In 2013, University of Toronto professor Geoffrey Hinton (center) is pictured with graduate students Ilya Sutskever (left) and Alex Krizhevsky (right).
Johnny Guatto/University of Toronto
Artificial intelligence has been revolutionized by several key advancements, with one of the most pivotal being the introduction of AlexNet in 2012. This neural network marked a significant leap forward in image recognition capabilities by computers.
On Thursday, the Computer History Museum (CHM), in partnership with Google, made the AlexNet source code created by University of Toronto graduate Alex Krizhevsky publicly available for the first time on GitHub, allowing anyone to review and download it.
According to the Museum’s organizers, they take pride in sharing the source code for the 2012 version of AlexNet, developed by Krizhevsky along with Ilya Sutskever and Geoffrey Hinton, a project that reshaped the landscape of artificial intelligence.
Krizhevsky’s innovation sparked extensive advancements and investment in AI, proving that neural networks could achieve significant progress with adequate data and computational resources. The source code, merely 200KB in size, consists of Nvidia CUDA code, Python scripts, and some C++ to outline how to build a convolutional neural network capable of processing and categorizing images.
Hansen Hsu, the CHM’s software historian, dedicated five years to coordinating with Google, the entity owning the rights to the source code, for its release, as detailed in his essay about the history of AI and the origins of AlexNet.
During this time, Krizhevsky was a student under the guidance of Hinton, who is a Nobel Prize-winning scientist in AI. Sutskever encouraged Krizhevsky to pursue the project, leading Hinton to later jokingly express that, “Ilya thought we should do it, Alex made it work, and I got the Nobel Prize.” Google maintains ownership of the AlexNet intellectual property due to its acquisition of the founding trio’s company, DNNResearch.
