Eunkyu Eunice Park

current Ph.D. Candidate at @SeoulNatlUni @SNUVLLab
current Visiting Researcher at @cmuhcii
prev @columbia @CUSEAS

[Eunkyu Eunice Park @uunicee_]

🧵Sharing our most-recent work! Critical or Compliant? The Double-Edged Sword of Reasoning in Chain-of-Thought Explanations

[Xuhui Zhou @nlpxuhui]

Does LLM really need to be a helpful assistant all the time?

No. If you want to simulate people, “perfectly helpful” could be the wrong objective.

Meet OdysSim, a journey toward LLMs beyond assistants, as behavioral foundation models (10B tokens of real human behavior; 23 sim benchmarks, finally in one place. new open models: outperform or on par with GPT-5.5, Gemini 3.1, or Claude Opus 4.7 in many behavior-sim dimensions).

Human behavior simulation is becoming essential.

Agent evaluation needs realistic users before real users show up. Medical and classroom training need realistic patients and students. Social science needs synthetic participants at scale.

But real people are not ideal assistants.

Real patients panic or ignore good advice. Real students misunderstand. Real customers are vague, picky, impatient, or simply leave. Human behavior is messy, diverse, and often imperfect.

Frontier LLMs are getting better at math, code, and long-horizon tasks. They are NOT getting better at simulating human behavior. If anything, they drift the other way: more assistant-ish, more homogeneous, fewer of the errors and quirks real humans show.

This is no accident. The whole pipeline is built for helpfulness and task success, not behavioral realism.
And you can't prompt your way out of that.

So we rethink the recipe from scratch and release:

🧠 The OdysSim corpus: 21.4M real human interactions (~10B tokens) from 62 sources, every conversation retrofitted with social grounding (who is talking, and why)
📏 SOUL-Index: 23 human-behavior benchmarks unified into one suite across 5 axes
🤖 OSim-8B: open weights; tops more SOUL-Index benchmarks than any frontier model, acts more like a real user than any of them on τ-bench (nearly matching real humans in the reaction dimension), and writes far more human-like text along the way.

[Xuhui Zhou @nlpxuhui]

Creating user simulators is a key to evaluating and training models for user-facing agentic applications. But are stronger LLMs better user simulators?

TL;DR: not really.

We ran the largest sim2real study for AI agents to date: 31 LLM simulators vs. 451 real humans across 165 tasks.

Here's what we found (co-lead with @sunweiwei12).

[Dayoon Ko @dayoon12161]

🎉 Our paper has been accepted to #ICLR2026! 😆💖
This work was done during my internship at LG AI Research – Superintelligence Lab. As summarized in the project:

Deep research requires broad evidence coverage and reliable synthesis.
HybridDeepSearcher achieves both by parallel retrieval for breadth with sequential reasoning for depth, supporting scalable search.

🔗 Project page: https://t.co/vKPfc0hAe1
📄 OpenReview: https://t.co/pWBlwdGjvL

Huge thanks to my mentors and co-workers for their guidance and support throughout this project. We also plan to release related work soon. Stay tuned! 😊

[Maarten Sap (he/him) @MaartenSap]

We are recruiting translators (any language/domain) for an interview study about translation technologies here!
Form: https://t.co/hBcZ1OLgiN
We will be grateful for all input!
1. Interviews will be on Zoom.
2. Participants will be given a $40 𝗔𝗺𝗮𝘇𝗼𝗻 𝗚𝗶𝗳𝘁 𝗰𝗮𝗿𝗱!

[Stella Li ✈️ ICML🇰🇷 @StellaLisy]

🤔💭What even is reasoning? It's time to answer the hard questions!

We built the first unified taxonomy of 28 cognitive elements underlying reasoning

Spoiler—LLMs commonly employ sequential reasoning, rarely self-awareness, and often fail to use correct reasoning structures🧠

[Eunkyu Eunice Park @uunicee_]

💞 A huge thank you to my amazing collaborators and mentors - @wes_deng, @VaseVarad, Mingxi Yan, @gunheekim, @MaartenSap, @meslami2 - whom I feel incredibly lucky to work with!

