Today, you hear something about artificial intelligence every five minutes. But have you ever wondered how we got here? This is largely due to AlexNet, a pioneering neural network from 2012. While it wasn't an immediate sensation, it eventually became the basis for the deep learning revolution we're experiencing today. Now, after years of negotiations, the original source code has finally been released to the public.

This is thanks to a partnership between the Computer History Museum and Google. The source code was originally written by Alex Krizhevsky, a graduate student at the University of Toronto, and has been uploaded to GitHub.

AlexNet is a neural network that marks a major breakthrough in computers' ability to recognize and classify images. By 2012, the theory behind neural networks, including the key backpropagation algorithm, had been around for decades. However, two key components are missing: the large data sets required to train these networks and the raw computing power required to process them. Initiatives such as Stanford's ImageNet project and NVIDIA's CUDAGPU programming ultimately provide these key elements.

These advances allowed Krizhevsky, under the guidance of artificial intelligence pioneers Geoffrey Hinton and Ilya Sutskever, to train AlexNet and unlock the full potential of deep learning. This is the first time that deep neural networks, large data sets and GPU computing have been combined and achieved breakthrough results. Each aspect is vital to the other.

Home computer with GPU used to create AlexNet

Finally, the AlexNet paper was published at a computer vision conference in 2012. At the time, most researchers dismissed it, but Yann LeCun, now recognized as an AI pioneer, immediately recognized its importance, calling it a turning point in the field. His prediction turned out to be correct. After the release of AlexNet, neural networks quickly became the basis for almost all cutting-edge computer vision research.

The breakthrough of AlexNet was to prove that training a relatively simple neural network can achieve superhuman performance on highly complex tasks such as image recognition. This marked the birth of the deep learning paradigm, in which machines acquire skills by ingesting and modeling massive data sets.

From that moment on, progress began to accelerate rapidly. Neural networks are advancing at an unprecedented rate, achieving many milestones, such as defeating human champions at the game of Go, synthesizing lifelike speech and music, and even generating original art and creative writing. However, the real turning point for generative artificial intelligence is the release of ChatGPT by OpenAI in 2022, which can be said to be the pinnacle of the evolution of deep learning.

Understandably, open sourcing such a historic piece of code is no easy task. The Computer History Museum had to negotiate for five years with Krizhevsky, Hinton (now at Google), and Google's legal team to get approval to release the original source files.