Workers Pay for AI Tools Themselves, but AI Productivity at Work Lags

British workers are buying and teaching themselves the AI tools their employers have not provided, with uneven results for AI productivity at work. A survey of 25,000 UK working adults published this week by Deloitte UK found that 63 percent knowingly use generative AI on the job, yet many are doing so without training, without company backing, and in some cases without their employer even knowing.

The survey, conducted by Ipsos between May 7 and June 10 this year and billed as the largest single-country study of workplace AI use to date, found adoption running well ahead of workplace policy. Nearly a third of AI users said they use the tools without their employer's knowledge, a practice known as shadow AI. One in six users pays for at least one AI tool out of their own pocket, which Deloitte estimates adds up to 958 million pounds in annual personal spending across the UK workforce.

That enthusiasm is not translating into clear productivity gains. About 31 percent of respondents said they had used AI at work but saved no time by doing so. Around 30 percent said the tools saved them at least some time each week, while just 7 percent reported saving five hours or more. On average, users said AI saved them 70 minutes a week, time most said they reinvested into work for the same employer.

Workers are moving faster than their employers

The survey suggests the bottleneck is no longer willingness. Nearly a quarter of workers said they use AI tools every day, and 46 percent said they use free tools to get work done. Among users, usage concentrates on routine tasks: 43 percent said they use AI to search for information or draft emails, and 31 percent use it to generate summaries.

Meanwhile, employer support lags. Around half of workplace AI users said they had received no formal training on using the tools safely and effectively, according to the survey. Deloitte's own analysis says most workers have had no training and only a minority feel confident that their leaders understand the technology. Just 35 percent of users said they hear their leaders speak about AI with a good understanding.

The gap has left workers to figure it out on their own. Deloitte UK's chief AI officer, Hayley McKelvey, said the challenge for businesses is no longer persuading people to try the tools, but meeting the demand in a way that is secure, responsible, and built to last. She said workers are showing they will not wait for permission, with many using free tools or paying for premium versions themselves.

The time savings are uneven

The productivity results break into two groups: A minority of heavy users report large time savings, while a substantial share see none at all. That split echoes the wider tension over AI and labor: while AI labs argue over whether the biggest labs colluded to slow AI development, workers in the survey were already dealing with the practical fallout. In information and communications, 21 percent of respondents said AI saved them five or more hours a week. In healthcare and social work, that figure fell to 5 percent.

Usage patterns also differ by demographic. Men were more likely than women to use AI multiple times a day, at 15 percent versus 9 percent. Workers in London used the tools more than those elsewhere in the UK, and marketing professionals were more than twice as likely as the average worker to use AI multiple times daily, at 26 percent compared with 12 percent overall.

The findings suggest that access alone is not the deciding factor. Deloitte's report says the technology's potential for AI productivity at work only shows up when organizations invest in people alongside software, through skills, confidence, and clear guidance on responsible use.

Stigma and fear of replacement linger

Even as use spreads, anxiety around it persists. Nearly a quarter of workers said they feel there is a stigma attached to using AI at work. Among weekly users, 64 percent said they worry that managers will conclude AI can do their jobs.

The leadership vacuum appears to feed that unease. Around 65 percent of users said their organization lacks convincing leadership on how AI should be used, and a separate figure from the LinkedIn coverage of the research put the share of organizations with clear AI policies at just 28 percent. One in three employees using AI without their employer's knowledge suggests that for many, the easiest path is to keep quiet. It fits a pattern seen across the jobs conversation, where the Gen Z versus millennials jobs debate has overlooked how technology is quietly reshaping day-to-day work.

The broader finding, according to the survey data, is that workplace AI use remains experimental. What improves AI productivity at work may depend less on the next model release and more on whether employers close the training and leadership gap their own workers have already spotted. Adoption is widespread but shallow, the report says: many workers have tried the tools, but few use them every day, and most use them for familiar tasks like writing emails rather than rethinking how work gets done.

For a deeper look at the numbers, see Deloitte UK's GenAI Workforce Survey 2026. Secondary reporting on the launch can be found at LinkedIn News' coverage of the findings.