AI mistrust costing businesses £29bn

Mistrust in AI could be costing businesses £29bn in lost productivity, according to technology company UnlikelyAI.

A survey by the firm found that 32% of employers are experiencing AI "burnout" due to the constant pressure to check outputs, as staff spend two-and-a-half hours per week either verifying or re-doing results.

Almost all respondents said they spent at least some time checking AI outputs per week, from quick sense checks (18%) to redoing some or all of a task manually (20%). Eighteen per cent said they ignored the output entirely.

As a result, only 57% of respondents said they were seeing a return on investment in time spent on AI. Thirteen per cent said, "it has been actively bad for the organisation".

UnlikelyAI's research also uncovered how employees were dealing with cognitive strains associated with relying on large language models (LLMs) such as ChatGPT, Gemini or Copilot.

Thirty per cent said they experienced "AI blindness", where they were losing perspective after repeatedly prompting AI tools and receiving inconsistent answers. A third felt they were becoming too dependent and losing their skills after routine use.

Another common symptom was "analysis paralysis", where employees felt they were unable to decide whether to trust AI or their own intuition, cited by 31%.

When asked why there were such high levels of mistrust in AI, 32% said explainability of outputs was an issue; 32% said security and safety; 31% questioned the consistency of answers; and 31% felt outputs were logically or factually incorrect.

More than a quarter (28%) encountered hallucinations, where AI produces entirely fabricated outputs.

William Tunstall-Pedoe, CEO and founder of UnlikelyAI, said the research highlighted a critical challenge.

"There has to be a better way to use AI. LLMs have strengths in specific, limited areas, but there's a huge lack of understanding about when to use them and when to look to other, less fallible models. That's where this trust gap is coming from," he said.

He advised companies to set ground rules for appropriate AI use, adding: "Clarity helps eliminate anxiety and uncertainty - so people are free to perform better, in the knowledge they're using a tool responsibly and within guardrails."

Managers should also train teams on different types of AI and their strengths and weaknesses, and encourage staff to prioritise explainability over novelty.

"When you're in a high-stakes business context, the long-term ROI on trustworthiness far outweighs the short-term gains of speed alone," he said.

Earlier this week, a survey by software company The Access Group found that seven in 10 employees are "experimenting" with AI tools in their daily work, but only one in five had received any formal training.