I’ve been spending a lot of time thinking about how we interact with computers to complete tasks. I believe that we will see a big shift in user interfaces over the next decade.

Historically changes in user interfaces have been led by big technology shifts, and I believe we are at the cusp of one of them. Here’s why.

A mental model for user interfaces

I like to think of user interfaces as: inputs, transfers and outputs.

First, you have inputs. A person needs to get information into a machine. For example, using a keyboard to draft a document.

Second, you have the output of that information. It’s presented to the user in the form of: text, a picture, a video, audio etc.

And in between, you have the transfer of information. There’s not much to say here because it isn’t visual. But, the capabilities available for the transfer of information have an impact on the other stuff. For example, if bandwidth were limited it makes little sense to input or output using video.

Greatest hits of UI

We’re going to look at some notable shifts in UI history. I’ll use the past to explain why I think the future might look different.

Punch cards

In the 60s/70s, information was input into computers using punchcards.

This was either done by hand (painful) or using a machine. The IBM 029, released in 1964, was one of the best punch card machines on the market. Computing power was limited at the time so this was considered okay. There were two inefficiencies though: first, it took time, and second, if you made a mistake you needed to start from scratch.

Command line interface (CLI)

The next innovation of note was the command line interface (CLI).

It had a request-response style, i.e. you make a request by typing something in and receive output in return. It was faster to get information. But the real game changer was that it allowed the user to make edits as they go. We take this for granted today but imagine what that felt like.

CLIs, while they look different today, are still used by programmers for a bunch of tasks.

Graphical user interface (GUI)

As computing power increased, we moved to graphical user interfaces. The Xerox Star was amongst the first.

The big change here was accessibility and increasing output. Being able to visualise information allowed you do more with computers. A graphical interface made computers mode accessible: it was easier to get information in.

Touch screens

Over the last two decades, touch screens have become other way for users to interact with data.

Lots of good stuff here: with mobile, you had the ability to consume and generate on the go. Touch screens allowed a bunch of actions that we didn’t have before: the swipe left / right, enlarge a picture, refresh your app like a slot machine.

The intent of this brief history lesson was to point out that major changes in UI are a result of technological advances. An increase in computing allowed us to stop punching holes and use graphical user interfaces. The ability to carry a computer in our pocket opened up a new input: touch screens. AR / VR allow you to input information using actions or movements.

Why I think UI will change

Over the next decade, AI will make inroads into most applications. Generative AI is all the buzz and is no doubt hyped, but let’s use a 10 year time frame to de-risk.

My general view is that we get true value from AI when it becomes action oriented. If we assume that AI becomes powerful enough to help us with tasks, user interfaces are going to look very different.

In my view, point and click takes a backseat. Text and audio become the main stay for interacting with computers.

Here’s why. Consider booking a holiday, which usually consists of the following:

1. Decide where to go

2. Check with your friends or family to coordinate schedules

3. Prepare a short list of schedules

4. Circulate to your group

5. Agree on the schedule

6. Book the tickets

7. Send the tickets to everyone

8. Get paid from everyone

There are two ways to achieve your outcome today. Use some combination of Skyscanner, Whatsapp and Monzo to achieve all the above. This is the point and click version.

There’s also the automated route. Your friend, an enterprising engineer, decides to automate the task using code.

Now, let’s imagine doing this in a world where AI is smart enough to take actions. Over the next 10 years, it may well be the case that we have a virtual AI assistant to help us with tasks. A Siri on steroids that can actually go out and do things for us. In this scenario, whilst point and click will exist, it won’t be the primary form of input.

It’s easier to describe in text or audio because your input can be richer: “Find me a holiday with XYZ for a maximum budget of $500. Text XYZ with options and ask them to approve. Once I approve the holiday, send everyone a payment request for their share.”

A little valley before the peak

My sense is that it will feel like we’re taking a step back before we take many steps forward. I’ve already felt this with a few conversations already.

The first reaction when you describe ChatGPT to someone is: how is this different from Google? Then you show it to them and they are like “Wow, this can replace my doctor”. Then they keep playing with it and hopefully realise that it can hallucinate.

The beauty of where we are at is that there is a lot of white space to design and craft experiences. It’s bit like when mobile applications first became a thing and we had to design them from scratch. Lot’s of interesting problems to think about and I’m excited for the innovation in this space.