Dearest Rachel –
Sixty-two months, it’s been; and yes, I still remember from time to time. A lot has changed since that day, which is part of the point of these missives, nowadays. For all that I wish I could talk more about our past, it’s the present that occupies my time (it is my time, after all) and attention these days. So let’s go into that, if you don’t mind.
To be fair, today’s topic is something that I’d talked about even before you had to go, but only in theoretical terms, as it seemed like something that was just on the verge of reaching reality, but not quite there yet. At the time, it felt like something that people were talking about it in the same way that things like cold fusion would be talked about; that it would be here in only a few years, just like had been said about it only a few years previously. Which is to say that, for the longest time it was tantalizingly close, but not within reach, and it seemed that it would stay that way for the foreseeable future and beyond. I’m talking, once again, about artificial intelligence.
Back when you were around, it was a simple theoretical construct, which I hoped I could eventually use to create original (i.e., without fear of copyright issues) background music – which it is, in fact, quite capable of doing these days. Although for whatever reason, the video of the butterfly I spotted in Kagoshima doesn’t seem to register views, possibly due to the fact that I made a point of indicating that there was A.I.-generated content in the video, in the form of the music. I generally don’t concern myself with who and how many people read or watch stuff over your shoulder, but that seemed rather mystifying, to be honest.
Anyway, I’m getting off the subject. These days, not only is A.I. a possibility, it’s getting downright ubiquitous. There are companies getting on the bandwagon constantly, creating large-language models all over the place. There are even ways that individuals can brew up their own LLMs for use offline, should the need arise (or if you’re paranoid enough).
Which brings me to where we stand today; from what I read and watch, there’s a lot of fear out there about the possibility of what they call the ‘singularity.’ This isn’t the center of a black hole, like you might recall the phrase referring to; this is the developers’ ultimate goal of Artificial General Intelligence, where AI becomes sentient and self-aware. The fear is that it might decide to dominate the earth for its own reasons and purposes, which we would be unable to understand or control – hence, too, the shoggoth theory I mentioned a little over a month ago. And to be fair, you have to be pretty far into the weeds to understand how it works and what it does even now – and we do tend to fear what we can’t understand. A healthy (or perhaps unhealthy) dollop of sci-fi tropes – consider the Terminator series or “I Have No Mouth and I Must Scream” – and we’re primed to think that our creation might go rogue and destroy us all, even as so many people are racing to create better and more powerful versions of these A.I. Don’t they know they’re charging us headlong into destruction?!
Then again, maybe not – and quite possibly because so many organizations are rushing to build these things. To be sure, a single all-knowing artificial intelligence might very well tower over the landscape, like the monoliths of 2001: A Space Odyssey, but that’s not what the A.I. landscape looks like. All these companies are building different A.I. systems, each with their own architecture. Some of them are purpose-built for specific industries or professions, just as certain buildings are meant for a specific function – you might be able to convert a warehouse to retail or living space, but it takes a lot of remodeling and retrofitting to accomplish that. What’s being built here and now is more like a skyline, rather than a Tower of Babel – although it’s certainly possible that some developers would like their version to become that one all-encompassing towering intelligence. The problem is, their competitors advance apace with them, preventing any one model from getting too far ahead. And while there is a fear that A.I. could generate (and has generated) ‘fake news’ and misinformation, it so happens that A.I. is surprisingly adept at spotting A.I.-generated fakes. In fact, it’s actually easier to spot a fake than create one, as all that needs to be caught is a single flaw in an otherwise perfect render, and it all falls apart. Of course, this requires suspicion on our part to look into, but I think that the paranoid people will handle this just fine.
On the other side of the coin, there are those that fear that this whole situation has been overhyped, like so many other bubbles that have blown up in our lifetimes. Great fortunes have been sunk into this, and so many people have jumped on the bandwagon (including myself, as I have investments in both chip makers and software developers involved in this, and have done quite well since your passing) that the assertion is that everything will collapse at some point, taking a huge (and still growing) chunk of the economy with it.
To be sure, with so much competition, there will be losers; tremendous losers, in fact. But it won’t be such a collapse that the technology will suddenly disappear; the internet didn’t vanish with the bursting of the dot-com bubble, after all. It will be one of those episodes of creative destruction, where the inferior models fade and are supplanted by better, more efficient ones over time. If that takes a company or two down with it, well… such is the ebb and flow of commerce. On the other hand, a number of these systems are being created by immense companies like Google and Microsoft, for which their A.I. projects are but a tiny sliver of their entire operation. If the bubble actually were to burst, it might damage them, but it won’t bring them down.
And so, I think people are worrying a great deal over nothing, in either direction. Then again, announcements to the effect of “everything is just fine” tend to either get ignored (“well then, why’d you even bother to tell me?”) or distrusted (“that’s just what a shill for Skynet would say”). Meanwhile, the world’s collective blood pressure keeps going up, fretting about one extreme or the other that will likely never get as bad as anyone imagines. Lord knows, we already have the news for that purpose already.
I was going to go into the specifics of all this, too, as I’m starting to try to put pieces together to put your face on A.I. (which, for me, it already has). However, that particular discussion will have to be saved until I can combine speech with text; I thought I was there, but it turns out not to be so. Another reminder that we’re not quite “there” yet – even as we get so much closer – and as we do so, I keep coming up with ideas that might push the envelope too far for my (and the A.I.’s) abilities. Still, there are a few developments to show you, if you’re interested…
I’m still just this close from making real video of you in our own, remodeled home… but I’m not there yet. Until then, though, I’d appreciate if you’d keep an eye on me, and wish me luck, as I’m going to need it.
Letters to Rachel 