About

About the AI Masterclass

An online educational program in artificial intelligence for students and lifelong learners worldwide.

Mission

The AI Masterclass teaches artificial intelligence at the level where the mathematics begins, with the equations, the numbers, and the methods researchers use in their own work. Most of what is written about the field stops at the point where a loss function, a gradient, or a probability would be needed. Our units keep the mathematics and explain every step of it, so that a participant with school algebra can follow how a network finds every gradient in one backward pass, how a positive test for a rare disease is turned into a probability, or what a scaling law does and does not predict, and then repeat the calculation on their own.

Participants receive one unit per week by email, as a PDF and as a web page. A quiz at the end of each unit shows whether the material has been understood, and the next unit follows once it is done. The program closes with a final exam of 50 questions in 60 minutes. Those who reach 70 % receive a certificate and a transcript, each with a code that a school, a university, or an employer can check on our verification page.

Vision

Everyone now uses systems built on machine learning, and almost nobody is taught how they work. School students who want to know meet the subject as a collection of product names; university students in subjects that have started to use these methods meet them as library calls; adults who work with the tools every day read about them in a language of hype that explains nothing. We want anyone with school mathematics and an internet connection to be able to work through the real methods at a steady pace, under clear standards, and to finish with a record of what they have achieved. This subject is written about with more exaggeration than any other, and the units are the place where the exaggeration stops.

We also want participants to come close to current research. Every unit introduces published work by researchers active today and lists papers that can be read free of charge, so that by the end of the program a participant can open a new paper on a training method or an interpretability result and follow the argument, including the numbers behind it.

The program

The curriculum has 16 units. It opens with the toolkit every AI researcher needs (vectors, probability, and derivatives) and continues through search, reasoning under uncertainty, learning from data, classifiers, neural networks and their training, and the big questions of the field, to convolutional networks, embeddings, transformers, reinforcement learning, and the four units on large language models. The Foundation masterclass covers the first 8 units, the Advanced masterclass the first 12, and Advanced + Large Language Models all 16.

Each unit is a self-contained chapter. Key equations, derivations, worked examples on data small enough for pencil and paper, and historical excursions are set apart in boxes, so that a unit can be read at the depth a participant chooses. No unit depends on code: every method is worked by hand, and the practical coding lives in the free tool of the field that comes with each unit. Research spotlights present the work of leading researchers, and a list of further reading follows each unit.

The weekly rhythm, the quizzes, and the final exam exist because people learn more when something is expected of them. The Advanced masterclass takes 12 weeks at the standard pace. Participants can take longer, and they can ask for the next unit early.

Development

The masterclass was developed in 2026 for students worldwide. Every unit was written for this program and typeset in LaTeX, so the PDF and the web version come from the same source and show the same equations. One notation runs through all sixteen units, so that the parameters a participant meets in Unit 4 are the same symbol in Unit 14. Parameter counts, dates, error rates, and benchmark results were checked against the original paper or the organizer's page, and the source is named where the number appears; where a number is not published, the unit says so. The quiz questions and the pool of exam questions were written together with the units, ten exam questions for each unit, so that the final exam tests what the units teach and nothing else. The material is in English and assumes no prior knowledge of artificial intelligence or programming.

Editorial team

The units, quizzes, and exam questions of the masterclass and the articles on our blog are published by the Artificial Intelligence Editorial Team. Blog articles explain the methods of the field for school and university students, each with a worked number, and link to the published sources they rely on. The researchers whose work appears in the research spotlights are not involved in the program and have not endorsed it. If you find an error in a unit or an article, please write to us at info(at)masterclassai.org and we will correct it.

Organizer

The AI Masterclass is organized by Edu.Harbour, an educational organization from Germany that runs international academic programs and competitions for school and university students. Edu.Harbour is responsible for the curriculum, the exams, and the certificates, and its academic office answers questions from participants and their schools. You can find more about its work at eduharbour.com.

Funding

The masterclass is funded by Tuition Fees and by sponsors. The Tuition Fee pays for writing and updating the units, for running the platform, and for issuing and verifying certificates. Sponsors contribute so that the fee can stay lower than the program would otherwise cost. They have no influence on what the units teach or on how exams are marked.