For AI Productivity Gains, Let Team Leaders Write the Rules

Corporations are rushing to invest in artificial intelligence. But corporatewide AI policies alone can’t transform work, according to our recent research. For AI efforts to pay off, executives must set clear guardrails while enabling teams to write the rules that make tool adoption real. Only then will AI deliver the meaningful returns that organizational leaders are pursuing.

In the productivity classes one of us (Robert) teaches at MIT, many executives are keen to learn about AI but unsure about how current corporate rules would let them use the new technology. This led to the design of a survey, taken by 348 business professionals — including senior executives, line managers, and team members from various industries — about who gets to make AI rules and how AI rules actually function inside their companies. The findings are clear-cut: Businesses have been overlooking the team dimensions of AI governance and implementation.

A striking 72% of the participants in the survey said that corporate headquarters should establish overall AI guidelines, but individual teams must set their own rules within those boundaries.

The Trouble With Centralization

Some organizations have responded to the AI craze by creating AI czars or centers of excellence, issuing corporate mandates, or establishing executive task forces. While these moves may feel decisive, they seldom change how real work gets done.

Centralization tends to clog the arteries. Teams wait for approval, workarounds proliferate, and some employees play it safe by circumventing AI productivity tools instead of harnessing them.

We’ve seen the challenges of new productivity tools before. However, the arrival of the internet didn’t lead to “internet departments,” and no one ever asked permission to open a spreadsheet. AI should be governed the same way: with principled guardrails at the corporate level and practical rules created at the team level.

AI tools have the capability to write, summarize, and analyze with speed. But AI cannot grasp context, apply judgment, or eliminate error. Responsibilities vary greatly by function, of course: The finance department is different from customer service or site operations. Because judgment is local, rules must be local and set by team leaders.