[Eunkyu Eunice Park @uunicee_]

🧵Sharing our most-recent work! Critical or Compliant? The Double-Edged Sword of Reasoning in Chain-of-Thought Explanations

[Eunkyu Eunice Park @uunicee_]

[10/10] 📌 Read the paper here: https://t.co/slPNepoPiZ
Looking forward, we hope this work motivates NLP and HCI communities to design explanation strategies and interaction paradigms that actively encourage critical engagement and uncertainty awareness. ❣️

[Wesley Hanwen Deng @wes_deng]

Hello colleagues and friends! ✨Please help repost and share ✨: I am officially on the job market for 𝘁𝗲𝗻𝘂𝗿𝗲 𝘁𝗿𝗮𝗰𝗸 𝗮𝘀𝘀𝗶𝘀𝘁𝗮𝗻𝘁 𝗽𝗿𝗼𝗳𝗲𝘀𝘀𝗼𝗿 and 𝗶𝗻𝗱𝘂𝘀𝘁𝗿𝘆 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗿𝗼𝗹𝗲𝘀 where I can apply my research to real-world AI safety challenges!!

My work combines empirical studies and system building to explore topics such as 𝗿𝗲𝘀𝗽𝗼𝗻𝘀𝗶𝗯𝗹𝗲 𝗔𝗜 (𝗥𝗔𝗜), 𝗔𝗜 𝘀𝗮𝗳𝗲𝘁𝘆, 𝗮𝗻𝗱 𝗵𝘂𝗺𝗮𝗻-𝗔𝗜 𝗰𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝗼𝗻. To this end, I develop tools and processes to support responsible AI in real world industry settings, particularly in the contexts of 𝗔𝗜 𝗮𝘂𝗱𝗶𝘁𝗶𝗻𝗴, 𝗿𝗲𝗱-𝘁𝗲𝗮𝗺𝗶𝗻𝗴, 𝗮𝗻𝗱 𝗶𝗺𝗽𝗮𝗰𝘁 𝗮𝘀𝘀𝗲𝘀𝘀𝗺𝗲𝗻𝘁.

[Wesley Hanwen Deng @wes_deng]

I am beyond trilled to present two 🏆 award-winning papers ✨ @ACM_CSCW in beautiful Bergen, Norway!! Broadly, I'm excited to chat about responsible AI, AI auditing and red-teaming, and human-agent interaction. Say hi if we run into each other!

𝐖𝐞𝐀𝐮𝐝𝐢𝐭: 𝐒𝐜𝐚𝐟𝐟𝐨𝐥𝐝𝐢𝐧𝐠 𝐔𝐬𝐞𝐫 𝐀𝐮𝐝𝐢𝐭𝐨𝐫𝐬 𝐚𝐧𝐝 𝐀𝐈 𝐏𝐫𝐚𝐜𝐭𝐢𝐭𝐢𝐨𝐧𝐞𝐫𝐬 𝐢𝐧 𝐀𝐮𝐝𝐢𝐭𝐢𝐧𝐠 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐯𝐞 𝐀𝐈
🏆 Best Paper Awards (Top 1% Submissions)
🔗 https://t.co/5HvRbMgGWk
📆 Mon, 20 Oct | PM, CET: Online & AI Harms

𝐒𝐨𝐜𝐢𝐞𝐭𝐚𝐥 𝐈𝐦𝐩𝐚𝐜𝐭 𝐀𝐬𝐬𝐞𝐬𝐬𝐦𝐞𝐧𝐭 𝐟𝐨𝐫 𝐈𝐧𝐝𝐮𝐬𝐭𝐫𝐲 𝐂𝐨𝐦𝐩𝐮𝐭𝐢𝐧𝐠 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡𝐞𝐫𝐬
🏅 Best Paper Honorable Mention (Top 3% Submissions)
🔗 https://t.co/1ZfuvuqwBu
📆 Wed, 22 Oct | AM, CET: Toward More Ethical and Transparent Systems and Environments
#CSCW2025 #AISafety #ResponsibleAI

[Maarten Sap (he/him) @MaartenSap]

I’m ✨ super excited and grateful ✨to announce that I'm part of the 2025 class of #PackardFellows (https://t.co/MUl0kGlC3h). The Packard Foundation and this fellowship will allow me to explore exciting research directions towards culturally responsible and safe AI 🌍🌈