This is the point where AI will change the world – it’s probably not what you think

Disclaimer: I’ll start off with a short piece of computer history. But bear with me. It might be worth it: there will be screaming goats, toilet paper and Apples. Here goes.

A little less than one year after I was born, in 1975, the first personal computers were introduced. With underwhelming (well, they were probably considered ‘hot’ back then) names like the MITS Altair 8800 or the IMSAI 8080, these machines opened a door to a new and exciting era.

What many people forget (or never knew, because they just weren’t around), was that these first computers were sold in “kits”.

B(uild)YOD

Today, you buy a PC, plug in it and “Tadaa!”, you have access to the world, an endless supply of screaming goat videos as well as all the applications you need. Back in the olden days, you had to assemble and build your own device. Even the legendary Apple I still needed some assembling, though it was a lot simpler in use than for instance the Altair 8800 which “had to be extended with separate hardware to allow connection to a computer terminal or a teletypewriter machine” (thanks Wikipedia!). ‘All’ you needed to add to the Apple I was a keyboard, power supply, an enclosure to the assembled motherboard and you were good to go. Needless to say that only the tech savvy people and Nerdiest of Nerds (who will have my undying respect until I die) bought these seminal types of PC kits.

It was the beginning of one of the biggest shifts in our history. But in that form, it was also not enough to make a dent in the universe. Not enough to set the network effects in motion. And certainly not enough to change consumer behavior, or the market.

The Apple II moment

The real tipping point came in the form of the Apple II. People tend to gush about how beautifully it was designed, about its affordable price or the fact that it had color graphics. But the real revolution was that the Apple II was not a collection of components. It was a completely finished product. You pretty much took it out of the box and plugged it in. That’s why it became the first successfully mass-produced microcomputer. Because you did not need to understand how it worked in order to use it.

I believe that artificial intelligence is still quietly residing in that Apple I moment, waiting to make that giant Apple II leap to change society in ways we cannot fathom. Yes, we all do “use” algorithms – they are in our chatbots, in the recommendations of Netflix, in our GPS, - but we don’t “use use” them. What I mean by this dreadful oversimplification is that machine learning, deep learning and neural networks aren’t consumer products. We for instance can't use them to create and personalize our very own AI assistant. In fact, it would probably be more correct to state that it's the algorithms consuming us instead of the other way around. (Thank you very much Facebook and consorts.)

The toilet paper principle

About I year ago, I learned about the ‘toilet paper principle’ from Tim Harford: “once a technology is cheap enough to wipe your bottom with, it’s cheap enough to change the world”. I thought that was bloody brilliant. We tend to indeed be biased about the intelligence and sophistication of technologies and as such overestimate their impact in the now: Roy Amara correctly taught us that “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.”. That is perfectly applicable to AI today: we tend to overestimate what it can do in this moment (oh yes, it can do great things, like beat the world's top Go-player, but we're still a long way from general artificial intelligence) but - once it will become cheap and consumer-usable (for lack of a better word) - we have no idea how big the impact will be.

So in order to have a technology create an actual seismic shift is society - like the PC or the internet - you need a completely underwhelming context of simplicity and economy of price, on top of an incredibly powerful technology:

1. Affordability to individuals, not just companies and experts - a toilet paper moment
2. Child-proof simplicity in usage - an Apple II moment

Like I said, the former 2 points tend to be easily overlooked when we are blinded by the power, potential and intelligence of a technology, which is exactly what is happening now. I think that’s probably why so many people are disappointed in AI, because we were not yet able to translate its potential into actual impact. We’re definitely not there yet. But when this “cheap ’n easy” moment arrives, that will be AI’s real tipping point.