AI IRL Podcast Episode 39: AI Makes Users Make Better Decisions
In the discussion of AI, we don’t talk about taking a user-centric approach nearly enough.
We get caught up in technology for technology’s sake or an implementation perspective or solving business problems.
Instead, we need to be asking: Are we delivering a better user experience through AI?
“One of the key components is how do we build these systems to be in service of people rather than the other way around?” Chris said.
“How are we doing a better job of making decisions as time goes on?” —Chris Butler
The Right Way to Approach Design
You need to make sure that the outcome or the system that you’re building takes everybody into account.
One problem: As we start to build teams to build these products, we limit the number of people in the room.
But that just limits the conversation.
“You also need to include in designers, product people, management and even customers or the people that are in some way impacted by your customer’s actions” he said.
“We’re no longer talking about one human being in front of one computer just typing away by themselves,” he added.
It’s an ecosystem that includes the whole organization and even the world at large. As machines are meant to do something for human beings, we shouldn’t leave humans out of the loop.
(Most of the time when we talk about machine learning doing something all on its own, there’s still some human impact somewhere.)
So, the right way to approach design?
Humans should always be at the center of the discussion.
“We’re no longer talking about one human being in front of one computer just typing away by themselves.” —Chris Butler
Getting More People Involved
AI designers need to acknowledge not only that more people need to be involved in the decision making process, but that AI needs to make decisions that always have human benefit in mind.
User-Centric Design Questions
- How do you deal with errors?
- How do you make sure the user understand the system well enough to make the right decisions?
- How does this tool fit within the larger experience for people?
“Things that are really important are not always easily measurable, but there is some way that you can either find leading or following indicators that can help you do a better job of including that in the decision-making process,” Chris said.
In one of Chris’s jobs, if he’d just built in a confidence score from the algorithm to measure machine learning effectiveness, people wouldn’t understand a) what that meant or b) the ecosystem-wide implications of the machine learning.
That’s what you have to test before you start collecting the data to build out the algorithm correct.
Which takes a ton of people on the design end — not just designers, but, you know, actual users.
“One of the key components is how do we build these systems to be in service of people rather than the other way around?” —Chris Butler
AI Can Help Identify Biases (for Better Decision Making)
AI isn’t so great at holding conversations — looking at you, chatbots.
But it is great at analyzing patterns that aren’t obvious on the surface.
“That’s where we’re starting to see where these types of highly targeted dashboard system models fit in,” Chris said.
“It’s not just about the idea of how do I present this information to someone but how do I work with the biases or the cognitive simplifications that human beings do on a regular basis to provide a better dashboard that tries to help me be a better decision maker?”
Spending design time on the user.
By involving more people in the design conversation.
To design a dashboard that allows users to make better decisions.
“Human beings are building tools that then help us make better decisions in some way,” Chris said.
“How are we doing a better job of making decisions as time goes on?” he said.
If you don’t use iTunes, you can also find every episode here.