ABCDEFGHIJ
1
Anyone is welcome to get started on any of these projects! Please demonstrate your own intiative, motivation, and skills.
Have a project idea? Submit it here: https://forms.gle/hmE5581EntQ8dJrm7
2
StatusProject NameBrief DescriptionDisciplines3 Steps to start withFaculty MentorIndustry MentorStudentsRecommended Readings/MaterialsRepository (AI Kitchen org members only)
3
CompletedFocus Mode: AI-Powered Distraction BlockerBuild a Chrome extension that uses AI to help users stay focused, detecting when browsing drifts off-task and intervening in non-annoying ways. Revive and extend the ideas from Stanford's HabitLab project.CS, Psychology, HCI, Behavioral Science1. Define distraction scenarios (5–10 real cases)
2. Build Chrome extension (basic detection)
3. Add AI nudges + test behavior change		Kai LukoffMisa Truong, Espy Aguilar, Ian Tan, Anh Pham- Kovacs, G. et al. (2018). HabitLab: Designing for Healthier Social Media Use. https://habitlab.github.io/
- Lyngs, U. et al. (2019). Self-Control in Cyberspace: Applying Dual Systems Theory to a Review of Digital Self-Control Tools. CHI '19.
- Roffarello, A. M. & De Russis, L. (2023). Achieving Digital Wellbeing Through Digital Self-Control Tools. ACM Computing Surveys.
4
CompletedCampusVal: AI Evaluation for University AdvisingDesign evaluation frameworks to test how well AI tools handle university advising tasks (course planning, prerequisite checks, policy questions). Build a benchmark dataset of real advising scenarios and measure where AI helps vs. misleads.CS, Education, Psychology, Any major1. Create dataset of advising scenarios
2. Test multiple AI tools
3. Build evaluation metrics (accuracy, risk, usefulness)		Kai LukoffShikhar Sisodia- Raji, I. D. et al. (2021). AI and the Everything in the Whole Wide World Benchmark. NeurIPS '21.
- Liao, Q. V. & Sundar, S. S. (2022). Designing for Responsible Trust in AI. CHI '22 Workshop.
- Ribeiro, M. T. et al. (2020). Beyond Accuracy: Behavioral Testing of NLP Models with CheckList. ACL '20.
5
IdeaAI for Parallel PrototypingMost AI development tools support serial prototyping: generate one thing, iterate on it. Explore how AI tools could instead support parallel prototyping, generating and comparing multiple design alternatives simultaneously, which research shows leads to better outcomes.CS, Design, HCI1. Choose design problem
2. Generate multiple AI prototypes
3. Compare outcomes + document insights		Kai Lukoff- Dow, S. P. et al. (2010). Parallel Prototyping Leads to Better Design Results, More Divergence, and Increased Self-Efficacy. TOCHI.
- Dow, S. P. et al. (2011). Prototyping Dynamics: Sharing Multiple Designs Improves Exploration, Group Rapport, and Results. CHI '11.
- Buxton, B. (2007). Sketching User Experiences. Morgan Kaufmann.
6
In ProgressAI Coursework Planner for SCU StudentsBuild an AI agent skill that helps SCU students plan their coursework across quarters, accounting for prerequisites, scheduling conflicts, major requirements, and personal preferences. Test whether AI-assisted planning leads to better outcomes than existing tools.CS, Education, HCI1. Map course requirements + constraints
2. Build planning logic
3. Test with real student scenarios		ACM is already working on this concept via http://scuschedule.com - reach out to them to get involved!- Chen, Q. et al. (2023). Large Language Models as Course Planning Assistants. LAK '23.
- Pardos, Z. A. & Jiang, W. (2020). Designing for Serendipity in a University Course Recommendation System. RecSys '20.
- SCU Course Catalog and Academic Requirements. https://www.scu.edu/bulletin/
7
In ProgressJoyful Workflow Automation (Campus Staff)Partner with SCU staff to identify repetitive tasks and build AI-powered automations that save time and reduce frustration. Focus on workflows that are tedious but don't require deep expertise.CS, Business, Any major (domain knowledge)1. Identify repetitive workflows
2. Build simple AI automation
3. Measure time saved + efficiency		Sandy Cai, Bridget Hestad, Nicholas Shen, Junfan Zhu - Amershi, S. et al. (2019). Guidelines for Human-AI Interaction. CHI '19.
- Yang, Q. et al. (2020). Re-examining Whether, Why, and How Human-AI Interaction Is Uniquely Difficult to Design. CHI '20.
- Zapier. (2024). The state of business automation. https://zapier.com/blog/state-of-business-automation/
8
IdeaAmbient Language LearningDesign an AI-powered tool that embeds language learning into daily digital activities (email, browsing, messaging) rather than requiring dedicated study sessions. Explore how ambient exposure can supplement formal learning.CS, Linguistics, Education, HCI1. Identify daily touchpoints (email, browsing)
2. Embed learning interventions
3. Track engagement + retention		- Duolingo Research. (2023). https://research.duolingo.com/
- Trusty, A. & Truong, K. N. (2011). Augmenting the Web for Second Language Vocabulary Learning. CHI '11.
