I’ve been thinking a lot lately about the user energy and excitement that is going into AI-based chat interfaces. I think we all understand pretty quickly why these experiences feel so compelling, but at some level, it’s curious: we live in a world with lots of rich media: tons of video, games, images, etc. Why are we so drawn back to the “retro” experience of plain (not even formatted) text?
I think it’s because, since the actual dawn of the computer industry, without really realizing it, we’ve been trying to talk to computers directly. It helps to think of them almost as this Star Trek style “alien species” we encountered, that can only communicate in binary arithmetic. There was clearly something compelling and rich there, but it saw the world in a very different way than we did, and we needed to figure out how to communicate across that gap.
So, we’ve spent a hundred years trying to develop a common language so we could speak directly to this new “species”. We started with binary codes and math, logic gates, things like that. Then we moved to assembly, then programming languages that looked a bit like our own languages but were structured enough that the machines could understand them, with some help from translators that we called compilers, and later, interpreters (hmmm). The languages got more abstract and more sophisticated. Layers kept getting built that let us express more complex ideas more naturally (in theory, at least).
And now we are getting really close! We can’t quite get the machines to do everything we want just by talking to them, but we can get them to do a lot. We can speak fairly naturally, and they sometimes seem to understand. There are still errors in understanding and miscommunications (hallucinations). But these chat experiences feel much closer to “just talking to” the computer. A non-programmer can express a complex intent to a computer with a decent chance of getting what they want, without having to “write code”.
We have a long way to go still, and many problems left to solve. But it seems now that this is what we have always been pushing for, in some way: the ability to “just tell the computer what to do, and it will do it”. That’s why “plain text” is exciting again.
This is SO cool