Stanford: A Free AI Course Beyond ChatGPT

A Revolutionary AI Course Offered by Stanford

Stanford University is offering a free online course that goes well beyond teaching current artificial intelligence tools like ChatGPT. This course, led by prominent researchers Peter Norvig and Sebastian Thrun, focuses on the fundamental ideas that truly enable an understanding of AI.

A Comprehensive Program

This course, taught by two pioneers in AI, is freely accessible and covers much more than just language models or prompt engineering. Participants should expect to invest between 75 and 100 hours to master the complex and enduring lessons offered.

Imagine a world where the transporter technology from Star Trek becomes a reality and is universally adopted. This would radically transform our daily lives, with transport platforms at every corner, allowing travel from New York to Los Angeles in five minutes for perhaps 20 dollars. This technological revolution would create new business opportunities, requiring training to install, use, and integrate these transporters across various sectors.

A Broader Vision of AI

However, despite the potential impact of this technology, most people would be content to learn how to use it without seeking to understand the underlying physical principles. Generative AI, while widely adopted, presents a similar situation. Teaching tools are multiplying, but AI is not limited to transformers, large language models, or modern applications. Understanding these aspects is crucial for developing solid AI skills.

In summary, generative AI is just one facet of a much broader field. It is in this spirit that Stanford offers an AI course dating back to 2011, which remains relevant today.

Learning the Fundamentals

Fundamental knowledge in AI does not age as quickly as product-related knowledge. While agentic AI is recent, the fundamental problems studied by researchers and the methods used to solve them are enduring and worth learning.

Udacity offers a free version of the course "Introduction to Artificial Intelligence," originally taught by Norvig and Thrun at Stanford in 2011. This course attracted over 160,000 participants during its initial launch, becoming one of the first MOOCs and contributing to the creation of Udacity by Thrun.

Peter Norvig is a renowned researcher at Stanford's Human-Centered AI Institute and a research director at Google. He co-authored "Artificial Intelligence: A Modern Approach," a reference textbook still used in university AI courses in 2026. Sebastian Thrun, founder and executive chairman of Udacity, is known for his pioneering work in autonomous vehicles and robotics.

In-Depth Exploration of AI

The course covers various topics such as:

• Problem-solving and search
• Probability and probabilistic inference
• Machine learning and unsupervised learning
• Knowledge representation through logic
• Planning and planning under uncertainty
• Reinforcement learning
• Hidden Markov models and filtering
• Markov decision processes
• Adversarial search, games, and game theory
• Computer vision
• Robotics and robot motion planning
• Natural language processing

These concepts are not ephemeral products but rather enduring mathematical and conceptual foundations, essential even as technology evolves. This course serves as an excellent complement to current training on generative AI.

An Expanded Understanding

Today, large language models do not solve all the challenges of AI. This course explores the expanded universe of AI, highlighting the fundamental problems researchers considered before the rise of chatbots. It demonstrates that artificial intelligence involves much more than just generating language and images. AI systems must represent knowledge, assess uncertainty, search for solutions, plan actions, perceive their environment, and learn from their outcomes.

Udacity does not provide a precise estimate of the time required to complete the course, which includes 22 main lessons, problem sets, introductions, and exams. A lesson on game theory, for example, contains 19 segments and takes about two hours to complete, including exercises. It is advisable to allocate between 75 and 100 hours to follow the entire course.

This course does not replace current training on generative AI. It predates transformers, LLMs, diffusion models, and contemporary development frameworks. It also does not address AI ethics or the major issues it raises. Consider it a complement to more recent sessions that cover current models and tools.

For those who simply want to use chatbots, a course on generative AI may suffice. But for those who want to understand enduring concepts and a broader perspective of AI, the course by Norvig and Thrun is a hidden treasure worth exploring.