How and to what extent do user value talking to AI?

A tool to measure the value proposition of AI dialogues

Based on the value terminology of Value-based Engineering and ISO/IEC/IEEE 24748-7000 we have investigated the value of AI and AI-based conversations. AI dialogues are value bearers that trigger a lot of emotional value reactions in users. Users may benefit from utility values generated by conversations if the AI information is true, if the AI is clear, attentive and provides useful information. But the AI is also judged based on its social acceptability. So values like coherence, sensitivity and politeness are playing a role.

We empirically investigated the values generated by a large public sector employment agency AI and developed what we call a “value monitor” (see figure 1). The monitor allows AI providers to see exactly where their technology stands in terms of ‘value proposition’. AI operators – based on this monitor – can see where they stand in terms of quality-of-service. They can improve their AI solution with the help of the monitor and trace advancements in the technology they provide.

Project participants

Lenart Motnikar

Alexander Kagan

Franziska Mayrhofer

Sarah Spiekermann

Aschauer IT