←
Cathy Fang271 posts
Cathy Fang avatar

Cathy Fang

@thecatfangs

spaces, embodiment, presence | PhDing @medialab | ex: @MSFTResearch; 🥽, AI/ML @Apple; @magicleap

◷ Joined December 2016▧ 62 Photos and videos
Cathy Fang avatar
Cathy Fang@thecatfangs·21 Mar 2025

Excited to share our large scale longitudinal study done in collaboration with researchers @openai on “affective use” of chatbots like ChatGPT. With AI chatbots becoming more anthropomorphic, how do extended interactions with these chatbots that involve emotional, personal matter impact individuals’ well-being? We looked at how different model and user behaviors shape people’s psychosocial outcomes like loneliness, socialization with people, emotional dependence, and problematic use. This was a team effort of many people on both sides and of many months, and this is just the beginning. We look forward to everyone’s comments and feedback.

Jason Phang@zhansheng21 Mar 2025

🧵I’m excited to share not one but two research papers, written jointly by researchers from OpenAI and the @medialab at MIT. We try to answer the following question: How do interactions with AI chatbots affect people’s social and emotional well-being?

Post attachment
Cathy Fang avatar
Cathy Fang@thecatfangs·Nov 11

me 👭 thimble 👭 claude artifacts everyday go try it out!!

I built thimble, an open source Claude Code plugin that makes it easy to understand what agents did. It adds explainers and interactive visualizations in your terminal. If Claude Code is for building, thimble is for understanding. Install thimble here! github.com/safety-research/t…

0:52
Cathy Fang avatar
Cathy Fang@thecatfangs·Oct 28

timely findings on deference to AI / AI's influence on judgment at the expert level! congrats @amacinho 🐐 who cooked up such a clever study🧑‍🍳

New Anthropic research: How AI agents shape experts' judgment calls When experts delegate work to an agent, do they still make the key decisions? We tested this in a randomized trial with 50 professional data analysts.

0:15
Cathy Fang avatar
Cathy Fang@thecatfangs·Sep 15

I loved going to media art exhibitions as a teen, and to be exhibiting myself felt surreal. Grateful to our team for turning the hackathon side project into something that ran reliably for a week and to the organizing team —especially Jeremiah, Alina, Tom— without whom this would not be possible. Manifesting a tour of “[AI]legory of the Cave”

Cathy Fang avatar
Cathy Fang@thecatfangs·Sep 15

We got Best-in-Show at @ArsElectronica Expanded Conference! Echoing the gallery’s theme of “expanded play”, seeing the surprises and pure joy on people’s face when playing with our piece was the biggest reward. ars.electronica.art/negotiat…

Post attachmentPost attachment

In Plato's Allegory of the Cave, prisoners chained to a wall perceive only shadows—and mistake them for reality. We extend this concept: a shadow man, controlled by an LLM, exists in 2D and perceives only silhouettes. His singular goal is to turn off the light that defines his existence. Users place objects in his path; he kicks them away. The cycle repeats. This mirrors the perceptual limits of AI. An LLM cannot see—it reads descriptions of sight. It processes language about the world without directly accessing the world itself. The shadow man embodies this gap. Both are confined to representations. The piece is a meditation on AI as cave-dweller—processing patterns without grasping the higher-dimensional reality that casts them.

0:49
Shayne Longpre avatar
⟳ Cathy Fang retweeted
Shayne Longpre@ShayneRedford·Aug 18

7/ Huge thanks to my co-leads @AnkaReuel + @zoeykii and our incredible collaborators @thecatfangs @megan_richards_ @zhipinghci @chuanyang_jin Jenn Mickel, Cedric Whitney, @ahmetustun89 @niloofar_mire @OjewaleV @ArielNLee @alex_pentland @tianshi_li @yuntiandeng Mykel Kochenderfer @sarahookr and @sanmikoyejo. And thank you to @nitashatiku @washingtonpost and @eileenguo @techreview for covering the work. Their pieces are well worth reading. washingtonpost.com/business/… technologyreview.com/2026/08…

Cathy Fang avatar
Cathy Fang@thecatfangs·Aug 18

Do some AI models refuse more user requests than others? Are people using AI to generate illicit content more or less over time? And which models do people go to for coding vs role-play? Unfortunately, existing reports of AI use from frontier labs are coarse and proprietary that do not help us understand the impacts and natural uses of AI. This prompted us to create a public AI observatory ai-observatory.org/

1/ How are people really using AI? Today, researchers across @MIT, @Stanford, + 12 academic institutions are launching the Public AI Observatory: public infrastructure for measuring how people actually use AI assistants in the wild. 🔗 ai-observatory.org/ 📜 dataprovenance.org/ai_observ…

Post attachmentPost attachmentPost attachment
Cathy Fang avatar
Cathy Fang@thecatfangs·Jul 10

Prior interventions to reduce AI overreliance often caused frustration and were less preferred. In contrast, our Priori interface was more preferred, a sign that the interface made exercising judgment easy and aligned with people’s incentive.

Post attachment
Cathy Fang avatar
Cathy Fang@thecatfangs·Jul 10

Compared to the plain chat interface, both intervention interfaces gave the user a greater sense of control over the model and felt more transparent while requiring no additional effort to judge the model output.

Post attachment
Cathy Fang avatar
Cathy Fang@thecatfangs·Jul 10

We ran a pre-registered study (n=160) where people were asked to discuss controversial topics using both a standard chat interface and one of the two intervention interfaces in random order: (a) Only seeing model reasoning (b) Priori, which allows seeing and changing models’ choices

Post attachment