Human first. PhD. Asst Prof at @uw_ischool Past @MIT @CCCatMIT @Harvard_IACS @dccuc @GoogleAI @WiMLworkshop Board & https://xitter.cf/t.co/ixad3x6I5l
Cambridge, MA
Joined May 2015
- Tweets305
- Following434
- Followers400
- Likes1.7K
Belén C. Saldías F. retweeted
Carlos Alcaraz on the possibility of missing Roland Garros
“I have a long career ahead of me. Forcing for Roland-Garros could penalize me enormously in the future”
“We'll see what the exams show. But I prefer to come back a bit later rather than rush.”
Belén C. Saldías F. retweeted
Got asked in a review for my ICML paper whether "there are realistic tasks where one needs to sample from a probability distribution given by its unnormalized density" (rephrased for anonymity). Are we cooked?
Belén C. Saldías F. retweeted
No one tells you that once you find the right routine for your neurospicy brain, you'll cry.
Because life WAS harder.
You can finally focus and do things easily without getting overwhelmed.
If you're struggling, try The Fabulous.
Belén C. Saldías F. retweeted
I interviewed 103 candidates for the MTS role in the last 6 months, we ended up only making single digit offers.
This is what I am looking for:
- how do they approach a problem. do they jump to the solution or first think about *how to measure* and how to set up evals
- how good are the fundamentals. everyones uses torchtune or verl but can you go beyond the wrappers and setup yolo runs from existing curves
- culture vibes. imo this is usually data hygiene, thorough documentation, and reproducible expts
Belén C. Saldías F. retweeted
How does training data shape model behavior? Well, it’s complicated…
1/10
Belén C. Saldías F. retweeted
As of June 2025, 66% of Americans have never used ChatGPT.
Our new position paper, Attention to Non-Adopters, explores why this matters: LLM research is being shaped around adopters, leaving non-adopters’ needs and key research opportunities behind.
arxiv.org/abs/2510.15951
ALT A circular flow diagram that compares current and proposed practices for LLM development using data from adopters and non-adopters. Three gray boxes represent current practices: “R&D,” “Chat Models,” and “Adopters’ Needs and Usage Data,” connected in a clockwise loop with black arrows. A blue box labeled “Non-adopters’ Needs and Usage Data” adds a proposed feedback path, shown with blue arrows, linking non-adopter data back to R&D and adopters’ data.
Belén C. Saldías F. retweeted
📢 Please retweet: We're recruiting **PhD students and Postdocs** at UC Berkeley & UCSF!
Please apply if you are interests in AI for healthcare, statistics/causal inference, or medical vision-language models.
For more details, check out this link: forms.gle/1rZHBoH8f8fu27F48
Belén C. Saldías F. retweeted
📣📣Do language models only trained to predict truth from falsehood still display a political bias? 🤔Surprisingly, it seems so! Our paper "On the Relationship between Truth and Political Bias in Language Models”, has been accepted to EMNLP 2024
Belén C. Saldías F. retweeted
📢 Who contributes most to politically polarized speech on Reddit and Twitter? As we come into the final stretch of the presidential election season, we present our work out of @CCCatMIT a large-scale analysis (2.5 billion tweets/comments) of politically polarized speech.
Belén C. Saldías F. retweeted
1 / Last week, MIT CCC graduate students @bcsaldias, @Schropes, @elinorpd_ , @wwbrannon, @hjian42, and CCC research engineer Doug Beeferman presented at the the 10th International Conference for Computational Social Science @IC2S2 at the University of Pennsylvania.
How could we empower communities and make of the internet a safer place for youth? Algorithms are increasingly shaping their lives.
At @IC2S2 covering design opportunities for our new decentralized social network!
📜 : arxiv.org/abs/2406.08420
@CCCatMIT : ccc.mit.edu/project/project-…
Belén C. Saldías F. retweeted
Still buzzing from yesterday’s workshop on LLMs in research, and looking forward to continuing the conversation at #CHI2024 🌺
That’s a wrap on the LLMs as Research Tools workshop at #CHI2024! Huge TY to our AMAZING participants for making today such a success @marianneaubin @Schropes @jie_gao_056 @_ziv_e @simonperrault @dmimno @roboticwrestler @barkhuus @hanlinlil1578
Belén C. Saldías F. retweeted
Para mis seguidores hispanohablantes!
"La #InteligenciaArtificial generativa no es el heraldo de la muerte del arte, sino más bien un nuevo medio con sus propias posibilidades", dicen científicos del @MIT en @ScienceMagazine
Hemos preguntado a una de las autoras, @Schropes, y a #ChatGPT
agenciasinc.es/Reportajes/Si…
Belén C. Saldías F. retweeted
Leonard Francis Vibbi, a grad student at the Media Lab + research assistant at @mitgovlab, has been selected as a Governance Innovation Research Fellow for a new initiative between MIT GOV/LAB, @MISTIatMIT, + the Priscilla King Gray Public Service Center. media.mit.edu/posts/leonard-…
Belén C. Saldías F. retweeted
During #TEDxBentleyU, @CCCatMIT PhD student @bcsaldias invited the audience to consider the cultural biases built into tools like ChatGPT, and shared some personal reflections on the hidden challenges of using AI. youtube.com/watch?v=p1DM51vq…