A new impact report from the School of Continuous Improvement follows five years of automated systems taking over everyday decisions --- fare-gate overrides, self-checkout flags, call-centre routing --- and measures how much staff and the public came to trust them.
A five-year impact report tracks trust in automated decisions from transit gates to call-centre queues, and finds notice alone rarely closes the gap
The School of Continuous Improvement has spent five years asking a question most organisations never formally get around to: once a system starts making a call a person used to make, who is keeping score on whether anyone still trusts it. A new impact report, covering 2021 to 2026, sets out what the School found across five such handovers, from a transit authority’s fare gates to a cooperative research consortium’s retired trust scores.
Five years in, we are more confident about how to measure a handover than about what to do once we’ve measured it. That is not the conclusion we expected to publish, and it is the one the evidence gave us.
— Associate Professor Casimir Beng, Lead of the Adaptive Metrics Lab
Compiled from partnerships with a metropolitan transit authority, a big-four professional services firm, a regional university network, a Nordic evaluation agency, and a cooperative research consortium, the report tracks what happened once each partner let an automated system take over a decision it used to make by hand — an override at a fare gate, a flag on a self-checkout screen, a call routed away from a person. In several of the five areas, staff trust the report measured six months after a handover still depended heavily on how much warning the system’s arrival was given, and how specific that warning had been.
Dr Solveig Adeyemi, who led the report’s call-centre work, said the finding “complicates the easy version of this story, where more warning is always better. Warning about the wrong thing turned out to be almost as unhelpful as no warning at all.”
The report’s method, in outline, pairs each partner’s own measure of trust or performance against an independently gathered figure the School collected without reference to the partner’s dashboard — the same discipline that has anchored the School’s evaluation-ecosystem agenda since its earliest work on the Living Dashboard and the Indicator Commons.
The full impact report is available from the University’s research repository under an open licence, doi:10.5555/slop.4fo82q.