musings on ai

Humans make mistakes. Computers don't.

When you look at something and realize: a human made this. A hand carved wooden butter knife. A child's drawing. An heirloom piece of furniture. An old masters painting. Even the most mass produced things: someone designed that Ikea furniture. One person, or several, sewed every seam in the t-shirt you're wearing. Countless people were involved in designing and manufacturing and assembling every piece that goes into your car.

The touch of the human hand is inherently valuable. Someone took time out of their day to make something. Maybe because it brought them joy, or maybe just because they needed to pay rent. They may have done the same thing wrong before, made mistakes, and learned.

The practice, the skill, the effort that it takes to have an idea and use your hands to shape it. To use your mouth to speak it. To watch someone else do something and learn how to do it better.

Every single thing made could be imbued with that value, if the onus wasn't instead to always make more. An ever expanding mass of output too big to be consumed. Every kind of thing you could ever imagine. And you should collect it all.

I want to say something about Luddites, about the fact that we can no longer weave as fine fabric as we could 200 years ago. We've lost that knowledge. I want to say something about the value of learning, of passing on knowledge, about trying even though you might fail.

I'm very afraid of failing, of falling behind, of being found out as an impostor, of asking questions. But that's a self-perpetuating cycle. Not asking questions when I don't understand, not trying because I might fail - that robs me of the chance to learn. To progress. I fall behind because I'm scared of going the wrong way when I go forward.

But the computer does make mistakes. Because otherwise it doesn't seem human. It hallucinates by design, because otherwise it would produce the same result every time - the most statistically probable.

And not even the most statistically probable is necessarily the truth - or right, or correct.

I want to work with my coworkers. To solve problems together. I want to do something valuable.

I don't want to tell an AI how to do my job and then wait impatiently as it approximates what a human would produce. I don't want to be 10x as effective when that doesn't result in more money in my pocket or more free time.

And I especially don't want it when I already have a good life and nothing is being done for people who are struggling.