Trying to further collect my thoughts on a presentation I gave about my experiences designing smart applications. The kind that use lots of math and lots of data to make recommendations to humans. The kind that learns from our behavior and adjusts those recommendations. The kind that can serve as a companion.
In my several years worth of designing big-data-driven projects I have consistently seen people treating the machines like humans. There is a dynamic between users and these systems that cannot be ignored or denied. Well, it could, but I’ve seen strong evidence that as digital experiences get smarter they become more human. This has a very large impact on the trust we place in these systems and that trust is tantamount to their adoption.
We treat machines like they are humans. Our devices are a social actors in our lives. The research performed by Clifford Nass has been influential in forming the presentation of my observations. My working hypothesis is that we treat machines as humans by default and adjust our expectations from there. Our human interactions set the bar for our expectations with digital experiences and if we think through capabilities as if they were taking place between two humans we can solve a lot of the design problems we are, and will be facing as systems get smarter.
I also always have a presentation given by Chris Fahey in the back of my mind. It was one of those “yeah, that’s what i was saying – but you said it so much more clearly!” kind of moments. I have actually implemented the approach and have lots of stories to tell and conversations to explore.
As a start I have explored seven human traits of a successful digital companion that can serve as guidance for designing these systems. If it were indeed a person, these are some traits of that person I have seen in successful systems.
- Observant/Smart
- Confident
- Transparent
- Humble
- Reliable
- Personable
- and a dash of awkward (a trait that truly endears me to these systems)
From a design perspective, there are many examples where having this “human” bar for the behavior of the system has helped come up with solutions. I’ve seen tactical examples, like the best way to implement a “please wait” spinner interaction. I’ve also seen strategic examples where we’ve set the bar for an entire experience on interactions and conversations between an advisor and a client.
A boy and his penguin 