Your team's AI productivity just hit a wall
BCG and the University of California, Riverside surveyed 1,488 workers and found a threshold most companies have already crossed. At one or two AI tools used simultaneously, workers report real productivity gains. At three, diminishing returns. At four or more, self-reported productivity collapsed.
The researchers call it "AI brain fry." 14% of workers reported it. They weren't the disengaged ones. They were the high performers. Marketing departments hit 26%. The pattern tracked with oversight burden: the more outputs you had to check, verify, and manage across tools, the worse it got.
Why It Matters
The average organization now runs seven AI tools. In 2023, it was two. Goldman Sachs reported in March that there is no meaningful relationship between productivity and AI adoption at the economy-wide level. ActivTrak found that daily task time increased 27-346% since AI adoption. Email time alone doubled.
Self-reported productivity has its limits. People can feel busier without being more effective. But when behavioral data from ActivTrak and self-reports from BCG point in the same direction, it's hard to dismiss.
Meanwhile, 66% of CEOs are freezing or cutting hiring while AI budgets climb 44% year over year. Companies are removing people and adding tools at the same time. If the tools are creating cognitive load instead of reducing it, the math breaks in both directions.
The Decision
Are you measuring how many AI tools your teams use, or how many they can actually absorb before the cost exceeds the gain?
What To Do This Week
- Ask three team leads one question: "How many AI tools does your team use daily?" Compare the answer to what IT provisioned. The gap is your real adoption number.
- Pick the department with the most tools and ask: "Which one would you drop?" The answer reveals which tools are overhead, not leverage.
- Count how many hours per week your team spends checking, verifying, or fixing AI outputs. That is your hidden cost of adoption.
What Not To Do
Don't measure AI success by the number of tools deployed. License count is not capability. Don't add a fifth tool before understanding what four are doing to your people. And don't assume the quiet ones are fine. The BCG study found the workers most affected were the ones least likely to report it.
Signal Boost
BCG x UC Riverside: AI Brain Fry Study - Read the section on oversight burden. That's where the real cost hides for companies scaling AI across departments.
