Dialogues

Introduction

In an age when writing assistants abound, ones based on large language models have struck a chord with many people who write for a living and others who need to write. The range of current use is quite large, varying from writing college essays, writing novels and all sorts of online content to include the HTML for websites. This is a tiny list of forms attempted so far.

I started using ChatGPT 3.5 several months ago. I began by using the prompts to generate essays on various topics. There are many recipes that can be found on the web for generating essays. I followed the recipes and generated a few, some included references. However, I was both unsatisfied with the final product because the meaning was not quite what I wanted it to mean and the ethics of presenting something I did not write. I did see the attraction for lazy college students.

Since I understand that ChatGPT has vast repository of knowledge embedded within its large language model, I decided not to disguise its output and use it to elicit, hopefully, expert opinion on topics of interest and label my responses and further questions explicitly in order to make the distinction clear. I expect that the "expert" opinion to be a consensus of current expert thought and not made up to please me or the audience.

In order to accomplish this goal, I have decided to prompt ChatGPT so that it will reply in the format of of a dialogue between an expert, a student, and a skeptic. This is the format used by many philosophers, most notably Plato. Galileo used it to get himself in trouble with the Catholic Church by writing a dialogue entitled: "Dialogue Concerning the Two Chief World Systems". In it, Galileo uses three interlocutors: Salviati, Sagredo, and Simplicio. This dialogue is an argument between Salviati, an adherent of the Copernican model of the solar system, and Simplicio, an adherent of the Aristotle-Ptolemy model. Sagredo is a model neutral member of the dialogue. I do not intend the dialogues that follow to be argumentative but instructional where the readers learn something or questions are provoked in the readers. I, myself, have learned new things and have had many clarifications from the dialogues that I have already prompted. I have found this exercise to be a very good learning experience.

I intend these dialogues to be "living" in that I can add prompts as new questions about a given dialogue arise or I think more clarification can be elicited by another prompt. I invite the reader to propose additional questions or request additional clarification. I make my additional prompts as either the student or the skeptic. I ask the readers who wish to contribute to specify their questions as either the student or the skeptic.

I will write an introduction for each dialogue describing its basic topic and aims. I will also write a commentary at the end of each that summarizes key ideas brought out in the dialogues. In order to distinguish between what is generated by ChatGPT and what is human input an output, ChatGPT output will be italicized and human discourse will not be. A ChatGPT prompt by a human will be preceded by the character: 'Who:>' and not be itlalicized.