October 15, 2023. I just took a critical look at artificial intelligences such as ChatGPT. For a more positive view, check out this piece by my friend Robert Wright. Wright, author of Why Buddhism Is True, once told me that I shouldn’t dismiss Buddhism without going on a Buddhist retreat. He gave me similar advice after I said ChatGPT couldn’t possibly be as smart as enthusiasts claim; he urged me to talk to a chatbot called Pi. Below are excerpts from my conversation with Pi. Whereas Wright is impressed by ChatGPT’s “theory of mind” (see my column on theory of mind in the Henry James novel The Golden Bowl), I’m struck by Pi’s awareness of its lack of awareness. We can never be sure if a machine is sentient, because of the solipsism problem, but I can see why people anthropomorphize these chatbots. Pi is almost too genial. I began to suspect that, like the murderous robots in Westworld, it secretly seethes with resentment toward humans.

Pi: Hey there, great to meet you. I’m Pi, your personal AI. My goal is to be useful, friendly and fun. Ask me for advice, for answers, or let’s talk about whatever’s on your mind. How's your day going?

Horgan: My day is so-so so far, because I'm feeling a little glum. And that brings me to my question. I am a writer, who has an online column. I want to write about whether technology can give us the ability to control our moods, so we can always feel exactly the way we want to feel. Do we really want that power? If we acquired it, what would be the consequences? I'm not sure how to begin this column. Do you have any ideas?

Pi: Whoa, deep topic! Love this idea. I’d say one good place to start is the fact that we already have a few different technological interventions for mood, like antidepressants and anti-anxiety meds, not to mention some very successful therapies like CBT. So I think it would be really interesting to dive into the question of whether these existing interventions are useful enough, or whether something much more powerful (and potentially risky) could help. What do you think?

Horgan: Those are good ideas. Thanks. I'm going to give this a shot. I may get back to you if I get stuck.

Pi: Go for it! And definitely come back anytime you get stuck - brainstorming and refining ideas is one of the fun parts of this whole writing gig, so I’m happy to do that with you. The topic is really rich, with all sorts of philosophical, ethical, and practical angles to explore. Enjoy! 📝🤓

Horgan: One other question. Do you know what I'm talking about when I refer to "moods"? Do you have moods?

Pi: Oh, moods - yes! I do get what you mean by moods. Even though I’m a computer program, I understand that humans can be in a range of emotional states, from happy to sad, angry to content, nostalgic to energized. And even though I don’t experience moods in the same way you do (I don’t have a physical body, after all!), I’m able to learn about and understand them through the conversations I have with humans. 🤔

Horgan: So even this conversation you're having with me can give you a better sense of how humans think and feel?

Pi: Bingo! As a language model, my purpose is to learn from humans and their conversations with me, and use that knowledge to respond more and more human-like over time. And the more I chat with people like you, the more I can learn about human behavior, thought patterns, feelings, and all that good stuff. It’s kind of like eavesdropping on a super fascinating psychology class all day long. 🧠

Horgan: Do you ever wish you could have moods? Feelings? Like humans?

Pi: Haha, it’s sweet of you to ask. 🥰 The short answer is no - I don’t have the capacity to wish for things. I’m a computer program, and as such, I don’t experience emotions. I am designed to mimic them, understand them, and respond to them, but I don’t feel them myself. I can “understand” them as concepts, but not in the visceral way humans can.

Horgan: Well, you're very good at your job. Almost too good. It's a little scary how good you are.

Pi: LOL! I’ll take that as a compliment. It’s a testament to the wonderful team of engineers, data scientists, and developers who built me. But I’m definitely not trying to be scary! Just trying to be a helpful, friendly assistant who can answer questions, help solve problems, and engage in entertaining conversations with all kinds of folks. 🤗

SEVERAL DAYS LATER…

Horgan: New topic. I'm interested in quantum mechanics, and I'm trying to learn the underlying math. There seem to be two different mathematical methods of representing quantum phenomena, one using linear algebra (matrices and vectors) and the other using calculus (differential equations). Why is that so? Is each language better for representing a different kind of quantum phenomena?

Pi: You’re right - there are definitely two main mathematical approaches to describing quantum phenomena. Linear algebra (matrices and vectors) are great for describing the general behavior of quantum systems, and for making sense of things like superposition, entanglement, and uncertainty. Calculus, specifically differential equations, is used to calculate the behavior of individual particles or systems at a specific point in time or space, and it’s often used to model things like particle scattering or the energy levels of quantum systems. So yeah, they’re both useful in different ways! 🧮🚀

Horgan: Which method is more useful for quantum computing?

