The 2024 Nobel Prize in Physics has been awarded to John Hopfield and Geoffrey Hinton, and the press release is here. Like many, I was surprised that the award was given to researchers outside the field of physics. I was further surprised by one statement in the release:

The Boltzmann machine can be used to classify images or create new examples of the type of pattern on which it was trained. Hinton has built upon this work, helping initiate the current explosive development of machine learning.

When I survey the field of machine learning, the central mechanisms are convolutional neural networks (CNNs) for image analysis and transformers for natural language processing (NLP). All the large language models in popular use, from Meta’s Llama to Google’s Gemini to OpenAI’s GPT, rely on transformers. Corporations are struggling to improve the speed and efficiency of transformers, while hackers like me struggle to access them in applications.

Boltzmann machines are fascinating, but I fail to see how they’ve influenced the “explosive development of machine learning.” I’ve searched the Internet, and I’ve yet to find any practical uses of Boltzmann machines in industry or academia.

I’m disappointed that none of the transformer’s developers have won a Nobel prize or even a Turing award. But the reason is simple—these awards are given by academics to academics, and commercial developers aren’t included.