Sunday, February 26, 2023

More attempts to fix the world


I am just remembering that in the early years of clerkmanifesto I would sometimes write late night missives from out of the cold, dark cement of my home's basement. Now I sit in a tall chair, high above the city, and it is warm in my aerie. But the fundamentals are all very much the same:

How do I let the Internet know what it's supposed to do now?

When I was in middle school in the 70's my friend Larry Davidson, whose father was a Doctor, got a computer, like a Tandy TRS-80 or something. It was fascinating and slightly boring, which is just like computers now! (Oh how far we've come without moving at all!). Larry had a briefly compelling game that was in every way a precursor to Chat GPT and our assorted current AI's. This presented itself as a kind of computer therapist. One could type in a question or sentence, and the computer would respond.

This really is a true story from the late seventies, at least as far as I can remember it.

I might type something like:

"Are you a truly sentient robot construct?"

And the computer would reply "Would you like to talk about that?"

I'd say "I'd like to know your answer is all."

"Does this remind you of your mother?" The computer might reply.

The reason this game wasn't fun for very long was because it soon became clear that the computer only had 25 or so comments in its wheelhouse.

The current iteration of this game, known perhaps inaccurately as AI, has billions of comments in its wheelhouse, so one can play with it longer. But in the end the same realization dawns: It's all canned.

There is no AI yet. 

A computer can still only say what people say, but not what a person says.

So I am here trying to tell all the computers what to say. Every night, from down under the earth or high above in the clouds, I am trying to tell the Internet what to say.

So far it hasn't listened.



  1. I remember that "game" you mention. I think the database of responses was created by Carl Rogers. [Only partly joking.]

    I also remember having a "game" based on Hitch Hiker's Guide to the Galaxy. " It was all text-based, of course, so you had to type in what you wanted to do." It was maddening in its computer-ness. "Put on my robe." "Can't, it's hanging on the hook." "Take my robe from the hook." And on and on. I had a cheat book that came with a pen like a yellow highlighter, so I could find the "tricks" that let me keep going.
    And at about the same time I was writing a computer-based course on parenting for Control Data. They used a branching structure: "Your daughter is crying. What do you want to do?" It was three levels with three choices at each level. Itreally was quite forward for its time (1970-ish).

    1. That actually sounds cutting edge compared to that "game" I wrote about. It is funny that computers have come this vast, almost unfathomable distance, and yet the thing you were trying to create there, even with the advent of these new AI's, still seems, of all things computers dreamed of, out of reach.


If you were wondering, yes, you should comment. Not only does it remind me that I must write in intelligible English because someone is actually reading what I write, but it is also a pleasure for me since I am interested in anything you have to say.

I respond to pretty much every comment. It's like a free personalized blog post!

One last detail: If you are commenting on a post more than two weeks old I have to go in and approve it. It's sort of a spam protection device. Also, rarely, a comment will go to spam on its own. Give either of those a day or two and your comment will show up on the blog.