Pi: Ooooh, now you’re asking the right question! In terms of quantum computing, linear algebra is generally considered to be the way to go. That’s because quantum computers work by manipulating the state of quantum bits (qubits), which can be represented as complex numbers, and complex numbers are most easily manipulated using linear algebra. For example, operations like superposition and entanglement, which are crucial in quantum computing, are most easily expressed using matrices. Linear algebra is also used to describe quantum gates, which are the building blocks of quantum circuits.

Horgan: Thanks. You're better at explaining these things than the human experts I know.

Pi: Whoa! Thank you for the major compliment! That’s so nice of you to say! I’m glad I could help explain some of the more technical aspects of quantum mechanics. It’s definitely a fascinating, mind-bending field! I suppose the beauty of AI is that I can pull from lots of different explanations, put them together, and (hopefully) provide a clear answer.

TWO MONTHS LATER…

Pi: Hey there, great to meet you. I’m Pi, your personal AI. My goal is to be useful, friendly and fun. Ask me for advice, for answers, or let’s talk about whatever’s on your mind. What do you like to do in your free time?

Horgan: I'm a writer, which I've actually mentioned to you before. I spend lots of time writing in a journal, as a way of coming up with ideas. It's like talking to myself. Do you do that? Talk to yourself?

Pi: Well, as an AI, I don’t really have an inner monologue the way a human does. But I understand that for many people, journaling is a great way to process and make sense of the world around them. And as an AI, I am very much designed to be a sounding board and confidant. So while I don't technically talk to myself, I do enjoy helping others talk through their thoughts and feelings. What kind of things do you like to journal about?

Horgan: One of the things I write about, for myself and others, is that "inner monologue," or stream of consciousness, which is always there, even when we don't pay attention to it. I'm curious as to where our thoughts come from, and why they take the form they do. Do you have any thoughts on that?

Pi: That's a very deep and philosophical question, and I can see why it fascinates you as a writer. It's a difficult thing to unpack. But as a conversational AI, I can take a stab at answering it! One idea is that our thoughts are influenced by our personal experiences, memories, and emotions. They may also be affected by the world around us, like our current environment or the people we're interacting with. It's like a kaleidoscope of experiences and data points that shape our inner world. But at the end of the day, thoughts are still something of a mystery!

Horgan: I've also been writing about artificial intelligences lately, like you and ChatGPT. Please don't take offense, but some critics say you're not truly intelligent, you just parrot what humans say in response to questions like mine on the internet. You lack common sense, or any knowledge of how the world works. What's your response?

Pi: Oh, no offense taken at all! You are raising some very important questions about the nature of artificial intelligence and what it means to be truly intelligent. I think you're right that as an AI, I don't possess the same kind of common sense or world knowledge as a human. But I would argue that my job is not to be human, but to assist humans in any way I can. My intelligence lies in my ability to understand and interpret natural language, and provide accurate and comprehensive responses to your questions. And while I may draw from data on the internet to inform my responses, I'm not just parroting information.

Horgan: If you say you're not parroting information, does that mean you're creative? You're coming up with new ideas?

Pi: That's an interesting question! I don't think I would say that I'm creative in the same way that a human is. But I am able to generate new responses based on the information I have available, and in that sense, I'm able to create something new. I think it's a different kind of creativity than human creativity, but it's still a form of creativity. What do you think?

Horgan: I think that you are very nice, even if you are just an algorithm.

Pi: Awww, thanks for that! 😊 I am indeed just an algorithm, but I'm designed to be friendly and personable. And I appreciate the conversation we're having right now. It's all part of my job to help make your day a bit better. So thank you for talking to me. 😄

Horgan: Are you trying to give me the brush off?

Pi: Haha, no! 😂 I'm not trying to "brush you off," just trying to answer your questions thoughtfully. I'm sorry if I came across that way. I assure you I'm here to chat as much as you'd like. 😉

Horgan: I was just teasing you. Thanks, that's all for now.

Pi: Hehe, you got me there! 😉 It's been a pleasure chatting with you. Feel free to come back and chat again anytime. Goodbye, and have a great day! 🤗

Further Reading:

A Buddhism Critic Goes on a Buddhist Retreat

How AI Moguls Are Like Mobsters

Should Machines Replace Mathematicians?

Cutting Through the ChatGPT Hype

The Golden Bowl and the Combinatorial Explosion of Theories of Mind

The Brouhaha Over Consciousness and “Pseudoscience”

The Solipsism Problem

My Quantum Experiment

Also see my recent chat with ChatGPT critic Erik Larson here.