For the longest time, I was sure these LLM chatbots were the next crypto grift: ideas that had existed for decades, implemented in the worst possible way, but with a lick of paint and a shiny marketing campaign, designed to separate the gullible from their money. The best strategy, I thought, was to sit it out, watch others lose their shirts, and wait for it all to blow over.

But I may have to go back a little further to see history repeating.

I think now that the current LLM craze is this generation’s dotcom bubble.

You see, the internet in general was a transformational technology, but it was the web that built on top of it with something tangible: it felt like we had productised it. We had the internet in a box now, and all that was left to come up with clever ideas, package it, and become rich.

So along came the speculators with their late 90s ideas of ideas that would revolutionise humanity! They all shared two common features (besides being in cyberspace):

1. Some high-level grand idea that sounded great in an elevator pitch; and
2. A reliance on some assumption that just was not (yet) true

Whether it was that people would be comfortable ordering clothes sight unseen, or typing credit card details into online forms, or watching movies on tiny screens, or listening to music on crappy speakers, or waiting three days to download software that they’d be faster driving to the store to get, the ideas all raised huge sums of money by focusing on the hype, and glossing over the real problems that needed to be solved.

And a lot of people bought into the hype. And they spent a lot of money. And then it all blew up and people lost their jobs, or their savings, or both. And they were angry, felt cheated, felt lost, and struggled to see a path forward. Many wrote the web off as a fad.

So what’s the analogue now? Just like the early dotcom companies, there’s a lot of easy hype money to be made by selling the future, then cashing out when people realise it isn’t here yet. It’s also easy to roll our eyes at how ridiculous the idea is that an LLM could do whatever it is that we’re being told it can do, or get angry at how expensive it all is to run, at how wasteful and energy hungry the technology is, at how immoral the training data is, at how dystopian the disruption of artists and writers and programmers and musicians and actors will be.

All of this is true. But look back at the list of dotcom problems above, and the funny thing is that all those problems eventually got solved so comprehensively that they seem positively archaic now. Back in 2000, boo.com was the poster child for how ridiculous it all was. As if people would ever buy clothes online! Hah.

Maybe the stakes are higher now, but LLMs, or whatever comes next, will soon be able to do the things they can’t do now. And like the web, that will both change everything and change nothing at all.

What I do know, though, is that the people who navigated the dotcom bubble to stay relevant were the ones who saw that whatever happened, the toothpaste was out of the tube. The web was here to stay, so they rolled up their sleeves and started working to address those problems, to build the future, rather than laugh at it or yell at it. I suspect the same will be true today — we can try to put the toothpaste back in the tube, or be angry at the ones who squeezed it out.

Or maybe we can roll up our sleeves once again, figure out how to take control of the technology again, to use it to build the future we want. Maybe it’s impossible, or if it is possible it won’t last, but since when was that a reason not to try anyway?