Why This Time Is Different

Before I began my interviews, I sought some practical context regarding the development of Artificial Intelligence (AI). Since a complete history of AI’s evolution since the 1950s would exceed the scope of this book, I looked for someone who had been actively involved. I found this individual in Jonathan Cohen, Chief Investment Officer of the Robocap UCITS Fund. Cohen graduated from my alma mater the University of St. Gallen and spent many years working in finance in Switzerland and the UK. He started investing in the robotics and AI sector back in 2015. At that time, very few public stocks were connected to AI. Prior to the rise of GenAI, he focused on computer vision, natural language processing, and machine learning. In 2021, he wrote about an early version of ChatGPT for the first time. What surprised him, as well as many of his advisors, was the remarkably fluent language output produced by ChatGPT once it could handle a large number of parameters. Two factors facilitated this. First, the tremendous increase in compute power. Today’s computer chips are a hundred times faster than they were 2 years ago and a million times faster than 10 years ago. The chips were designed by Nvidia, which previously provided chips to the gaming and crypto industry for mining. Second, Google’s research paper on Transformers established the groundwork for the new type of GenAI algorithm. Like everyone else, Cohen was struck by the massive capital expenditures triggered by the AI phenomeon…