that synthesize information from several existing @communitynotes and are written to foster consensus among a diverse set of users.
A thread 🧵... (1/10)
ALT Supernotes Framework
This project screams @kabirahuja004 in every way!
I had always wanted to work on something that can combine my love for fiction and NLP research, making this project a lot of fun. Huge thanks to the wonderful @melaniesclar and @tsvetshop!
We welcome any feedback and questions -- don't hesitate to reach out!
16/16
Soham De retweeted
Community Notes scale and speed up fact-checking, and do so in a more trustworthy way.
But to handle AI-scale misinfo slop, we’ll probably want AI assistants helping to write notes. “Supernotes” is one proposal of how this could work:
Beating a dead horse atp but Community Notes cited as exemplary use of AI for epistemics in Preparing for the Intelligence Explosion (pdf below)
Cc @jaybaxter
Soham De retweeted
My talk from @IASEAIorg on "Dynamic Preferences in AI Alignment: A Deliberative Democracy Lens" is up here
video.oecd.org/embed-or-en-f…
First pass sharing these ideas, more fleshed-out write-up to come~
Soham De retweeted
🧵 Thrilled to share our new @NatureComms paper on how different approaches to fact-checking affect echo chambers on Twitter/X! With the amazing @junsolk, Zhao Wang, @hshi420, and @hsinkengling.
When someone gets fact-checked by an individual (via direct replies), they retreat into echo chambers. But when they receive a collective fact-check from many people (via Community Notes), they're more likely to stay engaged with diverse perspectives.
Full paper here: rdcu.be/d7mbN
Soham De retweeted
In a new blog post, I discuss the emerging literature on simulating human behavior with LLMs.
doomscrollingbabel.manoel.xy…
