The Disposable Interface

In the last few months, thousands of tools have been popping up left and right promising to do a specific thing easier and faster using the help of AI.

AI to help you write your emails. AI to help you create images. AI to generate websites for you.

I want to focus on the last example - AI to generate websites - because it reminds me of something.

Steve Jobs famously introduced the iPhone by listing its core features: “An iPod. A phone. And an internet communicator.”

When he went through this list of features, the iPod and the phone received huge cheers from the crowd. “Internet communicator”, however, only got a polite applause.

Merging an iPod with a phone was exciting for people back then because these were two existing, familiar concepts. People knew what phones and iPods were, that both were useful, and that combining them into one product sounds great.

But they could hardly imagine what the advantage of a constant internet connection in your pocket would mean, beyond quick access to their emails maybe.

The current swarm of AI helper tools are like the “iPod and a phone”. They use a new piece of technology to make todays common tasks easier and cheaper.

The really interesting stuff, however, will only happen once we realise what this new piece of technology enables for the first time. Things that we could not do before.

AI’s “internet communicator”.

Traditionally, software was defined by the cost and time necessary to build it. To create a piece of software, you had (still have) to hire and pay engineers, designers, etc. for many months or even years. And to fund the effort, you design that software so it could be used by as many people as possible which would hopefully somehow pay for using it.

The result of these constraints is todays software landscape. Apps that stay mostly the same for years and try to cater to the needs and preferences of a wide range of people.

In a future with advanced AI, the cost and time necessary to create new software will be dramatically lower, which can have the following two consequences for how we think about interfaces in the first place:

Disposable

If software can be generated within seconds, we might quickly create it to accomplish a certain task and then throw it away again once we’re done. Why keep it if you don’t need it anymore and it costs nothing to recreate it once you need it again?

A good comparison is taking photos. When cameras where invented, taking photos was costly and time intensive. And not everyone could do it.You would dress up nicely and hire a professional photographer who would set everything up with their expensive equipment and take your photo. Taking multiple photos would have been even more expensive and take even more time, so people focussed on taking very few but very good photographs of themselves which were primarily meant to be shown to other people. So we ended up with a lot of identical looking portraits.

That’s exactly how software has traditionally been created.

Nowadays, everyone has a camera in their pocket at all times and taking a photo is as easy as pressing a button. Because of that, we now think about taking photos very differently.

People take multiple selfies, keep the ones they like the most, and delete the rest. They have become disposable, because the time and cost to take a photo has been reduced to zero.

The same will happen to creating software.

Personal

If software can be generated within seconds at no cost, we also don’t need a large user base anymore to fund its creation. That means that we will primarily create software not for other people, but ourselves.

To circle back to the photography comparison:

A large amount of the photos we take every day are meant for ourselves, and not to be shared with the public. We even found new applications for it, like taking a photo of someones phone number because it’s quicker than to type it in, or sending a quick photo of that brand of cheese to your partner to check if it’s the one they asked you to buy.

Just like photography, AI generated software will be very limited at first and not be nearly as good as software created by professionals. And maybe we will still depend on these professionals to create high quality software for us for quite a while.

But over time the results will get better and better, until they becomes good enough to cover most needs of most people and it will seem archaic that we ever had to pay and wait for other humans to create that piece of software we so desperately needed.