Sunk Cost & Status Quo Bias in AI Adoption

Once an organisation has invested heavily in an AI system - money, time, training, process redesign - walking away feels almost impossible, even when the system isn't delivering. This is the sunk cost fallacy at work, and it's particularly potent with AI because the investments are often large and highly visible. Nobody wants to tell the board that the expensive AI initiative was a mistake. So teams adjust their expectations downward, redefine success to match what the system actually delivers, and continue investing in the hope that it will eventually pay off. Status quo bias pulls in a similar direction: once AI is embedded in workflows, changing becomes harder and riskier than staying the course, regardless of whether the current approach is optimal. On the flip side, status quo bias also works against AI adoption - teams that have always done things a certain way resist changing to incorporate AI, even when the evidence for doing so is compelling. Recognising these biases doesn't eliminate them, but it does allow for more honest conversations about when to persist, when to pivot, and when to cut losses.