I was at a great AI & human centered design conference session at CES 2020 yesterday.
Takeaways:
š¦¾ āIntelligence Augmentationā is the future. AI augmenting human intelligence, not autonomous AI. This should be the main case & fully autonomous AI the edge case. Especially for mission critical applications.
š Itās critical for people to understand how AI is coming up with predictions. Users need to have a reasonable mental model for how algorithms work. [I think this is a good/key design principle in general]
āļø AI will fail. There must be mechanisms in the system to help the user understand why itās failing and provide an override (e.g. this was an issue with the Boeing 737 Max)
š Algorithms require constant updates. Outside stimuli can change (e.g. if lane marker colors change in one small town, driving AI will need to be updated)
š AI needs to know the userās context. (E.g. pausing notifications when driving) [I expect this is going to continue to be a huge area of focus for AI and smart product design, lots of room to grow here]
Dan McMillen 