AI Demystified

Generative AI is a general-purpose technology.

Originating in successful code-breaking efforts during the Second World War, artificial intelligence (AI) was always an effort to teach machines to think more or less like humans, but with the speed and accuracy computers are capable of at any given stage in their evolution. In its early incarnations, AI was defined by a set of rules it was programmed to deploy, and for that reason, the sophistication of the tasks AI was able to effectively perform was limited.

Generative AI is different. Its possible applications are manifold. Generative AI can produce novel ideas and content, and thereby solve problems in creative ways. Trained on vast amounts of data from text, images, video, and sound, generative AIs like OpenAI’s ChatGPT deploy large language models (LLMs), in particular, transformers, to determine the statistical relationship between text, visuals, or sounds in sequence and to predict what word, image, or sound will come next. Generative AI models can analyze data and make predictions in a wide variety of areas. And these models can be set up so they continuously improve as...