- Dearman, D. & Truong, K. (2012). Evaluating the Implicit Acquisition of Second Language Vocabulary Using a Live Wallpaper. CHI '12.
9
In ProgressAnnotated Repositories for Research through BuildingDevelop and test a method for turning GitHub repos into "annotated portfolios"‚ version-controlled codebases enriched with structured reflections that document what builders learn from deploying and maintaining real software. The deliverable is a shareable ANNOTATED-REPO.md skill file (compatible with Claude Code, Cursor, Copilot, etc.) that instructs an AI coding assistant to prompt builders for reflection at key moments, capture insights in a consistent format, and synthesize findings over time. Student teams will use the method on their own AI Kitchen Slow Roast projects, generating a corpus of annotated repos. The research question: what knowledge emerges from building, deploying, and iterating on real interactive systems that could not have been learned from design prototypes alone? Builds on Research through Design (Zimmerman et al., 2007) and annotated portfolios (Gaver & Bowers, 2012), extending them into the era of AI-assisted software development.HCI, CS, Software Engineering, Design, Education, Any major (as builders)1. Use annotation framework in repo
2. Capture reflections during builds
3. Synthesize learnings over time		Kai Lukoff- Zimmerman, J., Forlizzi, J., & Evenson, S. (2007). Research through Design as a Method for Interaction Design Research in HCI. CHI '07.
- Gaver, B. & Bowers, J. (2012). Annotated Portfolios. Interactions 19(4), 40-49.
- Gaver, W. (2012). What Should We Expect from Research through Design? CHI '12.
- Koskinen, I. et al. (2011). Design Research Through Practice: From the Lab, Field, and Showroom. Morgan Kaufmann.
- Sarkar, A. & Drosos, I. (2025). Vibe coding definitions and workflows.
- Hudson, S.E. & Mankoff, J. (2014). Concepts, Values, and Methods for Technical HCI Research. In Ways of Knowing in HCI.		https://github.com/The-AI-Kitchen/slow-roast-template/blob/main/ANNOTATED-REPO.md
10
In ProgressGazeReader: Gaze-Aware AI Reading AssistantA gaze-aware AI reading assistant that uses webcam-based eye tracking to detect which passage a reader is focused on, then provides contextual Q&A powered by Claude. Explores whether gaze can serve as a lightweight, implicit input for AI tools.HCI, CS, Computer Vision1. Implement gaze tracking
2. Map gaze to text
3. Add AI Q&A + test usability		Kai Lukoffhttps://github.com/The-AI-Kitchen/gaze-reader
11
In ProgressAI Kitchen Personalized Onboarding A two-part personalized onboarding system for AI Kitchen. (1) An AI chatbot embedded on the AI Kitchen landing page where visitors describe their background and interests, and the LLM suggests creative ways they could use AI tailored to their profile. Designed to make AI feel accessible and personally relevant to non-technical users. (2) A personalized onboarding email pipeline: when someone fills out the Get Involved interest form (with an optional LinkedIn URL for richer context), an LLM generates a custom welcome email suggesting specific AI Kitchen opportunities matched to their background, including Taste Test sessions, Cook-off topics, and Slow Roast projects they might join. Replaces generic welcome emails with something that makes people feel seen and gives them concrete next steps.CS, HCI, Communication, Business, Any major1. Build chatbot for interest capture
2. Create personalization logic
3. Generate tailored onboarding emails		Potential: Cameron Behar- Amershi, S. et al. (2019). Guidelines for Human-AI Interaction. CHI '19.
- Kocielnik, R. et al. (2019). Will You Accept an Imperfect AI? Exploring Designs for Adjusting End-User Expectations of AI Systems. CHI '19.
- Liao, Q. V. et al. (2020). What Can You Do? Studying Social-Agent Orientation and Agent Proactive Interactions with an Agent for Employees. DIS '20.
12
Ideaknow-me.md: Structured LLM Memory for PersonalizationA between-subjects deployment study comparing two approaches to LLM memory management. Control group receives a single generic memory instruction; treatment group receives a structured, hierarchical system: a top-level know-me.md (high-level preferences and context), domain-specific subdirectory files (e.g., family.md, work.md), and a privacy exclusion mechanism. Both groups use Claude Pro for ~4 weeks. Weekly surveys measure perceived model knowledge, usefulness, and personalization. End-of-study: participants prompt the model to share what it knows, anonymize, and review; plus a semi-structured exit interview probing mental models of LLM memory and privacy boundaries. RQs: Does structured memory lead to a more accurate user model? Does a richer user model improve perceived usefulness and personalization? How do users reason about what an LLM knows about them?HCI, Psychology, CS, Communication1. Design structured memory system
2. Run controlled experiment
3. Analyze feedback + results		Kai Lukoff- Amershi, S. et al. (2019). Guidelines for Human-AI Interaction. CHI '19.
- Liao, Q. V. & Sundar, S. S. (2022). Designing for Responsible Trust in AI. CHI '22 Workshop.
- Anthropic. (2025). Claude Memory Documentation. https://docs.anthropic.com/
- OpenAI. (2024). Memory and New Controls for ChatGPT. https://openai.com/index/memory-and-new-controls-for-chatgpt/
- Karpathy's LLMwiki (may make this idea obsolete!): https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47