← Retour au blog
tech 22 June 2026

Munich 1991: The Roots of the Current AI Boom

Today's explosive growth in Artificial Intelligence traces its roots back to pioneering work done in Munich in 1991. Discover how the innovative ideas from that era continue to drive current technological advancements.

Article inspired by the original source
Munich 1991: The Roots of the Current AI Boom ↗ people.idsia.ch

A Pivotal Year for AI

In 1991, at a time when computing was far less advanced than today, a laboratory at the Technical University of Munich, led by Jürgen Schmidhuber, laid the foundations of what has become the Artificial Intelligence industry as we know it. The work accomplished in just a few months laid the groundwork for technologies that are revolutionizing the sector today.

The Revolutionary Advances of 1991

The First Transformer

March 26, 1991, marks the publication of the first type of Transformer, known as the unnormalized linear Transformer. This innovation significantly reduced computational costs, making it feasible to analyze large amounts of data. Today, Transformers are at the heart of language models like ChatGPT.

Pre-training of Neural Networks

On April 30, 1991, Schmidhuber's team introduced the concept of pre-training for deep neural networks. This technique paved the way for advances such as deep learning, which is crucial for the development of modern AI.

Deep Residual Learning

Deep residual learning, also introduced in 1991, is a central component of Long Short-Term Memory (LSTM) networks and Residual Neural Networks (ResNets). These techniques are among the most cited in scientific literature and have greatly influenced current information processing methods.

The Lasting Impact of Munich's Work

The innovations of 1991 not only influenced the technologies of the time but continue to support current advancements. For instance, Generative Adversarial Networks (GANs), which are foundational for generative AI, trace their roots back to concepts developed in Munich.

Practical Applications Today

Today, companies like Google, OpenAI, and many others rely on these techniques to develop applications ranging from language modeling to image and video creation. Investments in these technologies reach hundreds of billions of dollars, illustrating their strategic importance.

Conclusion

The year 1991 in Munich was a real turning point for AI, laying the foundations for advancements that continue to transform the world. The innovative ideas of Schmidhuber and his team have not only stood the test of time but have also thrived, becoming pillars of modern AI.

Let's discuss your project in 15 minutes.

AI Deep Learning Transformers Jürgen Schmidhuber Munich 1991
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call