In boardrooms around the world, AI is still being sold as the ultimate productivity machine, a digital colleague that never sleeps, never complains and always delivers. But new research shows that the promise of efficiency is proving to be an expensive illusion. According to the MIT State of AI in Business 2025 report, 95% of AI projects do not deliver a measurable ROI.

The cause? AI promises efficiency, but often creates the opposite effect. In an article in the Harvard Business Review, BetterUp describes this as “workslop”: the flood of mediocre, AI-generated output that undermines the quality of work and causes productivity to decline.

Where AI once promised to save time, it increasingly introduces a new layer of inefficiency. While some employees use AI to refine and elevate their work, many rely on it to produce output quickly, output that often turns out to be incomplete, superficial, or devoid of essential context. The insidious effect of ‘’workslop’’ is that it doesn’t reduce the workload but merely shifts it: the burden moves from the creator to the recipient, who must interpret, correct, or even redo the work.

One respondent in the BetterUp survey put it this way:

‘Receiving this poor quality work created a huge time waste and inconvenience for me. Since it was provided by my supervisor, I felt uncomfortable confronting her about its poor quality and requesting she redo it. So instead, I had to make an effort to do something that should have been her responsibility, which got in the way of my other ongoing projects.’

According to a study by Gallup, the use of AI in the workplace has nearly doubled in the past year. But that increase has not translated into higher productivity. Employees now spend more time checking, improving and fine-tuning AI output than on their actual tasks.

As The Atlantic wrote back in 2008 in Is Google Making Us Stupid?, technology intended to make us smarter can actually impair our ability to think critically and deeply. What began with Google’s information abundance, fragmenting our attention across endless tabs and hyperlinks, has evolved into a new stage with AI. Where Google fragmented our attention, AI seems to be diluting our meaning.

The constant stream of automatically generated ideas and texts creates a sense of speed, but not of direction or humanity. AI produces output without experience or intention; it acts, but does not understand. As a result, depth is slowly disappearing from work: there is less room for reflection, context, and human judgment. What remains is efficiency without direction and progress without meaning.

Human growth, however, depends precisely on that friction, on the time it takes to think, to make mistakes, and to learn. As noted in an article by TrendsActive, our brains need effort and struggle to stay healthy and creative; convenience may feel efficient, but it weakens our cognitive muscles.

Accenture investigated this pattern and concluded that only 2% of companies that use AI know how to use it effectively, thanks to a solid data foundation and a genuinely human-centred approach. That is precisely where the key to success lies.

These 2% do things differently. They understand that the real challenge of AI is not technological, but about the human dimension. They use AI not to replace people, but to amplify what people do best: making sense of complexity, exercising judgement, showing empathy, and fostering creativity. In these organisations, AI is integrated into the flow of work to support reflection, collaboration, and better decision-making. By aligning technology with human values and behaviour, they turn AI from a tool for automation into an engine for learning, trust, and innovation.

This means new forms of organisation, in which judgement, empathy and creativity do not disappear, but rather make the difference. The future does not belong to companies that generate more, but to those that dare to select better, that choose value over volume. Ultimately, it will not be the smartest companies that master AI, but the most human ones.