I drive a car with a manual transmission. When people ask me why, I explain it like this: With a manual, I tell the car what to do. With an automatic, I make a request of the car, it considers it, and then it responds as best it can. However, the way it responds may not be what I want. For example, if I want to accelerate, maybe I want to drop to a lower gear first because I know that I need the extra burst. A car with an automatic transmission will generally assume I want to go faster in my current gear until it sees no, I really mean business.

Using a computer to generate these images is similar. Since I'm communicating in English, I can be more clear about what I want, but the machine may misinterpret at best, or refuse at worst. In this case, the machine refused. I'm not thrilled about how this comic turned out, because it's not showing what I want. The baby is supposed to be attempting to breastfeed on the tablet, not just putting its face close to the screen. Hopefully you understood that without my needing to be explicit. The title probably helped, also.

In order to even get what's here, I had to attempt a few different prompts with ChatGPT before I could get a baby and an tablet in the crib. Then I had to use a different model (Grok) to move the baby closer. ChatGPT knew what I was trying to do and wouldn't let me. Grok was a little more ok with it, but it still wouldn't remove or even reduce the amount of clothing on the display. What you see here is the best I could do.

I do not like the future where tools that are becoming more and more necessary impose limits on what we're allowed to do with them. I fear a consequence where people limit their thoughts in order to be able to simply function in the world. If that is the endgame, then the PC Police have won. I hope for a future where computing power is such that the margin between proprietary and open-source models is marginal, and we are liberated once again.

- Management