Seastone

Article

Where to learn more about AI, in three levels

These are the resources I give to AI at Work participants who want more: where to start, and where to dig deeper.

  • Tobias Sjösten
  • October 2026
  • 3 min

You don’t need to follow AI closely to get good use out of it at work. Some want to anyway, and these are the next steps I point them to. I put the list together for participants in AI at Work, but it should be useful to anyone who wants to understand AI better.

These resources come in three levels. Start at the top and stop when you’ve had enough. Everything is free unless I note otherwise, and in English unless I mark it as Swedish.

For everyone

  • Anthropic Academy: free short courses from the company behind Claude. Start with AI Fluency: Framework and foundations, which is written for people who don’t write code. It takes about four hours.
  • OpenAI Academy: the same idea from the company behind ChatGPT. Its AI Foundations course takes about an hour, and you need a ChatGPT account.
  • One Useful Thing: Ethan Mollick’s newsletter about AI in everyday work, written for people who aren’t technical. A new post every few weeks.
  • The course community, if you’ve taken AI at Work. I share what I see there that’s worth knowing.

For the curious

  • Release notes for Claude and ChatGPT: what changed in the app and when. Follow the one for the tool you use.
  • Anthropic’s prompting guide: written for developers, but the principles hold. It’s the source of one of the course’s golden rules — if a colleague with no context would be confused by your prompt, so would the AI. OpenAI has a guide for ChatGPT users too.
  • Generative AI in a Nutshell: Henrik Kniberg explains the whole picture in an 18-minute video.
  • Grundkurs i AI: short articles from Internetstiftelsen on what AI is and how ChatGPT works, 6 to 14 minutes each. In Swedish.
  • Elements of AI: a free basic course from the University of Helsinki, in Swedish. Plan for 15 to 30 hours. It covers AI and machine learning in general and predates ChatGPT, so read Grundkurs i AI for language models.
  • AI literacy under the EU AI Act: since February 2025, an organization that uses AI has to take measures to make sure its staff know enough about it. The European Commission’s questions and answers explain what that means. There’s a machine-translated Swedish version.
  • Att tänka på när du använder AI-tjänster: the Swedish data protection authority, IMY, on what to keep out of an AI service. In Swedish.

For the very interested

  • Simon Willison’s blog: the sharpest day-to-day analysis of new models and tools, with some technical depth. He posts daily.
  • Intro to Large Language Models: an hour-long talk by Andrej Karpathy from 2023. The step after Kniberg’s video, if you want to know how a model is built and trained.
  • 3Blue1Brown on neural networks: for anyone who wants to see what’s going on inside a model. Chapters 5 to 7 are about the transformers behind language models.
  • AI as Normal Technology: Arvind Narayanan and Sayash Kapoor of Princeton on what AI can and can’t do. A careful, skeptical counterweight to everything else on this page. Their book AI Snake Oil (2024, paid) makes the case at length.
  • Epoch AI: data on how fast the models are developing. The course’s chart comes from here.
  • Anthropic Economic Index: what people actually use Claude for at work, by task and occupation.
  • The Stanford AI Index: the annual overview of where AI stands. Long, but you can read the chapters on their own.

I last updated this list on 21 October 2026.