Me and My AI Ugh Field

Teasing apart conflicting career feelings 😩

Illustration of “The Turk,” a fake chess-playing automaton of the 18th century, from a book by Joseph Friedrich Freiherr von Racknitz trying to explain the illusion (1789)

In an attempt to make sense of the hot chaos one might call “my career,” I’m hoping to attend Mila, a well-known AI research institute in my city, for a graduate degree next fall. I’ve though this over for a long time and have a lot of reasons to believe it’s a smart move. But… there is a but. Which I need to dissect and tease apart, so I’ll use this blog post to do just that.

To write this, I’m actually repurposing a draft post from a few months ago, when I was working on my application to Mila. The draft started like this:

I’m applying to a graduate program in machine learning. This post is an attempt to coax myself into finishing the application, which is due at the end of this week.

The post went nowhere, and I ended up writing the application without needing that coaxing, thanks to the magic of ✨ deadlines ✨. But still, it is interesting that I thought I might need it. And now I have resurrected the draft to serve a similar purpose: force myself to revise machine learning notions so that I am well-prepared for an interview with a professor next week.

Considering that you’re reading about this writing process and not about questions such as “what is an eigenvalue” and “how does stochastic gradient descent differ from standard gradient descent,” I haven’t been successful at that goal either.

The truth is, whenever I decide to write about AI in a serious way, I enter an “ugh field” — a more or less subconscious flinching away from the task at hand (and towards procrastination, which these days takes the form of playing the farming simulation game Stardew Valley). Despite AI being my chosen field, and the one I work in every day at my current day job, there’s a strong force that makes me want to think about literally anything else when I’m my own time, including my writing time. As a result, I write about AI only very occasionally, and when I do, it’s fun, whimsical topics like the relationship between AI and the ✨ emoji, not stochastic gradient descent.

Is this worrying?

Maybe?

I mean, if I truly hated machine learning and then got stuck doing a two-year degree in it, that would be a pretty tragic mistake. But I don’t hate machine learning, otherwise I would have chosen a different career long ago and we wouldn’t be here today. It seems quite likely that once I’m in the program, once I’m surrounded by smart interested people and have clear projects to focus on, I’ll enjoy it. I’ve enjoyed similar situations in the past. I enjoy my current job. Well, I enjoy it as far as work goes —

— which may provide our first clue. Work, it seems, has a certain propensity to generate ugh fields. Even when we pick careers we like, it’s easy to end up being annoyed by the work we need to do. Vanishingly few of us wake up excited to spend the day working at something that also makes them money. More commonly, the excitement, if it occurs at all, gradually emerge as we dive down into a task and get into flow. Quite often we notice it only after it’s over — to have worked is a better feeling than to be going to work.

And then of course work provides a number of other useful things, such as money and a sense of purpose, which may, on their own, make ugh fields a lesser evil. It’s pretty clear, after all, that a strategy of avoiding ugh fields as much as we can would be a tragic mistake, too. In my case right now it would mean playing Stardew Valley all day and end up quite sorry for myself.

So, I’ll probably enjoy the Mila program and I’ll probably be happy to have done it. Yet, knowing this, the ugh field isn’t getting thinner. I’m looking at this webpage that explains eigenvalues and eigenvectors — and although I have learned what those things are at least three times, and it should therefore be easy to learn them just once more, I cannot muster the energy to dive into linear algebra again.

I cannot ever imagine myself writing a blog post about linear algebra, either (which is probably a good thing for you, dear reader).

I suppose one reason to have a blog is precisely that I can write about whatever I feel excited about, without worrying about whether someone needs me to, or whether it can make me some money. Nobody is ever going to pay me to write about worldbuilding or the perception of beauty. At least, I don’t think so, except, indirectly, the few generous souls among you who somehow support this blog with a paid subscription. In this sense, not writing about AI, or anything close to my current line of work, might be the expected result. Of course those are going to generate an ugh field, and ugh fields don’t write good posts any more than anger does.

Yet, with all that said, I have a feeling that the best careers are built on top of things we’re excited about. Things we can’t help think about or work on even if we’re not paid. If AI doesn’t excite me enough right now that I am compelled to write about it (or start a side project, etc.), I have a nagging feeling that might be bad. I’ll never have the same kind of success as the person who is obsessed by AI all day and night. Then again, are there really people like that? I don’t know. I guess there are, but also that they’re quite uncommon? And the rest of us, who tend to be obsessed by pointless fun things, like video games or TV shows or sports or partisan politics or blog posts about aesthetics, need careers too.

I’m running writing around in circles, it seems. I can’t decide what to think about this. In this sort of circumstance, I suppose the best strategy is to throw oneself ahead. It’s not like I have other great career plans I’d be renouncing anyway.

I will attend Mila (if they want me, anyway), and I’ll become more of a machine learning specialist — yet at the same time, I don’t expect you, dear reader, to learn much about whatever I end up doing. This article generated enough of an ugh field as it is, and I’m happy to be six words away from finishing it.

Another engraving from the same book