Eighty percent of employees say artificial intelligence makes them more productive, but only 37% of organisations see any measurable impact on earnings before interest and taxes, according to survey results cited in a Fortune commentary by Louisa Loran. Nearly two-thirds of organisations, the piece notes, have not captured any of the value employees report feeling. The gap, Loran argues, is not a shortage of data. It is a shortage of willingness to change how decisions are made.
The commentary points to a range of examples. In one aviation partnership, an optimisation solver was built to account for crew, aircraft, passengers and ground control at once. Adoption took about 18 months, moving from roughly half of the solver's recommendations being followed to nearly all of them. Controllers resisted the tool at first, and the airline's effort to expand it across its network stalled because of entrenched hierarchies. Loran notes that many operational constraints had hardened into fixed rules over time, and that those rules were upheld by people rather than by any technical necessity.
Consumer goods company Diageo offers a different illustration. According to the piece, the company combined sell-out figures, consumer sentiment, media activity and weather data by brand and by country to guide decisions. Even so, incentives and bonus plans continued to determine which targets people prioritised, showing that tools alone do not reshape behaviour when reward structures remain unchanged.
Workforce anxiety also plays a role. Loran writes that employees fear AI will be used to cut headcount, and that few can imagine what new roles might emerge from it. She cites a survey finding that 69% of executives use AI tools for less than an hour a week, a figure that suggests how shallowly many leaders have engaged with the technology themselves.
Shipping group Maersk is cited as a case of sustained transformation. The company was unprofitable in 2016 and reported an operating profit of $3.5 billion in 2025. Loran also describes an executive assistant whose most valuable contribution was not routine task automation but the human work of aligning colleagues around shared priorities.
To close the gap, Loran recommends four moves: start with ambition rather than tools, audit the constraints that shape operations, tie pay to shared results, and have leaders visibly model the change they are asking for. In her account, the companies that benefit from AI will be those willing to revisit long-standing assumptions, not simply those that deploy the most software.