‘This is a training and trust problem, not an AI problem’: Barely any workers are verifying information given to them by AI
AI chatbots are increasingly being used as the first stop for work-related questions, often ahead of colleagues or managers, yet very few workers are taking the simple step of checking whether the answers are correct. In a survey of 2,000 UK employees, only 24% said they verify an AI-generated response before moving ahead, while 39% admitted they have followed AI guidance that later proved inaccurate or unsuitable. At the same time, almost as many workers said they had received no AI training or guidance over the past year, highlighting a clear gap between adoption and support.
The findings suggest that AI may be spreading faster than the safeguards around it. Just 16% of workers said they had role-specific AI training, and 35% reported receiving no guidance at all. Rather than a situation where teams are fully supported, the picture appears to be one where people are expected to use AI and figure things out as they go. According to Alexia Pedersen, the issue is not the technology itself, but the lack of preparation around it. She said workers are moving ahead more quickly than their organizations can provide training and governance, leaving room for mistakes.
Pedersen argued that the solution is not to slow AI down, but to build better habits around its use. That means stronger training, clearer oversight, and a workplace culture that values verification instead of blind acceptance. The research points to a trust and training challenge more than an AI failure, with organizations needing to treat AI learning as seriously as any other workplace skill if they want it to deliver real productivity gains.
