British employees are spending nearly six hours each week 'botsitting' artificial intelligence tools — manually feeding them data, correcting their mistakes, and re-doing their work — according to a new study that casts doubt on the widely touted productivity benefits of generative AI.
The research, conducted by the technology consultancy UST, found that the average UK worker loses five hours and 48 minutes per week to tasks associated with managing AI systems. This includes time spent 'spoon-feeding' the AI with accurate prompts, verifying its outputs, and fixing errors that the software has introduced. The phenomenon has been dubbed 'botsitting' by industry observers.
For businesses, the findings present a stark warning. Many firms rushed to deploy generative AI tools — such as chatbots and automated content generators — in the hope of slashing costs and boosting output. However, the hidden labour of botsitting means that net productivity gains may be far smaller than anticipated, or even negative in some cases. 'If your staff are spending a quarter of their week babysitting a bot, you have to ask whether the technology is actually helping,' said Dr. Elena Marchetti, a digital workplace analyst at the University of Cambridge.
The implications for the UK economy are significant. With the Office for National Statistics reporting stagnant productivity growth, any technology that fails to deliver genuine efficiency risks wasting billions in investment. Moreover, the rise of botsitting raises concerns about employee burnout. Workers may feel pressured to adopt AI tools without adequate training, leading to frustration and reduced job satisfaction. The UK's Information Commissioner's Office (ICO) has already signalled that it will scrutinise AI deployments that place undue burden on staff, particularly where automated decisions affect employment terms.
From a regulatory perspective, the UK is navigating a complex landscape. While the government has opted for a 'pro-innovation' approach to AI regulation, rather than adopting the EU's AI Act wholesale, the ICO has made clear that existing data protection and employment laws still apply. Companies that deploy AI without proper human oversight or that fail to provide training could face legal challenges. 'The EU AI Act imposes strict requirements on high-risk AI systems, and even though the UK has diverged, many British firms trading with Europe will need to comply,' noted Sarah Llewellyn, a partner at the law firm Fieldfisher.
For consumers, the botsitting trend could mean slower service and more errors in automated customer interactions. If AI tools are not properly supervised, the risk of incorrect billing, flawed advice, or poor-quality content rises. Experts advise businesses to invest in proper onboarding and to treat AI as a tool requiring skilled human guidance, not a replacement for it. 'The key is to design workflows where humans and AI complement each other, not where humans become the AI's cleaner,' added Dr. Marchetti.