I’ve been on the record multiple times marveling at just how big of strides generative AI has been making since it first burst on the scene. Yes, I realize that “intelligence” is somewhat debatable with ChatGPT, but I also begin to wonder if it really makes much of a difference. It’s certainly possible that generative AI is just the first big step toward actual AI.
But.
I’ve been having a few additional thoughts the past while that have given me pause. Not enough pause to think this is all nothing more than a flash in the pan, but enough pause to wonder just how smooth the road ahead will be. It’s easy to look at how fast AI is going and assume it’ll be nothing but up from here on out. Assumptions like that rarely play out well, and I’m tending to think it’ll be the same this time.
First off, generative AI takes a ton of energy to create. It requires a veritable boatload of processing power to figure out what it’s going to say to any particular query. (Kind of ironic to see it take that much energy for mundane conversations a human can have using just a bowl of Froot Loops that morning.) Setting aside the fact that this is bad for the environment (because honestly, when has that stopped any business so far?), it also means that it takes an awful lot of money to keep it going. Right now, AI is swimming in money. Everyone is falling all over themselves to start a new AI venture or invest in some new AI application. So the money problem isn’t an issue yet, but unless AI starts producing tangible dollars in return on those investments, then that money’s going to go away sooner or later. Likely sooner.
(This adds further argument for my concern AI will quickly turn into some sort of ad-supported money grab, the same as the rest of the internet.)
My second concern is connected to the first: all those AI companies popping up all over the place. Cory Doctorow noted a few months ago that the existence of all those companies are very indicative of a bubble, and I don’t think he’s wrong. Everyone and their brother is starting a new AI company, despite the fact that I just don’t see that many actual applications for AI at the moment. I see the potential, yes, but not the actual results. Why would I need an AI subscription just for doing X when almost all the AI I’ve seen does just about anything? Yes, perhaps X is a tad more specialized, but I highly doubt anyone is going to be willing to pay more than a few companies for AI anything. Not people off the street, at least. Perhaps companies will have subscriptions to things, but again, that remains to be seen.
So far, AI does a passable job at writing passable text. It makes passable images. When you’re bad at writing or drawing, it’s a huge leg up over where you used to be, but it’s still nowhere near where people who are actually good at either one of those are already. The more I read text by AI, the more tired I become of it. Google’s now using AI to answer search queries, arguing that there’s so many garbage websites out there, it’s better for AI to just summarize it for you. This is a huge red flag, as Google was already monkeying around with the results. It no longer exists to help you find answers. It exists to make money. AI will let it hide how its answering that much more effectively, letting it make even more money. Yikes.
Does this mean AI is dead in the water? I definitely don’t think so. Processing power gets cheaper every year as it gets more and more advanced. So that cost will go down, and if there’s enough money devoted to it, it might go down more quickly. But I’m going to need to start seeing some real applications for the technology. Applications where it actually does a good job instead of just a passable one. Right now it’s like we found a talking gorilla, and we’re all impressed by how the gorilla can talk at all that no one’s really asking themselves why they should listen to a talking gorilla instead of an actual expert on something. Yes, talking gorillas are cool, but I don’t know that I’d take the advice of one until it proved it was a smart talking gorilla.
I think it would be foolish to ignore AI altogether. Yes, there are real obstacles in the way of it and true success, but there are also very big potential gains if it ends up succeeding. Knowing what you’re doing with it and how to use it well could be key to success in the future. One way or another, I see it impacting our society. Maybe this version won’t catch on, but even if it doesn’t, the cat’s out of the bag, and people will keep trying until something sticks. It’s scary not to know what that something will be, but it’s also tremendously exciting.
Kind of sums up life in general these days.