As as an IBM Distinguished Designer for the past 21 years, my current mission is to drive the development of IBM’s point of view on the practice of AI design.
When I first started on this type of work, I was told, “Figure out what it means to design for AI.” That’s it. It wasn’t a question about any particular brand—just, as a designer working with AI, what does that mean?
For something to be artificially intelligent, it has to have the basic building blocks for us to create meaningful lasting relationships between the human and the machine.
I knew enough to be dangerous—not in a good way.
Human-Centered Design and AI
During the first year of that work, I talked to as many people as I could, both inside and outside the company. I came to realize a handful of things.
- It is so easy for people to say, “Let’s put AI on this,” without understanding how much work needs to be done for that to be possible.
- Traditionally, if you are designing for digital regardless of discipline, you’re making artifacts so that a front-end developer can build whatever it is you’re working to build. When you are designing for AI, you are really making the artifacts for a data scientist, which means your data literacy has to grow exponentially.
- I have found the human missing from a lot of conversations about AI. It’s all focused on if we can make AI do what we want it to do. What was missing was a conversation about the human needs around AI.
For a system to be considered artificially intelligent, it has to be able to understand, learn, reason and interact. These are pillars of cognition, but these are the reasons we choose to build meaningful and lasting relationships from a human-to-human perspective.
So, for something to be artificially intelligent, it has to have the basic building blocks for us to create meaningful lasting relationships between the human and the machine.
The response isn’t programmed in a deterministic way but in a way that is trying to understand what we’re asking. And that’s the biggest and the best scenario for human-centered design.
Applying Ethics to AI Design
A large part of our work is centered on the ethics of designing for AI. This is something that designers generally don’t get trained in (which we all should).
There are ethics in terms of how we introduce bias through our actual design solutions in many mundane ways that have nothing to do with AI.
Putting the human at the center is so important.
The potential problems multiply when we get into artificial intelligence. Having people at the company who think deeply about this is important to me because of all the problems AI might represent in the future. I think there will always be bad actors, and some are made to be malicious.
However, there are many situations where the bias was unintended. It just happened because someone didn’t pay attention to the details or didn’t have some sort of framework to think about the solution.
This is why putting the human at the center is so important. Our position at IBM is that AI is to augment humans, not to replace them.
That, attitudinally coupled with practices that you should and could follow as a part of the design process, helps to keep that ethical conversation not just in the forefront but a part of the fabric of everything that we do.
* * *
For more on creating meaningful human-machine relationships, watch Adam’s Ted Talk below:
