@hankhplee
PhD student @CMU HCII | Prev: IBM Research, Microsoft Research, Brave I develop tools that assist AI practitioners in identifying, reasoning about, and mitigating privacy risks during the development of AI products. https://hankhplee.com/
Can LLMs really serve as "crash dummies" for security & privacy testing? We put this assumption to the test.
🚨New preprint 🚨: "How Well Can LLM Agents Simulate End-User Security and Privacy Attitudes and Behaviors?"
👇 THREAD 👇
[Link to paper: arxiv.org/abs/2602.184...
[1/n]
Excited and grateful to share that I’ve been named a Presidential Fellow by @cylab.bsky.social !
Endless thanks to my advisors @sauvik.me and Jodi Forlizzi, and all my collaborators for their constant inspiration as we work toward shaping the future of privacy-preserving AI.
Congrats to my student @hankhplee.bsky.social for being recognized as one of the 2025 CyLab Presidential fellows!
cylab.cmu.edu/news/2025/1...
@cylab.bsky.social wrote a great article on @hankhplee.bsky.social's #chi2025 paper on Purpose Mode.
He presented it today and did a great job!
PurposeMode reduces distraction on social media from 28% -> 7%. You can use it, too! It's open source :)
cylab.cmu.edu/news/2025/04...
Tomorrow (Wed), I will be presenting our @acmtochi.bsky.social paper "Purpose Mode: Reducing Distraction through Toggling Attention Capture Damaging Patterns on Social Media Web Sites" at:
- Wed, 30 Apr | 11:22 AM - 11:34 AM | Room G301
- Session link: programs.sigchi.org/chi/2025/my-...
Tomorrow (Mon), I will be presenting our #CHI2025 paper "The Impact of Generative AI on Critical Thinking" at:
- Mon, 28 Apr | 4:44 PM | Room G314 + G315
- Session link: programs.sigchi.org/chi/2025/pro...
(Reposts appreciated!)
Do you work on user-facing Artificial Intelligence (AI) products? We would like to hear from you!
Please fill out this survey to sign up for the study: cmu.ca1.qualtrics.com/jfe/form/SV_...
@yuxuanli1225.bsky.social has been producing great work since day one of his PhD. Can’t wait to see where his brilliant work leads next!
This work is done during my internship @msftresearch.bsky.social. Huge thanks to my incredible mentors and collaborators: @advaitsarkar.bsky.social @levlevlev.bsky.social Ian, Sean, Richard, and Nick!
We can boost motivation by positioning critical reflection as professional growth—not just extra auditing. Strengthen ability with AI reasoning explanations, guided critiques, and cross-references. Think of GenAI as a provocateur—encouraging workers to refine outputs confidently.
Our findings show GenAI tools should enhance motivators (like quality standards, skill-building) and mitigate inhibitors (time constraints, low awareness) to preserve critical thinking, e.g., proactive prompts can spotlight overlooked tasks, while reactive features offer on-demand assistance.
We observed a shift from “execution” to “stewardship.” GenAI automates tasks like info-gathering or content creation, but workers now invest effort in verification, editing, and alignment with project needs. This pattern emerges widely across different roles and tasks.
GenAI often lowers perceived effort for critical thinking—especially when users trust its capabilities. Interestingly, those confident in their own expertise tend to dig deeper, spending extra time verifying and refining AI outputs.
Their reports are in line with the longstanding balancing of effort at work. In high-stakes tasks requiring high accuracy, they report engaging in more perceived effort, including thinking critically. In low-stakes routine tasks with high time pressure, they report engaging in less perceived effort
Knowledge workers rely on critical thinking with GenAI to maintain work quality. They define it as setting clear goals, refining prompts, and verifying outputs against external sources and their own expertise. They’re driven by the need to avoid errors, improve work quality, and hone skills.
We surveyed 319 knowledge workers from diverse fields, collecting 936 real-world GenAI-assisted tasks. We asked: When is critical thinking necessary? How do they enact it? Does GenAI affect the effort of critical thinking—and how much?
We found that…
We ground our discussion using Bloom’s Taxonomy, focusing on knowledge, comprehension, application, analysis, synthesis, and evaluation. In a GenAI-driven workflow, each of these cognitive activities faces new challenges—and opportunities—for deeper critical engagement.
GenAI is everywhere in the knowledge workflow—from drafting emails to sparking new ideas. But as usage soars, how do we balance its benefits with mindful, reflective work? To design GenAI for better critical thinking, we must first understand how it is currently affecting knowledge work.
How does using GenAI tools reshape knowledge workers’ critical thinking? Our #CHI2025 paper studied 319 knowledge workers to dive into this question. w/@advaitsarkar.bsky.social @levlevlev.bsky.social Ian, Sean, Richard, Nick @msftresearch.bsky.social @hcii.cmu.edu
www.microsoft.com/en-us/resear...
Not at all Ketan! I really appreciated your thoughtful comments on the paper! This is also an amazing opportunity for me to really understand how people feel about our work. I am truly humbled and appreciated!
I'm absolutely thrilled to see my recent CHI paper go viral! I’m humbled by all the support and grateful for the thoughtful critiques.
Critical thinking in the Generative AI era is more important than ever — stay tuned for more insights soon!
Massive thanks to my advisor @sauvik.me, my brave.com internship mentor Philipp, and my amazing collaborators Yi-Shyuan, Lan, and Stephanie! We’re on a mission to challenge the trillion-dollar attention economy — one toggle at a time!
Purpose Mode helps you remove cluttered layouts, notifications, infinite scroll, color saturation, and autoplay. It’s fully open-sourced on GitHub — give it a shot and reclaim your attention:
github.com/hankhplee/pu...
🎥 youtu.be/AWY8HQ_z_-c?...
So what is the bottom line here? Removing these subtle yet powerful design traps through tools like Purpose Mode drastically lowers perceived distraction. It puts you back in control — and makes your social media browsing more purposeful.
Overall, users’ subjective perceptions, such as sense of agency, overshadowed ACDPs as immediate predictors of users’ perceived distraction. Yet when we asked them to reflect on losing focus, they often blamed these dark patterns, suggesting habituation conceals their influence.
In the second week, they were introduced to Purpose Mode, a browser extension we built to allow users to “toggle off” ACDPs. Participants reported feeling distracted only 7% of the time and spent 21 fewer daily minutes browsing these websites.
In the first week, our participants reported feeling distracted in 28% of the EMA questionnaires they answered. Their subjective perceptions of the browsing experience, such as a sense of control and satisfaction, were highly correlated with their perceived distractedness.
We ran a 2-week study with 29 daily users of Facebook, X, LinkedIn, & YouTube — tracking ACDPs (dark patterns) and their perceived distraction via ecological momentary assessment (EMA) questionnaires when browsing these websites. We found that…
Inspired by Monge Roffarello et al’s Attention Capture Damaging Patterns (ACDPs) taxonomy (attentioncapture.com):
RQ1: How do ACDPs impact users' perceived distraction on social media websites?
RQ2: How does Purpose Mode that removes ACDPs affect people's use and experience on these websites?
Social media platforms wage a trillion-dollar battle for your attention, deploying infinite feeds, autoplay, and flashy visual cues that prey on your cognitive biases to keep you hooked. How can we build for a more purposeful and less distracting browsing experience on social media?