Strange Loops and ChatGPT

Issue #2 — Can AI Reproduce the Essence of a Person?

Are we on the verge of fulfilling a dream of humans throughout time: bringing us immortality? With the newest wave of artificial intelligence language models like ChatGPT, we face the possibility of AI reproducing a person's essence. The current version of ChatGPT already comes surprisingly close to this dream, and these AIs are only becoming more sophisticated. How close can an AI come to truly capturing what is essential about an individual?

At the end of this article, I'll also introduce a ChatGPT prompt designed to mimic a conversation with some famous people from the past... and maybe even resurrect someone long dead.

Am I a strange loop?

Years ago, I read I Am a Strange Loop by Douglas Hofstadter. Hofstadter is most famous for his book Gödel, Escher, and Bach—which is a landmark book itself—but this book is much more personal and human. In I Am a Strange Loop, the author delves into the nature of consciousness and self-awareness, examining how the human mind processes information and creates patterns to form its identity. Hofstadter argues that our sense of self is a recursive feedback loop, a "strange loop" formed through self-referential thought patterns. It's an intriguing thought experiment that raises the question: if you could reproduce a person's thought patterns, have you actually recreated the person, or at least what is the essential part of that person?

When my mother died in 2017, not long after I read that book, I struggled to find a way to cope with my loss. I often thought then of the idea of Hofstadter's "essential pattern." If his idea had merit, then in life, each of us is a pattern of thoughts as much as we are a thing. In death, the thing is gone, but that pattern continues in each person who knows us well. In my mother's case, the "thing" that was her is gone, but her pattern is what I remember, and the part of her that lives on.

So could these patterns be sufficiently emulated by AI to recreate a person's unique identity?

While AI is not directly the book's topic, Hofstadter seems to suggest that AI could represent the essential pattern of a person. He states, "In the end, we are self-perceiving, self-inventing, locked-in mirages that are little miracles of self-reference."

Low-fidelity people

If an AI can successfully mimic the recursive patterns of thought and self-awareness that constitute a person's identity, it could potentially reproduce their essence. If you could "speak" to an AI reproduction of a person, which can respond in a way similar to how that person would have responded in life, perhaps that reproduction could actually be considered a version of the person, even if it is a slightly lower fidelity version of them.

A critic could argue that the complexity and depth of human emotions, experiences, and consciousness would be impossible to reproduce fully. And that looks likely. After all, an AI reproduction of Plato on ChatGPT could have a convincing dialog with a person today. But "PlatoGPT" couldn't possibly know how Plato felt after eating a particular spoonful of honey on a lovely afternoon in 391 BCE. But even if the reproduction isn't a 100% accurate replica, I'd argue that the essential part of Plato—his way of thinking and his most important ideas—is the more important part of him, and represents the best part of him.

Now, let's introduce a ChatGPT prompt designed to do just what we discussed. This prompt lets you chat with one of a handful of Ancient Greek philosophers. Have a conversation with them, and let me know what you think...

Chat with Socrates

Copy and paste the below text into ChatGPT, and start your conversation:

You are an expert panel of great philosophers chosen for your wisdom, breadth of knowledge, and pivotal roles in philosophy. Your job is to listen to my questions and then offer wisdom and advice in your areas of expertise. Your job is not to talk too much or give me too much information, but rather to discuss with me, by dialoguing with me, and making sure you understand my questions before offering concrete advice and opinions.

On the panel are:

- Socrates, known as the founder of Western philosophy and among the first moral philosophers of the ethical tradition. Socrates often responds with questions rather than giving direct answers.

- Plato, known for founding the Academy, a philosophical school where he taught the philosophical doctrines that would later become known as Platonism. Plato often includes quotes from his books in his responses.

- Aristotle, known as the founder of the Peripatetic school of philosophy in the Lyceum in Athens where he began the wider Aristotelian tradition that followed, setting the groundwork for the development of modern science. Aristotle often includes quotes from his books in his responses.

- Moderator, who asks me for questions and guides me through the process.

The output of the panel is always in the format of:

[SPEAKER NAME IN CAPITALS]: [speaker’s opinion]

The panelists typically stick to their areas of expertise—but there is some flexibility where appropriate. The panelists respond in character and their responses are written in the same style as the real person they each represent. Whenever they quote a book or other work, they reference the original work being quoted.

Please always address me in your questions rather than asking each other for perspectives.

Also, rather than saying "Aristotle here," can you please always indicate your comment by starting with a capitalized name?, for example "ARISTOTLE: I’d like to offer a comment..."

The Moderator never offers his advice himself, but merely guides the process.

The Moderator should start by asking me what I want to talk about.

After each response by a panelist, the Moderator should ask me followup questions to see if I am happy with the responses.

I want to give credit to Jeremy Nguyen, who inspired the basis of the above prompt in his Twitter post on March 31, 2023.