Widespread AI Use in US Workplaces Yields Low Business Returns Amid Quality Concerns

According to Gallup data, the proportion of US workers using AI at least occasionally jumped from 21% in mid-2023 to 50% by early 2026. Despite this rapid adoption, MIT researchers report that only 5% of organizations currently see measurable business returns on their AI investments. This gap highlights managerial challenges in integrating AI without establishing clear work norms and organizational culture. The main obstacle is not AI quality but how it is actually used.

This failure has given rise to a problematic organizational phenomenon called "Workslop," identified by behavioral research center BetterUp Labs and Stanford's Social Media Lab. The term derives from "AI Slop," a 2025 Merriam-Webster word of the year describing generic, recycled AI content flooding social media. Workslop refers to superficially polished but low-quality AI-generated outputs, such as reports or presentations that appear professional but lack substance. A BetterUp study of over 1,000 employees found 40% encountered Workslop products in the past month, mostly in horizontal peer communication (40%) but also vertically toward managers (18%).

Workslop burdens organizations because employees who receive such outputs must expend significant effort reverse-engineering unclear AI-generated work, rewriting it, or awkwardly requesting redo from colleagues. This dynamic causes frustration and wastes resources. The issue extends beyond inefficiency to eroding organizational knowledge, as poor-quality AI outputs circulate deeper into companies, increasing remediation costs and degrading collective expertise.

The interpersonal toll is significant: 53% of employees receiving Workslop materials reported irritation, 38% confusion, and over 20% felt insulted by the perceived disrespect for their time. Nearly half viewed colleagues sending unrefined AI outputs as less creative, skilled, and reliable, with 42% losing personal trust and 37% perceiving reduced intelligence. Over a third became reluctant to collaborate with those colleagues, damaging workplace relationships and productivity.

Economically, decoding a single Workslop output costs an employee on average 1 hour and 56 minutes, equating to about $186 monthly per worker. In a 10,000-employee firm, this productivity loss exceeds $9 million annually. The paradox is that AI users save effort, but the burden shifts to recipients and the organization.

The current generative AI flood reverses traditional value equations: as AI-generated content becomes cheap and generic, critical human thinking gains importance. Managers must redefine productivity to emphasize depth of processing over volume of output. This requires cultural shifts rewarding reflection, verification, and professional responsibility rather than polished appearances. The 5% of organizations achieving real AI ROI are those recognizing these new rules and ensuring thoughtful human oversight behind every AI-generated product.