Corporate AI pilots are multiplying faster than leaders can manage

Companies rushing to adopt artificial intelligence are creating a new management problem: too many pilots and too little focus. Executives at Fortune’s Brainstorm Tech conference in Aspen said the spread of AI experiments can slow companies down when projects lack clear business value.

Bret Greenstein, chief AI officer at consultancy West Monroe, said corporate leaders often know a small number of priorities will affect performance, while teams across the business may be pursuing hundreds of AI use cases. Greenstein said at the roundtable that those ideas are not all equally important, even if each has internal support.

The discussion centered on the gap between AI enthusiasm and execution. Managers and workers have broadly heard that AI should become part of their jobs, according to the participants, but that message can lead teams to start projects that do not get completed or do not serve a larger strategy.

Sean Bruich, senior vice president and chief technology officer at Amgen, said companies should give more attention to the business and internal politics of AI adoption, rather than treating the work mainly as a technical challenge. Bruich, who described himself as a data scientist, said a surge of pilots can weigh on a company because each project can come with its own sponsor, team, metrics and data engineering support.

Bruich said the volume of pilots can make it harder to see which AI efforts produce business results. He also said some companies move too slowly to shut down efforts that are not working.

Dan Gill, chief product officer at Carvana, said his company has faced similar lessons in a business that handles vehicle transactions between millions of buyers and sellers and runs a large financing operation. Gill said completing one priority is more valuable than pushing several projects only partway along.

Gill said AI has lowered the cost of building prototypes, writing documentation and generating code. That ease, he said, can tempt companies to advance many ideas at once instead of finishing and refining the most valuable work.

Nizar Trigui, chief technology officer at logistics company GXO, said AI efforts need direction from business leaders, not only technology teams. Trigui said boards, CEOs and executive teams must be involved in the work if companies want AI adoption to change how the business operates.

The executives said that shift can be difficult because older measures of leadership success still shape behavior inside companies. They pointed to headcount, spreadsheets and performance metrics as examples of signals that may conflict with a sharper approach to AI adoption.

Despite those tensions, Gill said companies no longer have a realistic path back to the pre-AI workplace. He said the future remains promising, but the discussion made clear that companies will need to choose fewer projects, finish more of them and cut the ones that do not deliver.

This story draws on original reporting from Fortune.