Why I’m writing this down

I’ve been more excited about technology four times in my life.

The first was finding out computers existed. The second was the day one came home. I was not a good student for most of my schooling, not for lack of ability, I just wasn’t one, and my parents set two conditions. I had to do exceptionally well in the national exam, and I had to choose between a full printed set of Encyclopaedia Britannica and a family computer. I wanted the Britannica badly. A salesman had done his job on me, and I read far more then than I do now. I got the highest available grade in every subject, and I still don’t know whether that was because of the computer or a combination of things. But the choice itself was never a choice. It was an Intel 286 desktop PC, a beige box under a CRT monitor with one megabyte of RAM, which is a few thousand times less than the phone in your pocket. It wasn’t close.

The third was reading about the internet in the Straits Times, Singapore’s main newspaper, which arrived at our apartment in Brunei along with the local one. It ran recurring pieces on what the internet was and what you could do with it, and I read them almost every day for years before I could touch it. For a while it also carried a column by Bill Gates answering readers’ questions. In one of them he said that anyone who could get through Donald Knuth’s The Art of Computer Programming should send him a résumé. I didn’t know who Knuth was, but I remembered the name. I knew what the internet was long before I knew what it felt like.

The fourth was finally getting online, in New Zealand, on a dial-up account I split with a schoolmate across the road. Neither of our parents knew what the internet was. Only one of us could be connected at a time, so logging in kicked the other off, and so did my mum picking up the phone.

Then for a long time nothing came close. There were interesting things and useful things, a whole career’s worth of them, but not that feeling. This is the fifth.

I didn’t have to make myself learn about AI. I couldn’t stop. That’s the honest version of why I’ve spent the last stretch of time going as deep as I can, into training runs and model internals and the maths I skipped the first time round, knowing full well that absorbing all of it is impossible and that making the effort anyway is a different thing from absorbing it.

There’s a second reason, and I’d rather name it than pretend it isn’t there. I’ve always wanted to build something of my own that’s worth paying for. That hasn’t gone away, and it never does. The pull toward the technology and the pull toward making something out of it are two separate things that feed each other, and neither is the real one.

Looking back at that first choice, I think I was already picking the thing you make answers with over the thing that contains them. I’m making the same choice now. So that’s what I’m doing: building things with this technology, and writing down what happens.

I want to be clear about what this isn’t. It isn’t a success story, because there isn’t one yet. I’m not going to tell you what I’ve learnt about building AI products, because so far I’ve mostly learnt what doesn’t work, and not enough of it to be useful. What I can do is keep a dated record of what I tried, what broke and what I changed my mind about, and let it accumulate into something.

Everything here will have a date on it, because what I think today is not what I’ll think in a year. When I change my mind I’ll say so, in the open, next to the thing I used to think. Let’s see what happens.