Elements of AI: Review

A PM’s confession: I almost skipped this because it’s a reading course

I’m Tom, product manager, professional non-engineer. When vendors pitch us AI features, my job is to smell the difference between capability and marketing. I wanted a course that sharpened that nose. Elements of AI kept getting recommended, with one warning attached: it’s text. Almost no video. I nearly bounced off that — I watch courses, I don’t read them — and I’m glad I didn’t.

Some background, because it’s genuinely interesting: this was Finland’s gift to the EU during its council presidency, an AI literacy course for every EU language, built by the University of Helsinki and MinnaLearn with Professor Teemu Roos as academic lead. It’s now at roughly two million learners across 170 countries. Part 1 runs about 10–30 hours depending on reading speed, no deadlines, no cohort, and everything is free — including the LinkedIn-postable certificate, which is rare. A Part 2 (Building AI) adds basic Python, and Finnish learners can even get 2 ECTS credits.

What it covers

Six chapters. Definitions and philosophy first (the Turing test shows up), then problem-solving and search, then probability and Bayesian methods to ground AI in the real world. Chapters four and five introduce machine learning — with a careful, jargon-free explanation of how ML differs from traditional programming, which is exactly the distinction I need when engineers explain things to me — and neural networks. The last chapter is implications: bias, surveillance, employment.

That last chapter is why I’m recommending this over the flashier options. Reviewers consistently call it more substantive than the ethics lip-service free courses usually phone in, and I agree. As a PM I sit in meetings where “is this creepy?” is a real product question. This course gives you vocabulary for that conversation.

The exercises deserve a mention too. Around 1,000 Class Central reviews average 4.8/5, praising questions that test understanding rather than recall. Fair warning from those same reviews: most quiz questions allow exactly one attempt. “Unforgiving” is the word people use. Read carefully before clicking.

What annoyed me

The material predates the LLM era, and it shows. Generative AI, ChatGPT, prompt engineering — minimal coverage. For a course about AI’s place in the world, that’s a real hole in 2026; the tools I actually use at work barely appear. The dense text format is also genuinely tiring in places — some learners find it dry, and I had evenings where I felt that. No hands-on projects, no forum, no peer support. And while the free certificate is nice for LinkedIn, reviewers note it carries little weight in technical hiring.

Who it’s for

People who will live with AI decisions without implementing them: managers, policymakers, educators, marketers, humanities students. No math, no code, no prerequisites for Part 1. Not for developers targeting ML roles, not for anyone who wants to train models, and not for people already fluent in generative tools looking for depth.

The neighbors: AI for Everyone is shorter (~6 hours), video-based, more corporate-strategy flavored — pick it for speed, pick Elements of AI for rigor and ethics. Google AI Essentials is the generative-native option if “AI” to you means ChatGPT-class tools. And CS50’s AI with Python is the technical upgrade if Part 1 makes you want to build.

Bottom line

Free, zero-prerequisite, rigorous where its peers are fluffy, honest about ethics where others gesture. Still one of the best starting points for AI literacy anywhere. But its pre-LLM content means you now need a companion course on generative AI to get the whole story.