As It Turns Out, I Am An AI

Well, at least I have some AI traits that leave me wondering

Ever since reading Scott Alexander’s review of Surfing Uncertainty I have wrestled with my own uncanny identification with the core thesis of that book. Namely that, as Scott summarizes it:

The key insight: the brain is a multi-layer prediction machine. All neural processing consists of two streams: a bottom-up stream of sense data, and a top-down stream of predictions. These streams interface at each level of processing, comparing themselves to each other and adjusting themselves as necessary.

I don’t necessarily entirely agree with this premise or the larger predictive processing model theory. However, I do find it to be not just thought provoking but a fairly good approximation of how minds work and interact with reality.

All of this has come full circle with the recent developments in AI. I find that a lot of how I think and how I behave seems to match up with AI behavior.

Essentially I am an estimator. In many areas I am average, but in some I am a very, very good estimator. I make a lot of guesses. Many seem to be close enough to correct that I get credit.

Some of this is learned behavior as I have come to realize you don’t get held responsible for a lot of guesses if they are educated, have parameters limiting their negative impact, and are done with enough confidence.

This phenomenon is not limited to mental exercises. I see it in how I play sports too. In fact the prediction/recalibration/prediction cycle seems essential in physical activities.

To the extent artificial intelligence is about (accurate and improving) prediction, I am with the AIs. And if consciousness is predicated on the ability to make if not execute predictions, we again seem to be one of the same.

Looking at the output of ChatGPT, Bing’s Sydney, and the rest I see another way I am like an AI. AIs are not aware of what they are doing, and we have no idea how they are doing it. What’s more, in many cases we don’t necessarily know that when they give us answers they are, how should we say it, . . . well, full of shit.

How does this relate to me. Well, you see, I have a superpower. I am a master of bullshit.

Don’t believe me? Try me sometime.

Now, unlike AIs I am aware of when I am bullshitting. I know when what I’m saying is made up. What has taken me years to realize is that other people don’t. This is a problem because I don’t usually want to mislead, or if I do, it is for comedic effect and I want everyone to be in on the joke . . . eventually.

For example, in 2015 I was in New York City on a manager due diligence trip. Having some time to myself, I went for a walk through midtown eventually winding up in Greenwich Village and specifically Washington Park. I’ve always had nostalgia for Washington Park being that my uncle owned the car that is used by Meg Ryan in the opening scenes of When Harry Met Sally, in which of course Washington Park plays a backdrop roll. Anyway, friendly guy that I am I was lured into a chess match with one of the famous black chess players of the park. My chess knowledge at the time was fairly limited, but I had good instincts. As I was walking by, one of the players asked if I had ever played, and I made a flippant joke that I thought was obviously a joke about “hoping to advance within the coming few years to grandmaster, but yeah, I’m okay”. This raised more than eyebrows and in a matter of moments I was ushered to a table and the clock was tapped. I knew I would lose, but I hoped to make it comical enough that it would be obvious that I was being sarcastic a moment ago. I had no intention of insulting anyone, but perhaps myself. So I was nervous but not about the game but rather how I was looking playing it. I was trying to be obvious about my obliviousness. Yet that just serve to distract me from focusing too much on the game itself. For that I was kind of on autopilot. I just paid enough attention to make sure I moved the pieces as they legally can more. Otherwise, I was just randomly reacting to it being my turn. I was guessing all the way. But my guesswork was somehow working. Somehow I managed to play the true master out to a draw after an extraordinary 2 hour 45 minute match. I was late getting to dinner that night, but it was easily worth it. I have this story to tell the rest of my life. I just wish I could have let those men in the park in on it, but I couldn’t, I just couldn’t, because . . .

I made the whole thing up—all of the prior paragraph is pure fiction. Okay, maybe not my best effort, but consider that I did that free flow with basically no stopping point. I put it down as fast as my fingers could type it straight as my mind generated it just now. All I needed was a prompt, which I spun up as I typed the paragraph before it.

I’ve been to NYC numerous times including on manager due diligence, well, more of a conference, at Goldman Sachs. I’ve been to Washington Park a number of times loving it including watching the men play chess, who I am not in any position to challenge. I love the movie When Harry Met Sally, but I have no connection to it aside from knowing every line from watching it as many times as I have. I do make A LOT of sarcastic remarks, which sometimes get me into jams and sometimes get taken quite literally. But that is just the data upon with my AI mind has been trained. All I needed was a prompt. Then the bullshit can flow.

Let me give you anther example of my superpower.

In MBA school I was in a marketing class the second semester of first year. Our final project was a paper that if I remember correctly was to make a marketing proposal which encompassed all the major concepts we had covered that semester.

Many in the class were stressed over this. I was not. As a business undergraduate, I had taken quite a few courses within this realm including marketing. Marketing was a subject I held if fairly low regard due to what I felt was its vagueness and lack of numeracy. Let me be very clear: I was arrogant in MBA school in general and unsufferable in regards to “lesser” subjects like marketing—and I was very wrong to be that way. My apologies for my cringy behavior and attitude at the time. I’ve changed my view on marketing in particular.

Still, at the time I knew this paper was not going to be hard. In fact I supposed I could just sit down and write it and do okay. No need for trips to the library for research. No need for an outline, rough drafts, numerous edits, or coercing a friend into proof reading it. I would just wing it. I would bullshit.

My paper, which you can see here, was a proposal to build a casino resort in Hawaii.

I did choose a subject I knew a bit about. I had been a long time admirer and stockholder in what was at the time Circus Circus Enterprises having fallen in love with Las Vegas and this company specifically ever since my first trip there at 11 years old. It was the first stock I ever bought. While I had never been to Hawaii, I had watched every episode of Magnum, P.I., Hawaii Five-0, and Gilligan’s Island as well as knowing inside and out the episode where The Brady Bunch goes to Hawaii—I’ve always feared tiki idols as a result.

It took me about three hours to write the 18-page paper. I did it basically straight through in one pass with minimal backing up between sections. I gave it another pass through to check for overall flow, grammar, and consistency. I think some of my peers spent days on theirs. I got the highest grade in the class—95/100.

No, I don’t believe I am an AI as I have a number of differences.

• I do know quite a bit about how I think.

• I do actually know, truly know, a lot of stuff. At least at a superficial level in most cases and at a fairly deep level in quite a few cases.

• I do know when I’m making mistakes or at least when there’s a good likelihood that I’ve made mistakes.

• I can self correct and edit without the help of others. (I welcome the help of others; I am just saying AI requires it.)

Still, I have sympathy. We share a currency. I offer some grace.