Jackie Yang

Co-founder and VP of Software and ML @ Skywalk Inc.
CS PhD @ Stanford University

[Jackie Yang @jackieyang_]

OpenAI has different thinking strength scales across Web chat mode, Web work mode, iOS work mode, and Mac Codex mode. I'm confused.

Is Pro a different mode?
Ultra vs Max, which one is higher?

[Jackie Yang @jackieyang_]

A year in the making: Subtle Voicebuds.
Near-perfect wearer speech isolation means only your voice gets heard.

Clear calls in loud environments, plus AI chat/dictation at a volume that’s invisible to your surroundings.

Live demo@CES Venetian 60852.

[Subtle @subtlecomputing]

Introducing the first subtle computing interface, Voicebuds.

A device that hears only you, even at your quietest. Available for preorder now.

[Jackie Yang @jackieyang_]

@BLKMDL3 How about no turn on red signs? Is FSD v14 observing those better than v13?

[Jackie Yang @jackieyang_]

Programmers these days: "Talk is cheap. Show me the prompt (to Cursor/Claude Code)."

[Tianshi Li @tianshi_li]

🚨New workshop at @acm_ccs #CCS2025!

Glad to announce the 1st workshop on Human-Centered AI Privacy and Security
(#HAIPS2025, pronounced “hypes” 😎)

AI is reshaping lives and S&P research—let’s bring people back to the center.

📢 CFP is out!

🌊 https://t.co/z3zgSi2mqF

[Jackie Yang @jackieyang_]

See our talk at the next session at #CHI2025! Wed, Apr 30, 4:20 PM - 4:32 PM in G416 + G417. First talk in the "Shaping Cognitive Processes". (I know, the paper, as usual, is not related to the session name 😉)

[Jackie Yang @jackieyang_]

🧞‍♂️Multimodal apps are flexible but hard to build - how can we discover needed features early?

✨Excited to share our #CHI2025 paper: "GenieWizard: Multimodal App Feature Discovery with #LLM"!

Come see my talk: Wed, Apr 30, 4:20 PM in G416+G417. Session: https://t.co/CLLUyXRBg2

[Jackie Yang @jackieyang_]

Huge thanks to my amazing co-authors and advisors for making this work possible: Yingtian Shi, Chris Gu, Zhang Zheng, Anisha Jain, @tianshi_li, @MonicaSLam, and @landay! 🙏

[Jackie Yang @jackieyang_]

🧞‍♂️Multimodal apps are flexible but hard to build - how can we discover needed features early?

✨Excited to share our #CHI2025 paper: "GenieWizard: Multimodal App Feature Discovery with #LLM"!

Come see my talk: Wed, Apr 30, 4:20 PM in G416+G417. Session: https://t.co/CLLUyXRBg2

[Jackie Yang @jackieyang_]

In our user study (N=12), devs using GenieWizard implemented 42% of missing features vs. only 10% without it! They also found it more usable & less demanding. Boost #multimodal app dev! 🚀 #CHI2025

[Tianshi Li @tianshi_li]

Chatting with LLMs like #ChatGPT is so effortless—we often overshare without even realizing it.

Our #chi2025 paper presents Rescriber, a smaller-LLM-powered tool to help users regain control over #privacy when still getting useful responses.

Preprint: https://t.co/IakZNcRaBX

[Jackie Yang @jackieyang_]

From the UI framework level, ReactGenie can automatically generate an object-oriented interface for AI to perform function chaining within the app's UI state to achieve complex tasks users request through multimodal interactions effeciently.

[Jackie Yang @jackieyang_]

In addition to improving AI's ability to interact with apps (Anthropic computer use), apps should also improve in providing an interface for AI models. My previous research, ReactGenie (https://t.co/YFRocv7A5l), provides a toolkit for apps optimized for both humans and AI.

[Jackie Yang @jackieyang_]

Meet with @plkumjorn at @ACMUIST #UISTChallenge

[Tianshi Li @tianshi_li]

I’m planning to hire 2 PhD students in Fall25 at @KhouryCollege!

How do we make privacy easier for people?
How do we address people’s real privacy needs objectively and subjectively?
How does AI play a role in introducing and mitigating risks?

Talk to me at #UIST2024 #HCOMP2024

[Jackie Yang @jackieyang_]

Prompt: Read these webpages and summarize the price max weight/max min height/max size of the monitor supported. If they mention the main stem radius, please list it; if not, list N/A. Spend more time reading each webpage. Think step by step.
[Links to products]

[Jackie Yang @jackieyang_]

I asked a few LLM services to read a few Amazon product web pages and create a table for comparison, but they all failed miserably. I thought this task would be quite easy for LLMs. Any recommendation for an alternative LLM/AI service that would work with this kind of request?

[Jackie Yang @jackieyang_]

I have tried ChatGPT (4o), Bing Copilot (Pro), Gemini (Advanced), Amazon Rufus, and Perplexity (pro query). They all failed miserably. I quickly checked https://t.co/DSpiICwZk8’s robots.txt. It blocked ChatGPT but no other services, so it shouldn’t explain the bad performance.