Telia, a Swedish telecom company provides fiber for costumers, private and corporate. They will connect you to the fiber-net and if there is none connected to the property they will dig a hole, place an optic cable there and connect you to the net.
Costumers contact Telia to find out if they need to pay an extra single payment for the digging or if it is covered by the monthly fee. The customers are generally more inclined to sign a contract if there is an extra single payment involved. So, this needs to be calculated fast. Is there or is there not an extra single payment involved?
Administrators working with this assignment do their calculations manually, it is slow and for every assignment they complete, 2-3 new have been added. In the long run, Telia wants to cut down the number of employees in this department, only having people handle difficult cases and supervising an AI. Everything else is to be handled by the AI named Glen.
My assignment was to design a simple user interface where the user will input an order number and the output will be the single payment amount. Either 0 indicating, no extra single payment, and the amount the customer would need to pay. Also, Glen will present the facts that were significant when deciding the amount. That way the user can provide feedback to Glen and help him improve his predictions.
One of the major challenges was to decide whether the users needed to understand that the software they used “was an AI”. If they were to know, how much of the AI function would need to be displayed and to what extent would the users be able to control it.
The users of Glen are young, energic, playful. Many of them are students working part-time. They expressed their desire for an easy and playful system that is fun to use. They wanted the AI to be personalized, viewing him as a person. The current systems they currency used were expressed as “boring and slow”.
Based on that, I decided that I would be clear in the design that they were communicating with an AI, and that the AI needed their help to learn and make better predictions. The users were excited about working with an AI, “Glen the robot” as they called him, and they were looking forward with enthusiasm on helping Glen by reporting bugs and miscalculations.