Pinned Tweet
Most of what I actually need help with, I never think to tell a model. But why is it on me to remember?
Our new paper asks: what if AI could proactively specialize to individuals and the tasks they’re carrying out at this very moment? 🧵
Our new paper asks: what if AI could proactively specialize to individuals and the tasks they’re carrying out at this very moment? 🧵
13
43
280
51,097
Say hello to us at UIST in Detroit!! See you at the Tues afternoon coffee break to chat about research, PhD apps, and maybe weigh in on my lab name 👀
Interested in a PhD in HCI at UT Austin CS? Meet us at #UIST2026!
Join Karthik Mahadevan (@karthikm0), Michelle Lam (@michelle123lam), and me to chat about research and ask questions about PhD applications. Our interests span a range of technical HCI areas (see the flyer).
KGFafAw
🗓️: Tuesday, November 3, from 3:30 PM (during the afternoon coffee break)
📍: Renaissance Foyer East, Renaissance Center
Looking forward to meeting you!
1
4
35
1,737
Michelle Lam retweeted
I’m recruiting 2~3 funded PhD students for Fall 2027 to join the Symbiotic Interfaces Lab at UT Austin CS (@UTCompSci)!
Check out our research:
symbiotic-interfaces.cs.utexas.edu
Apply by Dec 15, mentioning my name:
cs.utexas.edu/graduate/apply
Reposts appreciated!
(1/2)
Check out our research:
symbiotic-interfaces.cs.utexas.edu
Apply by Dec 15, mentioning my name:
cs.utexas.edu/graduate/apply
Reposts appreciated!
(1/2)
4
64
297
30,630
Michelle Lam retweeted
🚨🚨🚨UIST Workshop Announcement: The Personalized Computer for the 21st Century
Join us in Detroit on November 2nd for our morning workshop! 💻🤖☀️
personalized.computer/uist-26/
Join us in Detroit on November 2nd for our morning workshop! 💻🤖☀️
personalized.computer/uist-26/
1
14
45
7,473
After over 10 years at Stanford, it's time to leave :)
I will be joining UT Austin's CS department as an assistant professor in fall 2027! If you're excited to envision the future of interaction with AI, I'm recruiting PhD students this cycle. Come join me!
I will be joining UT Austin's CS department as an assistant professor in fall 2027! If you're excited to envision the future of interaction with AI, I'm recruiting PhD students this cycle. Come join me!
103
90
1,271
102,517
If you're interested, I'd love to get to know you: forms.gle/r3VNjpPHgXfMUH…
Thank you to my advisors @msbernst & @landay, without whom I would certainly not be on this path of HCI research. Thank you @StanfordHCI, my home for all these years. You all make it hard to leave!
Thank you to my advisors @msbernst & @landay, without whom I would certainly not be on this path of HCI research. Thank you @StanfordHCI, my home for all these years. You all make it hard to leave!
3
2
61
4,890
Michelle Lam retweeted
People are increasingly worried that AI tools make us overreliant.
But how do we actually measure this? We introduce Offloading Score, a measure of reliance based on the fraction of cognitive effort offloaded to AI while completing a task.
In a controlled user study, Offloading reliance under time pressure, while several common alternatives do not.
(1/9)
But how do we actually measure this? We introduce Offloading Score, a measure of reliance based on the fraction of cognitive effort offloaded to AI while completing a task.
In a controlled user study, Offloading reliance under time pressure, while several common alternatives do not.
(1/9)
7
77
230
94,898
Michelle Lam retweeted
We upgraded Tabracadabra 🎉 to bring an entire context-aware assistant (not just tab to autocomplete!) to any textbox. It's pretty great if you hate switching between the chat interface and what you're working on. We're also open-sourcing, so you can try it out!🧵
13
36
189
43,353
Michelle Lam retweeted
This is one of my fav figures in our paper. You can:
1. Identify a user's objective by observing general interaction with their computer.
2. Use it to construct a "just in time" rubric.
3. Sample bunch from model and SCALE TEST TIME COMPUTE ON LITERALLY ANY OPEN-ENDED TASK?!?
1. Identify a user's objective by observing general interaction with their computer.
2. Use it to construct a "just in time" rubric.
3. Sample bunch from model and SCALE TEST TIME COMPUTE ON LITERALLY ANY OPEN-ENDED TASK?!?
Once you have JIT objectives, you can embed them into various LLM architectures via existing generators and evaluators. Evaluations on N=205 participant-provided inputs show that JIT objectives produce user-preferred outputs, whether generating experts, tools, or feedback.
0
6
37
10,270
In hour-long in-lab sessions where users applied the system to their own writing tasks (N=17), Poppins produced specialized outputs that are unique to each participant and are rated as significantly higher quality than that of a standard LLM chat tool.
1
0
12
1,521
Thanks so much to co-authors @oshaikh13, Hallie Xu, @azguo2, @Diyi_Yang, @jeffrey_heer, @landay, & @msbernst! We’re grateful that this work received a CHI best paper honorable mention!
paper: arxiv.org/abs/2510.14591
waitlist: forms.gle/rVzWo5asJ2Xv6Q…
talk: Th 4/16 12pm @ P1-111
paper: arxiv.org/abs/2510.14591
waitlist: forms.gle/rVzWo5asJ2Xv6Q…
talk: Th 4/16 12pm @ P1-111
0
3
40
3,025
Michelle Lam retweeted
What’s the point of a “helpful assistant” if you have to always tell it what to do next?
In a new paper, we introduce a reasoning model that predicts what you’ll do next over long contexts (LongNAP 💤).
We trained it on 1,800 hours of computer use from 20 users.
🧵
In a new paper, we introduce a reasoning model that predicts what you’ll do next over long contexts (LongNAP 💤).
We trained it on 1,800 hours of computer use from 20 users.
🧵
17
83
323
126,976
Michelle Lam retweeted
Is the only way we can create algorithms that people understand to make them trivially simple? We argue, no.
People can predict the behavior of algorithms that are arbitrarily complex, if and only if they are available, compact and aligned.
arxiv.org/abs/2601.18966
People can predict the behavior of algorithms that are arbitrarily complex, if and only if they are available, compact and aligned.
arxiv.org/abs/2601.18966
1
17
56
14,611
RT @KaitlynZhou: As of June 2025, 66% of Americans have never used ChatGPT.
Our new position paper, Attention to Non-Adopters, explores…
Our new position paper, Attention to Non-Adopters, explores…
0
38
0
1
Michelle Lam retweeted
🚨 70 million US workers are about to face their biggest workplace transmission due to AI agents. But nobody asks them what they want.
While AI races to automate everything, we took a different approach: auditing what workers want vs. what AI can do across the US workforce.🧵
While AI races to automate everything, we took a different approach: auditing what workers want vs. what AI can do across the US workforce.🧵
14
133
657
119,577