@GhateKshitish

PhD student @UWCSE | @MSFTResearch Intern | MLT Grad student @LTIatCMU | Human-AI interaction and alignment

kshitishghate.github.io/
Joined October 2022
  • Tweets76
  • Following277
  • Followers135
8Media
🚨New paper: Reward Models (RMs) are used to align LLMs, but can they be steered toward user-specific value/style preferences? With EVALUESTEER, we find even the best RMs we tested exhibit their own value/style biases, and are unable to align with a user >25% of the time. 🧵
1
18
72
7,065
Kshitish Ghate retweeted
Shuhaib Mehri
@shuhaibmehri8h
User simulators are only as faithful as the user representations they are conditioned on. Missing details force simulators to fill gaps with assumptions, leading to generic behavior⚠️ Introducing UserRepGen-4B, which generates representations optimized for simulation fidelity.
3
24
6
105
5,309
Kshitish Ghate retweeted
Andy Liu@uilydnaOct 13
LLM developers train their models to follow prosocial values, but post-training can influence model behavior in unexpected ways. In “Predicting Alignment Generalization with Value Representations”, we ask whether downstream effects of value alignment can be predicted in advance.
3
19
5
77
7,956
Kshitish Ghate retweeted
Chan Young Park@chan_young_parkOct 8
Second piece of news, I’m looking to recruit multiple PhD students to work on human-centered AI/alignment/LLM social reasoning this cycle @UTiSchool While I’m sadly missing #COLM2026 I'll be heading to #EMNLP2026 🇭🇺 Whether you'll be there or not, reach out if you're interested!
4
32
1
200
10,223
Will be presenting PPol at #COLM2026 Social Sim Workshop. Come say hi!! Generate realistic user personas with PPol, which is now a pip-installable package. Check it out and let us know what you think :)
Harshita Chopra @ COLM 🌉@chopra_harshitaOct 5
Excited to present Persona Policies (𝗣𝗣𝗼𝗹) at the #COLM2026 Social Sim Workshop this Friday! 🌉 More news: PPol is now super easy to use! Start simulating more realistic user behaviors using 👉 𝐩𝐢𝐩 𝐢𝐧𝐬𝐭𝐚𝐥𝐥 𝐩𝐩𝐨𝐥 👈 Check out our GitHub repo for examples!
1
23
1,087
Kshitish Ghate retweeted
Stella Li @ COLM
@StellaLisyOct 2
Deep Reasoning in General Purpose Agents via Structured Meta-Cognition Room: Continental Ballroom 4 Fri 10/9 - 4:00PM Workshop on Agent Behavior arxiv.org/abs/2605.11388

Deep Reasoning in General Purpose Agents via Structured Meta-Cognition

Humans intuitively solve complex problems by flexibly shifting among reasoning modes: they plan, execute, revise intermediate goals, resolve ambiguity through associative judgment, and apply...

arxiv.org
1
2
5
229
Kshitish Ghate retweeted
Shuhaib Mehri
@shuhaibmehriAug 25
MultiSessionCollab has been accepted to #EMNLP2026🎉 Memory systems are typically optimized for information recall and question-answering, and we show that this doesn't really help with improving user interactions. Instead, we optimize memory for improving human-AI collaboration over time with: (1) MultiSessionCollab: a benchmark that evaluates how well agents can learn user preferences and improve collaboration over time (2) A training framework that leverages user feedback signals to improve agent memory It's super important to build AI that's useful to people in the real world. But it is challenging, and benchmarks don't always capture that very well. This work is our attempt to take a step in that direction, and I'm really excited to keep building on it!
3
5
1
45
2,677
Kshitish Ghate retweeted
Jihoon Tack
@jihoontackJul 27
Excited to share my first work after joining @MSFTResearch! LLMs have entered the agentic era, and we now collaborate with agents on complex tasks over many turns of interaction. But does your agent actually follow what you intended? We show where agents get lost: LLMs Get Lost in Evolving User Intent. 📄 arxiv.org/abs/2607.20734 🧵 1/N
4
41
6
216
30,324
Kshitish Ghate retweeted
Elinor@elinorpd_Jul 23
Pluralistic alignment is thriving as a research agenda yet failing at its goal: making the AI systems people actually use more pluralistic🌈 🚨New position paper: we argue adoption in deployed models should be the fields main goal & we provide a roadmap of how to get there 🧵1/
7
28
10
122
35,792
Which cultures do LLMs best align with on open-ended questions? Why? Can we change that default? We study this in our new work: "Steering LLMs for Culturally Localized Generation" (COLM 2026), from my time at Google DeepMind! 🧵 arxiv.org/abs/2603.23301
1
11
1
87
9,020
Kshitish Ghate retweeted
Devansh Jain
@devanshrjainJun 25
i like my training like i like my agents: memory-native 💤
Letta
@Letta_AIJun 25

How can we build agents that learn like people?

