Supremacy: AI, ChatGPT, and the Race that Will Change the World

by Parmy Olson

The T in ChatGPT stands for “transformer.” This has nothing to do with the alien robots that morph into eighteen-wheelers but a system that allows machines to generate humanlike text. The transformer has become critical to the new wave of generative AI that can produce realistic text, images, videos, DNA sequences, and many other kinds of data. The transformer’s invention in 2017 was about as impactful to the field of AI as the advent of smartphones was for consumers. Before smartphones, mobile phones couldn’t do much more than make calls, send texts, and play the odd game of Snake. But when touch-screen smartphones hit the market, suddenly people could browse the internet, use GPS, take high-quality photos, and use millions of different apps. Transformers also broadened the scope of what AI engineers could do. They could handle far more data and process human language much more quickly. Before transformers, talking to a chatbot felt like talking to a dumb machine because the older systems operated on sets of rules and decision trees. If you asked a bot something that wasn’t already written into its programming (which was often likely), it would be stumped or make a peculiar error. That was how digital assistants like Apple’s Siri, Amazon’s Alexa, and even Google’s Assistant were originally designed. They treated each query as a single isolated request, which meant they were terrible at context. They couldn’t remember questions you’d previously asked in the same way a pers...