Blog: Designing for AI
3 tips to design AI and Automation Technologies that people want to use
AI Experiences can be scary
I work closely with people who are designing Automated and Artificially Intelligent services, and they are acutely aware that the people they interact with are scared that a computer is going to steal their job. In fact, a large part of their job is assuring people that this is not the case.
As Designers, we should help to reinforce this idea, by providing ways of thinking and working that can guide design decisions about how these types of experiences show off the positive aspects of Automation, and remove the negative aspects. This article takes you through 3 useful principles to apply.
Technology should serve us
As someone who has been involved in making products and services for the last 20 years, I am a firm believer that life is what you make it, especially so when it comes to technology. Technology should serve us and not the other way round.
So when I hear phrases from colleagues who work closely with Artificial Intelligence and Automation technologies that “we’re in the 4th industrial revolution”, it got me thinking about how we can make this new revolution work harder for us than in previous “revolutions” (it worked really well for people who owned the means of production, but not so much everyone else).
I got a dose of inspiration recently when I was fortunate to attend and be asked to speak at the excellent Service Design Summit in Barcelona. It was here that I heard some thought-provoking words from Dina Krumstroh of the TUI Travel Group and Olli Mannerkoski of Nokia about ways to balance the needs of people alongside design choices one has to make when designing for AI and Automation products and services.
After reflecting on Dina and Olli’s talks, I thought it would be useful to share my 3 key takeaways for designing for AI —
- #1 Let humans take credit for helping another human being
- #2 Give me options, not orders
- #3 Nobody wants more UI
Design Tip #1 — Let humans take the credit
When it comes to the benefits of AI, and why it is a good thing, an argument I hear regularly is that AI and Automation will free us up from the boring mundane tasks that nobody wants to do, leaving us to live a life of leisure.
This sounds great in theory, but in practice, imagine you’ve worked with your colleagues to identify a whole bunch of repetitive and mindless tasks — what kind of principles can you then use to prioritise what receives attention first? what factors are considered desirable for people using this new, intelligent automated mode of working?
One useful principle is to avoid automating those tasks where we can directly take credit from another person for helping them.
Why? because we place more value on succeeding in tasks where another human says “thanks for helping me” compared to simply completing a task successfully on a computer.
So, given that we spend most of our days in 3 modes —
- Mode 1 — in our own heads (thinking)
- Mode 2 — interacting with other people (human-to-human)
- Mode 3 — interacting with a digital device (human-to-CPU)
It’s clear that Mode 3 is likely to be the most desirable and feasible area to consider for automation, provided that there’s nothing that Humans can take credit for. This will leave automation to take care of the bad/boring/repetitive stuff within a human-to-machine scenario, leaving people to handle and take the credit for the human-to-human stuff, creating a self-reinforcing loop.
Design Tip #2 — Give me options, not orders
My children do not like being told what to do, especially when they are tired and hungry. But give them some options, or the illusion of options eg. “do you want to switch off the TV yourself and have some dinner? or would you like me to help you switch it off so you can have dinner sooner?” and their perception changes completely.
The same principle can be applied to interacting with an algorithm that is suggesting a course of action, a bit like you might find when someone interacts with a chatbot.
People are more likely to react favourably if they are given options to consider, especially if those options are calculated and offered based on thousands or millions of data points.
Conversely, if those data points are used to provide you with a single option, you run a greater risk of people rejecting the lack of control that has been imposed on them and reinforcing negative attitudes to AI and automation.
So when you’re designing anything that can offer options, aim to offer options that put the control into hands of the person driving the experience.
Design Tip #3 — Nobody wants more UI
User Interfaces can sometimes be a thing of beauty, a potent mixture of imagery, words and ideas that can transport people to a magical place that they weren’t expecting. But no matter how brilliant a user interface is, it’s the thing at the end of the journey they take you on that is most valuable.
If you’re able to get people from A to B faster and make the steps to get there easier and less mundane, then you’re onto a winner.
So my final tip is to focus your design efforts to see at how you might use Artificial Intelligence and Automation to simplify your User Interfaces.