Is AI really paying off? The hidden gap between adoption & RoI

Organizations are investing heavily in AI with the promise of faster work, better collaboration, and higher productivity. From automating routine tasks to accelerating knowledge work, AI is being positioned as the ultimate efficiency booster. But as adoption scales across teams, a tougher question is emerging. Are companies actually getting the Return on Investment they expected, or are the gains being quietly eroded along the way?

Are organizations getting the production they thought they would with AI adoption?

AI is a Fast Work Colleague

As per an Atlassian report on AI adoption, a remarkable 92% of Indian professionals agree that AI will improve the speed and quality of their team’s work. Globally, about 27% of work week is wasted searching for information each year within a Fortune 500, as more than 17 hours per week is spent tracking down information. This is something AI can really help with.

87% of respondents believe AI is helping them show up as a better teammate. 51% of knowledge workers believe they could work faster if their teammates used AI more.

“The future of work is here, but to fully exploit AI, businesses must invest in the tools and training needed to unlock productivity, collaboration, and innovation.” — Dr. Molly Sands, Head of Teamwork Lab, Atlassian

Dr. Molly Sands, Head of Teamwork Lab, Atlassian commented, “Team silos hinder collaboration, create inefficiencies, and lose opportunities. Without alignment, achieving objectives and adapting to market changes becomes difficult. To remain competitive, companies need to prioritise transparency and collaboration. Connecting information and integrating AI are essential: organised information allows AI to reveal key information, enabling teams to find what they need, at the right time. The future of work is here, but to fully exploit AI, businesses must invest in the tools and training needed to unlock productivity, collaboration, and innovation.”

The observation is that while organizations are rushing to deploy AI, they are failing to capture real ROI, because they’re prioritizing speed over quality. Recent data from Workday shows 40% of AI time savings are lost to correcting errors and rewriting low-quality outputs. Also, 77% of daily AI users spend just as much time (if not more) reviewing AI work as they do human work. And, 66% of leaders call skills training a priority, but only 37% of the employees struggling with the most rework actually receive it.

“Too many AI tools push the hard questions of trust, accuracy, and repeatability back onto individual users,” — Gerrit Kazmaier, President, Product and Technology, Workday

“Too many AI tools push the hard questions of trust, accuracy, and repeatability back onto individual users,” said Gerrit Kazmaier, President, Product and Technology, Workday.

New Thing to Deal With: Workslop

And it’s not like AI is doing a flawless job. Productivity might be up, but the output is, at times, far from desired. New research from BetterUp Labs and Stanford’s Social Media Lab calls it AI workslop. While AI?generated output looks polished with slides, summaries, reports, even code, it lacks substance, hence transferring the necessary effort from the sender to the receiver.

The research found, in a survey of 1,150 American full?time employees, 40% received workslop in a month with only around 15% not needing any change.

And there is a cost. Averaging that around two hours are required to redo the work, the research authors put it at around US$186 per employee per month, i.e., over US$9 million a year for a 10,000?person company. Moreover, the sender of such slop earns a rep as an untrustworthy worker, which corrodes collaboration and adoption of AI.

Shifting the Effort Rather than Eliminating it

AI is undeniably speeding up work, but speed alone doesn’t guarantee value. When time saved is lost to error correction, rework, and low-quality outputs, the ROI story becomes far more complicated.

The rise of “AI workslop” shows that without proper training, governance, and quality standards, AI can simply shift effort rather than eliminate it. To unlock real returns, organizations must treat AI not as a shortcut, but as a capability, one that requires clean information, skilled users, and accountability. Otherwise, the promise of AI productivity risks turning into an expensive illusion.