Wolfram U is the education site of Wolfram Research, the company behind Mathematica, Wolfram|Alpha and the Wolfram Language. It offers open interactive courses, live training events, video lessons and certifications, and nearly all of it is free. The artificial intelligence category is a small, tidy corner of the catalogue. The core courses are three short instructor-led classes. ML201 Introduction to Machine Learning in Wolfram Language and ML202 Introduction to Neural Networks in Wolfram Language each take about two and a half hours.
ML215 Wolfram Language and LLMs is the newest and most relevant for 2026. Its syllabus covers Wolfram AI access with no external accounts, the AI Assistant chat inside notebooks, prompt engineering, built-in functions such as LLMSynthesize, LLMFunction and LLMExampleFunction, building tools an LLM can call, and the Wolfram Prompt Repository. It is marked intermediate and assumes basic Wolfram Language. The next live session listed is on Thursday 7 January, run on BigMarker.
There is also a longer ML810 course, Exploring AI Foundations with Wolfram Tools, at about seven and a half hours, and an interactive course on convolutional neural networks for computer vision. Certificates work in two layers. Each course gives a completion certificate once you have watched the videos and passed the auto-graded quizzes. Those certificates count toward Level 1 and Level 2 certifications: Proficiency in Neural Networks at Level 1, and Applied Expertise in Neural Networks and Applied Expertise in Wolfram Technology for AI at Level 2. The Level 2 AI certification requires an independent project, which gives it a bit more weight than a quiz badge. What is it like to learn this way? Pleasant, honestly.
Wolfram Language's high-level functions mean you can train a classifier, inspect a neural net or wire an LLM into a function in a few lines, and the notebook interface is good for exploration. Instructors are Wolfram staff who know the tools well. But the abstraction cuts both ways. You come away knowing which function to call, not what the optimiser is doing, and the skills do not move easily to PyTorch, Hugging Face or LangChain, which is where the jobs are. So our recommendation depends on who you are.
Researchers, engineers, maths teachers and students whose institution already licenses Mathematica should take ML215 at least, because putting LLM calls inside the tool you already use is practical. Anyone else looking to learn AI for work should start with Python-based courses and come back to Wolfram only if a project calls for it.