Three-quarters of London businesses are already using AI. That statistic should read as a success story. Instead, it is the most uncomfortable number in a new survey: the firms running AI tools are the same firms reporting they cannot find the skills to run them well. Adoption outpaced capability, and the gap between the two is now measurable. ## The Numbers, and Why They Are Harder to Dismiss Than Most Surveys A Survation poll commissioned by BusinessLDN surveyed 2,043 London business leaders and found that the share of firms reporting significant AI skills and capacity gaps has nearly quadrupled, reaching 15 percent, according to reporting by the BBC and BusinessLDN. That is a striking directional signal in any single year. Equally telling is what happened on the confidence side: the proportion of leaders who believe their workforce has the skills needed for the AI era dropped from 63 percent to 50 percent over the same period, as reported by Resultsense citing the same survey. These are not projections or modelled scenarios. They are self-reported assessments from people who sign off on hiring budgets. The sample size matters here. More than 2,000 respondents is enough to take the directional signal seriously, even accounting for self-selection in business-leader surveys. What the data does not tell us is which roles are most acutely affected or which sectors are driving the numbers. That granularity matters enormously for anyone making a career decision, and it is the honest limit of what this survey can answer. ## What "Already Using AI" Actually Reveals The detail that deserves more attention is buried in the headline statistic. Three-quarters of surveyed businesses are already deploying AI tools, according to BusinessLDN. That means the widening gap is not primarily a story about laggard firms refusing to engage with technology. It is a story about organisations that moved fast on adoption and are now discovering that deployment and capability are different problems entirely. Installing a tool is not the same as building the judgment to use it, audit it, or adapt workflows around it. This distinction matters enormously for anyone trying to position themselves in the job market. The firms feeling the pinch are not looking for someone to introduce them to AI. They already have it running. What they are discovering they lack is people who can work effectively alongside it: employees who understand where outputs need checking, how to prompt for reliable results in a specific domain, and how to flag when a model is producing confident nonsense. That is a workflow skill, not a credential. ## What This Means If You Are Deciding Where to Invest Time The temptation when looking at data like this is to reach for the nearest certification program. Resist that instinct long enough to ask a more useful question: what does the firm already have, and what is it missing? A business running AI tools for a year but still struggling likely does not need another person who can describe transformer architecture. It needs someone who can QA outputs in a legal context, reconcile AI-generated analysis against source data, or document the prompts and workflows the team has actually settled on. Those are skills you can build without a postgraduate degree, but they require working with real tools on real tasks, not completing modules about AI concepts. For career changers at any experience level, the more productive read on this data is not panic but positioning. The gap is widest among adopters, which means proximity to tools in your current role is worth more right now than a certificate that sits on a profile. If your employer is already using AI and you are not the person steering that workflow, that is both a risk and an opening. The People Management reporting on the UK's broader tech skills shortage adds useful framing: demand is rising fastest not for abstract AI literacy but for applied capability in existing business functions. ## The Honest Limits of One Survey, and What to Watch One Survation poll, even a well-sampled one, cannot tell you whether London's gap is larger or smaller than comparable cities, which job families are most exposed, or whether the confidence decline reflects genuine skill shortfalls or simply higher expectations as firms learn more about what good AI use looks like. Both interpretations are plausible, and the answer probably differs by sector. What the survey does establish clearly, according to the BBC and BusinessLDN coverage, is the direction: confidence is falling, reported gaps are rising, and the firms driving that trend are not beginners. For learners deciding where to invest, the next useful data point to watch is sector-level breakdowns from follow-on research, particularly anything that disaggregates professional services, finance, and public-sector adjacent roles from pure tech firms. The AI Skills for Life and Work Rapid Evidence Review published by the UK government is one place that framing may develop further. Until then, the most actionable read on this survey is simple: adoption without capability-building creates demand, and that demand is now large enough to show up in a poll of two thousand business leaders. ## Sources - Half of London firms report skills gap amid AI boom

Sources