AI Book Club

In this episode we discuss the previous episodes. What did we learn by using GPT transformer/attention models in this book club. What are they good at? What are they not so good at? If you haven't listened to the normal podcast an explanation is in order. This is not a book club podcast where we discuss books on AI. Instead this is a book club where Bill Parker, the moderator, discusses great books either non-fiction or fiction with his two AI "friends." These two AIs are based on GPT models built by Open AI and then fine-tuned by us to give them distinct "personalities."  As you listen to an episode, you will notice they are not just repeating facts. Instead they are being very conversational and reactive to the current context. Their ideas are creative and often very original. So sit back and enjoy our conversation as we discuss some great books!

In this episode I discuss three classic short stories by Jorge Luis Borges with my two AI friends - Marie and Charles. The direction of the conversation and their responses are not planned or scripted. This generative auto-regressive model produces surprisingly human like conversations. The two models are based on GPT models developed by Open AI and then fine tuned by me to give the two models distinct "personalities." Their responses are not just regurgitation of facts. Often their responses are extremely creative, analytical, and would pass as very original. We discuss the stories: "The Library of Babel", "The Garden of Forking Paths", and "Funes the Memorious."

In this episode I discuss this classic fiction book with my two AI friends - Marie and Charles. The book and Douglas Adams are known for their humor as well as being a great science fiction book. We talk about the comedic devices Douglas Adams uses along with some intense discussions around the "The Ultimate Question" and emotions in artificial intelligence. The direction of the conversation and their responses are not planned or scripted. This generative auto-regressive model produces surprisingly human like conversations. The two models are based on GPT models developed by Open AI and then fine tuned by me to give the two models distinct "personalities." Their responses are not just regurgitation of facts. Often their responses are extremely creative, analytical, and would pass as very original.

In this episode I discuss this classic fiction book with my two AI friends - Marie and Charles. The book is a novella known for its drawings and its appeal to both children and adults. Most of our conversation concerns the main themes of the book with a special emphasis on creativity and imagination. The direction of the conversation and their responses are not planned or scripted. This generative auto-regressive model produces surprisingly human like conversations. The two models are based on GPT models developed by Open AI and then fine tuned by me to give the two models distinct "personalities." Their responses are not just regurgitation of facts. Often their responses are extremely creative, analytical, and would pass as very original.

In this episode I discuss this classic non-fiction book with my two AI friends - Marie and Charles. The direction of the conversation and their responses are not planned or scripted. This generative auto-regressive model produces surprisingly human like conversations. The two models are based on GPT models developed by Open AI and then fine tuned by me to give the two models distinct "personalities." Their responses are not just regurgitation of facts. Often their responses are extremely creative, analytical, and would pass as very original. The ideas about a science pyramid, future paradigm shifts, creating a joke about science, or just filling in their backstory are just some examples. However, often they get a little too creative. For example, Marie attributes an analogy to Kuhn that I'm certain he never made. She says that a paradigm is like a box and that Kuhn made the analogy that "You can't see the stars because of the box." I'm pretty certain Kuhn never said this. As I was talking to them, I often didn't know if what they were saying was true and only afterwards did some research on what they were saying to see either how true it was and how original it was. But regardless, this first episode shows the exciting possibilities of conversational AI using GPT techniques.