A working list of the tools, readings and people worth your attention this semester. Tool links change often; readings age better. The full annotated collection lives in the course resource database.
Tools
Language Models & Research
- ChatGPT — General-purpose assistant; the baseline to compare the others against.
- Claude — Strong on long documents, and builds working code and interfaces in the conversation.
- Google Gemini — Tied into Google Workspace and Google’s own image and video models.
- Perplexity — Search-first, answers with linked citations. Use it to verify claims.
- NotebookLM — Answers only from sources you upload, citing back to the document.
- Hugging Face — Open models, datasets, and browser demos. The model cards are worth reading.
Image Generation & Editing
- Midjourney — The strongest house style of the major tools — study what it does uninvited.
- Adobe Firefly in Photoshop — Generative Fill and Expand, trained on licensed and public-domain material.
- Reve — Built around editing rather than one-shot generation — live layers, direct manipulation.
- Krea Realtime Canvas — Draw rough shapes and watch them resolve live. Prompting that feels like sketching.
- Magnific — Sketch to Image — Real-time sketch-to-image; sketches are not used to train their models.
- Stable Diffusion — Open-weight models you run yourself — the foundation under many other tools.
- Figma Weave — Node-based workflows chaining models and edits — Comfy’s idea without the local install.
- Comfy (ComfyUI) — Node-based workflows. The only tool here that makes the pipeline visible.
- Replicate — Run thousands of open-source models from your own code. Pay per second of compute.
- Ollama — Run open language models on your own machine, entirely offline. Free and open source.
Video & Motion
- Runway — Video, image and audio generation and editing in one workspace.
- Higgsfield — Video generation built around camera control — the platform the two camera references above feed into.
- LTX Studio — Script to storyboard to shots — forces pre-production decisions before you generate.
- Google Flow — Shot planning, generation and natural-language editing, running Veo.
- Autodesk Flow Studio — Live-action footage into editable CG scenes; markerless motion capture.
- AI Camera Movements — 46 camera-movement prompts with video previews. Free reference.
Audio, Voice & Music
- Suno — A finished song from a text description. The sharpest case study on training data.
- ElevenLabs — Reference-quality text-to-speech, voice cloning and cross-language dubbing.
- Adobe Podcast — Enhance Speech — Makes a bad recording sound like a studio. Repair rather than invention.
Synthetic Humans & Likeness
- Synthesia — Corporate video from AI avatars — actors who licensed their likeness.
- HeyGen — Avatar video, plus the translation and lip-sync work it is known for.
Design, Interfaces & Code
- Figma Make — Prompt from a Figma design to a working prototype, on your own libraries.
- Claude Design — Prototypes, decks and one-pagers you can edit directly; imports your design system. Beta, paid plans only.
- Uizard — UI from a prompt, a screenshot, or a photo of a hand-drawn wireframe.
- ml5.js — Machine learning for the browser, built for p5.js. AI as a component, not a prompt box.
- Impeccable — Design-systems tooling for AI coding environments. Addresses drift across screens.
- Lovable — Describe an app in plain language and get working software — database, auth and hosting included.
Prompting
- Prompt Engineering Guide — The most complete open reference on technique. Maintained and current.
- OpenAI Prompt Engineering Guide — Short and concrete — but tuned to OpenAI’s models specifically.
Provenance, Rights & Verification
- Content Credentials (C2PA) — The industry standard for attaching provenance to media. Includes a verify tool.
- Fairly Trained — Certifies AI companies whose training data was properly licensed.
- EU Artificial Intelligence Act — Explorer — Searchable text, plain-language summaries, and the implementation timeline.
- This Image Does Not Exist — Two-minute game: human-made or machine-made? Most people score worse than expected.
- PhotoGuard — MIT research on immunizing photos against AI editing — and why it doesn’t hold.
Readings
Bias & Ethics
- Excavating AI: The Politics of Images in Machine Learning Training Sets — Crawford & Paglen, 2019. Training sets are political acts, not technical artifacts.
- Humans Are Biased. Generative AI Is Even Worse. — Bloomberg, 2023. Thousands of images measured; skew well past real-world statistics.
- Google explains Gemini’s AI pictures of diverse Nazis — The Verge, 2024. A fix for one bias producing an obviously wrong result in another direction.
- OpenAI’s GPT Is a Recruiter’s Dream Tool. Tests Show There’s Racial Bias — Bloomberg, 2024. Identical résumés, different names, consistently different rankings.
- Israel used AI to identify bombing targets in Gaza — The Verge, 2024. Ranked lists and a human clicking approve, at lethal scale.
Labor, Energy & the Future of Design
- Design in Tech Report — John Maeda. Eleven years of annual reports on design and technology; the 2026 edition argues the job is shifting from UX to AX, and every past report is downloadable.
- Why A.I. Isn’t Going to Make Art — Ted Chiang, The New Yorker, 2024. Art is thousands of small choices; a prompt supplies too few.
- After Automation: Why AI progress creates more work for humans — Dan Shipper, Every, 2026. The opposite case to Chiang. Read them together.
- Does Amazon’s Just Walk Out rely on 1,000 workers in India? — USA Today, 2024. The interface was built so you wouldn’t wonder.
- Berlin Debates AI Threat to Hollywood Jobs — The Hollywood Reporter, 2024. “Fear the person who uses these tools” — worth interrogating.
- How much electricity does AI consume? — The Verge, 2024. The per-prompt cost, and why it’s hard to pin down.
- Google’s emissions climb nearly 50% in 5 years due to AI — The Guardian, 2024. From Google’s own environmental disclosure.
Misinformation & Verification
- Does AI Fight or Fuel Election Disinformation? — Brennan Center, 2026. Half of all chatbot responses contained an inaccuracy.
- How a Small Iowa Newspaper’s Website Became an AI-Generated Clickbait Factory — WIRED, 2024. Every credibility signal still pointed at something that no longer existed.
Design Practice
- Incorporating AI into design — a framework of considerations — Tom Castle. Design for people, transparent value, failure, learning, ethics.
- Avian Visitors — Teddy Warner. AI as one material inside a designed system rather than the whole point.
- Aesthetics Imperfections — Gianpaolo Tucci, Slanted Publishers. Experimental typography made with AI.
Follow & Keep Up
- Harper Carroll (@harpercarrollai) — How models actually work, explained by an AI engineer. Instagram.
- Jeremy Carrasco (@jeremyfindsai) — How to spot synthetic media — detection rather than production. Instagram.
- Evolving AI (@evolving.ai) — A large community feed of new tools, techniques and generated work. Good for scanning what’s arriving week to week. Instagram.
- enigmatic_e (@enigmatic_e) — AI video and animation workflows, start to finish. Instagram.
- The Neuron — Daily AI newsletter, skimmable in two minutes.
- That AI Collection — Hand-checked directory of thousands of AI tools, updated weekly.
- awesome-generative-ai — Curated list of tools, papers and courses — with a section on critical perspectives.
More
This page is a selection. The full database — every resource, with tags, reading time and notes on what to be skeptical about — is here.