This post was originally published on this site.
Artificial intelligence has crossed an important threshold in British business. It’s no longer something companies are merely discussing, testing in innovation teams or watching from a safe distance. People are using it, and they’re using it a lot.
The latest UK Business Data Survey found that 41 per cent of businesses handling digital data now use AI for at least one purpose. Among large companies, that figure rises to 82 per cent. Separate research from the Office for National Statistics suggests adoption may be moving even faster among workers themselves, with 55 per cent of employees reporting that they use AI for work or education.
Those numbers definitely sound impressive. But they also raise a more difficult question for British business. If we’re adopting AI this quickly, when do we start seeing the transformation we’ve been promised?
For investors, business leaders and policymakers, that question matters much more than the number of people who’ve opened an account with ChatGPT or had Copilot rewrite an email. Britain doesn’t have an AI awareness problem anymore. Increasingly, it has an AI integration problem.
Using AI isn’t the same as changing a business
The government’s own data makes the distinction quite stark. Among businesses already using AI, only 21 per cent say their AI tools are integrated into existing business systems.
When you look at what companies are actually doing with the technology and the picture becomes clearer. The most common reported use of AI is researching information, cited by 28 per cent of businesses handling digital data. Another 21 per cent use it to summarise information or draft reports and correspondence.
These are obviously useful applications. I use AI tools myself and can see the value they offer in removing some of the friction from everyday work. But we need to be careful about describing every efficiency gain as some sort of ‘transformation’.
As an investor, I’m much more interested in what happens when AI moves deeper into a company. Is it changing how customers are served or how products are developed? Can it shorten a process that previously took days or more to hours? Is proprietary company data being used more smartly? Can management make better decisions because information that once sat in different systems can now be understood together?
These are harder changes to make. They’re also where the economic value is likely to become much more significant.
Only 5 per cent of AI-using businesses in the UK Business Data Survey reported using automated decision-making systems. Just 6 per cent said they use data to develop, train or improve AI or automated decision-making systems. Much of British business, in other words, is still near the beginning of this process.
Workers are moving faster than their companies
The gap between individual and corporate adoption is particularly interesting. ONS research found that 55 per cent of employees were using AI for work or education, while 35 per cent of businesses with ten or more employees reported using at least one AI technology.
There’s something encouraging about that. Technologies often spread because people discover that they solve a real problem, rather than because somebody at head office tells them to use one. But it creates challenges too, particularly when individual experimentation moves ahead of the systems and rules surrounding it.
The UK Business Data Survey points to that governance challenge. Its detailed findings show that only 5 per cent of businesses using AI have a formal written policy governing its use or development. Among large businesses, however, the figure rises to 56 per cent.
That’s a remarkable difference. It suggests that the emerging AI divide in Britain isn’t simply between companies that use the technology and those that don’t. There’s also a divide between businesses with the resources to integrate and govern it properly and those that are largely figuring things out as they go.
The answer shouldn’t be for smaller businesses to slow down with layers of bureaucracy. They don’t need an AI committee for the sake of having one. But they do need to understand what information employees are putting into external systems, where decisions remain subject to human judgement, and which uses of AI carry genuine commercial, legal or reputational risk.
Good governance should make adoption easier, not harder.
Productivity is appearing before revenue
There are already signs that AI is delivering economic benefits. Research published by the Department for Science, Innovation and Technology earlier this year found that 56 per cent of businesses currently using AI reported increased employee productivity.
But there’s another number that deserves at least as much attention. Some 77 per cent of businesses using AI said they hadn’t yet seen any change in revenue. Only 12 per cent reported an increase.
I don’t find this particularly surprising. Productivity gains should appear before many of the larger commercial benefits. A member of staff saving an hour on a task has value, but it doesn’t automatically create a new customer, a better product or a new source of revenue.
The next stage is turning those accumulated efficiencies into something more substantial. Businesses need to ask what they can now do that they couldn’t do before, rather than simply how they can do existing tasks slightly faster.
This distinction will become increasingly important for investors too. Asking a management team whether it “uses AI” is already becoming a fairly meaningless question. Before long, almost every company will be able to answer “yes”.
I’d rather know where AI sits inside the business, which processes have changed because of it, what measurable improvement has followed and whether competitors could easily reproduce the same advantage. Those questions tell us much more about whether AI is creating lasting value.
Britain doesn’t need to build everything
The debate about Britain’s position in artificial intelligence often gravitates towards comparisons with the United States and China, which is understandable. Frontier models, computing infrastructure, chips and research capability matter enormously, and the UK should remain ambitious about its role in all of them.
But Britain doesn’t have to dominate every single layer of the AI economy to benefit from it. Nor should its success be measured solely by whether the next global foundation model is built here or not.
The country already has considerable strengths in financial services, life sciences, professional services, advanced research, creative industries and technology. It also has millions of smaller businesses whose productivity matters enormously to the wider economy. For many of those companies, the opportunity isn’t to become an “AI company”. It’s to become a better company because of AI.
This may sound like a small distinction, but economically it could be the more important one. The productivity benefits of a technology don’t come only from the companies that invent it. They spread when businesses in other sectors reorganise around what the technology makes possible.
The latest numbers suggest Britain has become rather good at experimenting with AI. The challenge now is to move beyond experimentation and embed it into the far less glamorous machinery of business, from operations and customer relationships to product development, finance, logistics and decision-making.
If it can do that, the most important British AI story may not be the creation of one spectacular company. It may be thousands of existing companies becoming more productive, more competitive and better able to grow. That’s a harder transformation to capture in a headline, but it’s the one that could matter most.
