Designing for understandable predictive intelligence
Weekly triggered thought
This week's "triggered thought" comes from a personal experience. It's about a conversation I had regarding my parking situation. The sophisticated parking space was built almost 20 years ago. It's a system where you drive your car onto a pallet, get out, and then use your badge to have the system move your car underground.
Last weekend, a situation occurred when I called for my car using the service to call in advance, but someone else arrived at the garage and used their responder to collect their car without the software service. This made me think about how AI could influence such systems in mundane situations.
For example, a more intelligent system could recognize when someone has called in person and prioritize their car as it is closer to the exit, even if someone else had placed the call first.
It's similar to modern elevator systems in skyscrapers, where you choose your destination floor before entering the elevator. This allows the system to optimize everyone's journey based on the floor it travels to, disconnecting your actions from the behavior of the automated system.
This can be even more intelligent. It uses not only the current situation of the different elevators but also patterns from the past and predictive knowledge from similar elevator systems. Then, we have predictive relations that create strategies for operating based not only on the context but also on expected behavior.
The behavior of the thing (elevator, parking automat) is disconnected from the user's control. This raises interesting questions about user experience and trust in AI systems. How do we design these systems to balance efficiency, user understanding, and intentions? We will need conversational AI interfaces that can explain decisions, offer the possibility of driving, and create tangible moments of interaction.
This weekly “Triggered Thought” is written as part of the Target is New newsletter, which offers an overview of captured news from the week, paper for the week, and interesting events. Find the full newsletter here.
About the author; Iskander is particularly interested in digital-physical interactions and a focus on human-tech intelligence co-performance. He chairs the Cities of Things foundation and is one of the organizers of ThingsCon. Target is New is his “practice for making sense of unpredictable futures in human-AI partnerships”.