A toxic hellstew of wrong predictions

As an industry, we've never once called a major trend ahead of its arrival. Why do we keep on trying to predict things?
Back to Blog

Back in the day, way before I started working, I remember saying to a good friend of mine that the internet “was ridiculous” and that “there was no way that it would catch on”. I’m still embarrassed about that, and nearly thirty years on that same friend still reminds me of what I said.

Part of what we do as technologists is try and work out what the next trends are going to be. Why we do this baffles me. We seem to look as an industry way off into the future, whereas what’s important in the here and now – what our customers want, and what our peers are having immediate success or not with, is much more important.

As an industry, we get most long-term predictions wrong. Or we overegg things to the point where we should see the ridiculousness of it. We never see the things that are actually going to work. People laughed at the iPad when it was released. None of Apple’s competitors thought the iPhone would be disruptive. No one predicted Snapchat, or Twitter, or even Facebook. Same with Bitcoin, AWS, Uber, Airbnb, and so on. No major success story of the past 30 years has had the red carpet rolled out for it, because none of us knew it was coming.

What’s keyed me into writing this post is – in particular – the endless noise around AI, AR, and IoT.

Each of these ideas has grains of probability within. For IoT, it’s true to say that the ability to put some network-connected intelligence in previously “dumb” devices is getting cheaper. But if my kettle breaks, do I think, “Oh great, now I can buy that IoT kettle that I’ve been waiting for an excuse to buy?” No. Because normal people don’t think like that. They want an electric cylinder for heating water, not a… what would an IoT kettle even do? Skype you when it needs descaling? You can see IoT as much as you like – no one gives a damn.

Anyway, IoT isn’t the biggest delinquent in this trio.

AR. Yep, OK, pretty helpful for seeing how an IKEA chair might look in my lounge. Is it helpful for anything else? At work or at home? Not really.

If AR takes off it won’t be because every hacker geek and their dog is out there building pitch decks about it. It’ll be because someone builds some new version of some Snapchat-type thing that everyone will love and use and the founders will end up running around trying to cash in on whilst in the meantime it’ll have like 98% market share and Facebook will try and copy it.

It doesn’t matter than Apple and Google, and even Microsoft is obsessed with this stuff. Great, put it in the OS if you like. No one cares.

So now let’s think about AI. Artificial Intelligence. Intelligence. It’s not though.

There are some branches of computer science where AI is really AI. Things like speech and image recognition. But some of what we now call AI is nothing more than pattern matching. And its pattern matching that is usually ridiculous. If I go shopping for a lawnmower and buy it, every marketing algorithm that’s latched onto me that thinks I’m now some sort of obsessive lawnmower collector. Or how about Google Maps that, despite me having a pattern to how I travel each week, nine times out of ten can’t properly predict the next leg of my journey.

Or even how most of the software that we built doesn’t need AI. Why do I need AI to post an invoice to Sage, or show the details of your upcoming hotel booking, or show you a list of shows on at the Brighton Fringe, or take a copy of this post and automatically post it on Medium?

I do think that in 20-30 years the labour landscape will be changed forever by AI. There is a strong argument that we can automate a lot of knowledge worker-type activities in the same way that 200 years ago we managed to automate a lot of manual worker-type activities. That’s a risk to anyone who wants a job, or anyone who’s trying to raise kids who might like to make a living when they’re in their 30s, but for now, what we’re calling AI in 2017 – it’s little more than a comedy of errors. Some clockwork simulation of a brain that does little more than tick through a few tricks in rote.

I can’t see this changing. As an industry, I think we’ve completely lost the plot, and are utterly obsessed with the idea that we know what “the next big thing” will be. We ignore what customers tell us, and all we have to do is listen to what works for them.