You know what the missing piece of the puzzle was? It was that AI art isn’t for artists.

Like, it can be. But for the most part it isn’t.

Okay so here’s a “thread” I wrote back in October, in the pre-ChatGPT era, right when everyone was freaking out about AI art, and I’m pulling it out for a third time to add some commentary on which of my thoughts have shifted and which I still think are relevant.

Ever since I decided to get myself involved in academic research two years ago, I’ve spent a lot of time thinking about the habits and behaviours of the different researchers I’ve interacted with. Most researchers are crazy, but they’re crazy in different ways because their actions are driven by different forces.

Understanding your particular brand of influences and craziness as a researcher and comparing it with those of others may help you to understand why your collaborations with some researchers work very well and why others drive you insane. Therefore, I hereby present to you the “Researcher Alignment Matrix”.

Last year, I wrote a blog post about why I think half-hour meetings suck. You should read it first before reading this one – it’s short. Also, I still stand by a lot of it.

The tl;dr is that most things that require meetings actually take 15 minutes, 45 minutes, or an hour, and that almost everything else shouldn’t be a meeting in the first place.

I am now going to elaborate and add a whole bunch of caveats.

I’m starting to think that maybe essays and poems are two sides of the same coin, where they’re both attempts to explain things, but an essay is what you get when you follow a thread through to the end, and a poem is what you get when you can’t quite isolate the thread and you don’t know where it ends.

I think fiction has a slightly different purpose. It’s for the things that need to be expressed but probably shouldn’t be explained outright.

I’m becoming increasingly concerned by this trend where we’re letting these AI companies convince us that we can use these systems to streamline our lives by flattening the complexity out of them, as if our problems are a standard dataset we can summarize our way out of instead of this massive nightmare jumble of causes and effects we don’t always see and that we definitely do not understand.

And I find it paradoxical how the word “context” has become so associated with the use of LLMs when context is precisely the thing that they are stripping out of our lives. Knowledge isn’t supposed to exist in a vacuum; it’s supposed to be developed through being in conversation with other people and their needs and with the real world and with other knowledge, and it is through the specificity gained through that kind of conversation that information becomes more meaningful. We act as if this act of seeking and internalizing context is grunt work that should be offloaded toa machine in order to make room for the real work, when it is, in fact, the real work.