How to Overcome Algorithm Aversion in 2021
Algorithms help humans make better decisions. Yet one survey found that as few as 13% trust AI. Algorithm aversion is the study of why humans reject AI. Tools that help humans build trust in algorithms will be more important in 2021.
In Overcoming Algorithm Aversion, the authors* shed light on methods to overcome algorithm aversion. One study allowed participants to adjust algorithms. That interaction, even just a small amount, dramatically increased their willingness to use its output. Another study showed increased trust when participants were allowed to intervene when they suspected the algorithm was wrong.
It’s perhaps an obvious yet profound lesson: trust in AI builds with contact and collaboration.
Tools can help bridge the divide. One example is embedded AI. Embedded AI puts a human interface on algorithms like the steering wheel and dashboard of your car. Michelle Lacy from Bayer Precision Agriculture calls it “putting the algorithm behind the button.”
Here’s an example of embedded AI in Spotfire. The map displays hundreds of thousands of data points. The human operator chooses an area on the map and sets evaluation criteria. Like your car’s dashboard, the screen lights up with AI-powered predictions. We’ve put the domain expert behind the steering wheel and dashboard of the algorithm.
Embedded AI is just one way to help humans reach out and touch AI. Others include Model Ops, Automation Analytics, and Dynamic Learning.
Embedded AI is like algorithm aversion therapy. Explore it in 2021!
FOOTNOTES:
* Dietvorst, B. J., Simmons, J. P., & Massey, C. (2016). Overcoming Algorithm Aversion: People will Use Imperfect Algorithms If They Can (Even Slightly) Modify Them. Management Science, 64 (3), 1155-1170. http://dx.doi.org/10.1287/mnsc.2016.2643