Article · @Letta_AI
1
4
186
Excited to share our new paper on Persona Policies (PPol)! We evolved user simulator personas that are diverse, human-like, and task-preserving. Training agents on PPol rollouts improves task performance with out-of-distribution users vs. training on default simulators!🧵
Harshita Chopra @ COLM 🌉@chopra_harshitaMay 16
Are you training or evaluating agents with LLM-based user simulators? Most simulators inherit the behavior of their underlying models: cooperative, clear, and homogeneous - which is unrealistic! 🤖 Humans are messy: they falter, forget, push back, and behave in ways that are difficult to define. Manually writing personas becomes brittle and hard to scale. 😣 We introduce 𝗣𝗲𝗿𝘀𝗼𝗻𝗮 𝗣𝗼𝗹𝗶𝗰𝗶𝗲𝘀 (𝗣𝗣𝗼𝗹): an evolutionary framework that automatically discovers behaviors and instructions to generate diverse human-like user personas for any given task – ✨grounded in real dialogue traces✨
Scatter plot of humans, existing user simulators, and our PPol in the behavioral feature space.

ALT Scatter plot of humans, existing user simulators, and our PPol in the behavioral feature space.

1
2
26
3,687
Joint work with @chopra_harshita @aylin_cim @yoshi_kohno @chirag_shah @natashajaques !
2
112
We are extending the deadline to May 8th! One more week to go! We look forward to seeing everyone at #ICML2026.
Pluralistic Alignment Workshop@pluralistic_aiMar 30
🚨We are excited to announce the 2nd Pluralistic Alignment Workshop at #ICML2026 in Seoul! Submissions: openreview.net/group?id=ICML… 🗓️Deadline: May 3 More details pluralistic-alignment.github… We invite work on pluralistic alignment across technical, philosophical, societal perspectives!
5
16
2,951
Kshitish Ghate retweeted
Andy Liu@uilydnaApr 20
I'll be in Rio this week for #ICLR2026 to present "Generative Value Conflicts Reveal LLM Priorities" (Friday morning, P4-#4105). Happy to chat anything related to LLM alignment, human-AI interaction, or multi-agent systems - feel free to DM if interested!
3
1
29
4,198
🚨We are excited to announce the 2nd Pluralistic Alignment Workshop at #ICML2026 in Seoul! Submissions: openreview.net/group?id=ICML… 🗓️Deadline: May 3 More details pluralistic-alignment.github… We invite work on pluralistic alignment across technical, philosophical, societal perspectives!
1
13
4
57
21,311
Kshitish Ghate retweeted
Natasha Jaques
@natashajaquesMar 20
The paper I’ve been most obsessed with lately is finally out: nbcnews.com/tech/tech-news/a…! Check out this beautiful plot: it shows how much LLMs distort human writing when making edits, compared to how humans would revise the same content. We take a dataset of human-written essays from 2021, before the release of ChatGPT. We compare how people revise draft v1 -> v2 given expert feedback, with how an LLM revises the same v1 given the same feedback. This enables a counterfactual comparison: how much does the LLM alter the essay compared to what the human was originally intending to write? We find LLMs consistently induce massive distortions, even changing the actual meaning and conclusions argued for.
46
393
71
1,473
262,415
Kshitish Ghate retweeted
Zora Wang@ZhiruoWFeb 11
‼️Position: AI coding agent research needs recalibration. We've heavily optimized for solo autonomy, and far less for designing agents that empower the humans using them. It’s time to build human-centered coding agents. 🧵
25
63
10
316
51,170
Kshitish Ghate retweeted
Andy Liu@uilydnaFeb 4
Our work eliciting which values LLMs prioritize via simulated value conflicts was accepted to #ICLR2026 ! See you in🇧🇷
Andy Liu@uilydna2 Oct 2025
🚨New Paper: LLM developers aim to align models with values like helpfulness or harmlessness. But when these conflict, which values do models choose to support? We introduce ConflictScope, a fully-automated evaluation pipeline that reveals how models rank values under conflict.
xkcd.com/1613

ALT xkcd.com/1613

1
2
1
30
5,065
Kshitish Ghate retweeted
Ethan Shen
@ethanlshenJan 27
Today, we release SERA-32B, an approach to coding agents that matches Devstral 2 at just $9,000. It is fully open-source and you can train your own model easily - at 26x the efficiency of using RL. Paper: allenai.org/papers/opencodin… Here’s how 🧵
Ai2
@allen_aiJan 27
Introducing Ai2 Open Coding Agents—starting with SERA, our first-ever coding models. Fast, accessible agents (8B–32B) that adapt to any repo, including private codebases. Train a powerful specialized agent for as little as ~$400, & it works with Claude Code out of the box. 🧵
26
88
13
680
97,269
Kshitish Ghate retweeted
Andy Liu@uilydna24 Oct 2025
Replying to @jifan_zhang @johnschulman2 @sleight_henry @TheAndiPenguin
Congrats on the paper! We've been working in a similar direction (evaluating value prioritization in LLMs - arxiv.org/abs/2509.25369), would love to chat if you're still thinking about this

Generative Value Conflicts Reveal LLM Priorities

Past work seeks to align large language model (LLM)-based assistants with a target set of values, but such assistants are frequently forced to make tradeoffs between values when deployed. In...

arxiv.org
7
22
2,103
Load more