Lawted Wu

Updated May 24, 2026

I teach the AI for Social Sciences track at Generation AI 2026 — the joint AI Maker program of Stanford Center at Peking University.

AI is finally cheap enough, and tools are finally good enough, that a teenager can ship a game a stranger actually wants to play. This course is built around that fact.

In eight weeks, students build a playable, AI-powered interactive narrative game about one social issue that quietly shapes Chinese teenagers — academic pressure, family relationships, or the online environment they grow up in.

The reference project is PUA Game — a black-comedy survival game I built that simulates a graduate student under an abusive advisor. Each student ships something in the same spirit, on a real public URL, with real players.

No prior code experience is required. You vibe-code with Claude Code and Codex, both running in the terminal. You describe what you want; the AI agents pair with you to scaffold a Next.js + Tailwind + Supabase app and ship it to Vercel.

Outside the course, I'm a research assistant at Stanford HAI, a research intern at Tsinghua HCI Lab (co-author of an ACM CHI 2026 paper), and incoming Harvard MDE '28. Before that, three years of frontend at Alibaba, MiniMax, and Tencent.

You can reach me at Email, lawted.tech, CV, GitHub, or X.

Background

- 2026—28 Harvard University · MDE Master in Design Engineering. Incoming '28 cohort. Incoming
- 2025—now Stanford HAI · Cohen & Sahami Lab Research assistant on HCI × AI-supported academic motivation. Stanford
- 2025—26 Tsinghua HCI Lab Co-author of an ACM CHI 2026 paper on proactive AI for collaborative writing. CHI '26
- 2022—25 Alibaba · Lingyang Frontend engineer on the enterprise data platform. SQL editor, design system. 3 yrs
- 2022 MiniMax · Glow Initial frontend for Glow, one of China's earliest consumer LLM products. Intern
- 2021—22 Tencent · PCG Search Internal platforms for AI & algorithm teams. Intern
- 2019—23 Queen Mary University of London · BUPT BSc (Engineering), Telecommunications Engineering & Management. BSc

Method

- 01 Reproduce, then deviate Start by rebuilding PUA Game piece by piece. Then break the pattern. The fastest path to taste. Week 2
- 02 Ship every week Public URL by Week 2. Each week ends with something you can show a stranger. Velocity beats ambition. Weekly
- 03 Defend every choice Don't just make it — explain why this NPC, this branch, this ending. The same skill carries through the oral defense. Always
- 04 Players, not graders The work is judged by whether real testers feel something playing it. Not by whether I like the slides. Week 7

Ideas

- Academic The Gaokao Countdown A time-pressured RPG where every choice leads to a slightly worse outcome. The player learns there was never a winning move. Idea
- Family Family Dinner A conversation game where you can never say what you actually mean. The NPC parents react to what they hear, not what you said. Idea
- Online The Feed An infinite scroll that always knows what to show you next. The exit button is real, but never visible. Idea
- Yours Bring your own The strongest games come from a student with a personal scar around one of the three issues. Pitch me yours. Idea

Outcomes

- Game A playable live game on a real URL Strangers actually play it. Not a slideshow, not a Figma file, not a "demo on my laptop." Aug
- Paper A 3000–5000 word paper, APA style Defended at the Stanford × Peking University academic conference. The kind of artifact admissions readers actually remember. Aug 2
- Study A user-testing report on your own work Real testers sit down. You watch. You take notes. This is the social-sciences method, scaled to one classroom. Jul
- Skill Fluency in Claude Code, Codex, and the modern web stack Skills that compound. After this, you can ship anything you can describe in language. Forever

Background

Method

Ideas

Outcomes

Elsewhere