UK’s AI readiness gap caused by skills underinvestment, says research

Coursera and Udemy’s global ranking places the UK 19th among 98 countries

The UK risks falling behind in the global AI race because investment in AI infrastructure and policy is outpacing workforce skills development, according to the latest Coursera and Udemy Global Skills Report.

The UK’s strong AI policy and technology base is being undermined by weak workforce preparation, according to the latest Coursera and Udemy Global Skills Report.

The research, which is based on a subset of the millions of learners on the Coursera and Udemy platforms, ranks the UK 19th among 98 countries in its new AI-Human Skills Synergy Index. Germany, France and Japan took the top three places.

Despite placing in the 97th percentile for government capacity, policy infrastructure and technology readiness, the UK scored less strongly for learner preparedness. Speaking at an event to discuss the research findings Marni Baker Stein, Chief Content Officer at Coursera and Udemy, attributed the result largely to “an underinvestment in lifelong learning, in workforce development”.

Professor Michael Wooldridge, Ashall Professor of Artificial Intelligence at the University of Oxford, argued that investment in skills should be matched by investment in technology.

“Underinvestment is the headline,” he said. "Solving the UK's long-standing productivity problem means giving people the skills that they need to be productive but also giving them the technology to enable them to be productive.”

Demand for training rapidly accelerating

Demand for training is nevertheless accelerating. Across Coursera and Udemy, generative AI enrolments exceeded 45 per minute, up from 25 per minute in 2025. Coursera’s total AI enrolments rose 67% year on year, while generative AI enrolments increased 132%. AI security learning grew 665% and agent-based workflows 546%.

But the speakers warned against the idea of equating AI competence to skilful prompt writing. “It’s a lot more than writing prompts,” Wooldridge said. “It’s really about thinking creatively about how you can get AI to do the work for you.” He, and Lila Ibrahim, Google DeepMind’s first Chief AI Readiness Officer added that organisations also need sound data strategies and an understanding of where AI is and isn’t useful.

Answers are cheap, good questioning is much harder

The research introduces the AI-Human Skills Synergy Index which is a first-of-its-kind measure of how learners are blending technical skills with the very human qualities employers keep telling us are more important than ever.

The report found that learners are pairing technical knowledge with critical thinking, ethics, governance and leadership. Critical-thinking enrolments rose 107% and complex problem-solving 126%. Lila Ibrahim said the company increasingly looks for curiosity, collaboration and evidence of continued learning. “Asking, getting answers is super cheap these days,” she said. “But knowing how to ask really good questions is getting harder and harder.”

When asked about the tangibility of generative AI improvement Professor Wooldridge offered an example from his own work: an AI system checks a textbook he is writing every morning and emails the authors a list of errors. He said the process was “knocking months off the schedule of the book.” When the LLM was updated to the newest version, it immediately found three pages of additional problems.

As an anecdote, it serves to illustrate both the productivity gain on the table from agentic AI and the pace at which workers must adapt.

The Doomers are wrong

In a riposte to the AI doomer cult who insist that the technology will render the majority of white collar work redundant (a message enthusiastically amplified by the AI labs) Baker Stein said disruption is happening faster at task level than at job level.

“What we’re finding with our enterprise partners is that task disruption is happening a lot faster than job disruption,” she said. “Yes, there may be certain jobs that are that are drying up in the moment, but a lot of those jobs still exist. But what's changing underneath them is the tools and the tasks and the workflows that that really represent what that job does.”

That requires workers to identify which activities can be delegated to AI and which new responsibilities they can take on.

The panel was cautious about predictions of mass job losses. Wooldridge argued that AI is more likely to automate parts of jobs than replace entire roles, although some employers may cut entry-level posts prematurely or AI wash job cuts made for different reasons, like over hiring five years ago.

Governance gaps and micro-credentials

The gap between AI deployment, governance and staff skills are wide, and growing. Nearly three-quarters of organisations participating in the research had deployed or were testing AI, but only 18% said most of their workforce had received AI training in the previous year.

In higher education, more than 95% of surveyed students and educators used AI, while only 26% of educators said their institution had a formal policy.

Wooldridge argued that organisations couldn’t wait for government to provide frameworks. “You can’t wait for the regulation to arrive,” he said, urging organisations to adopt their own guidelines and established frameworks.

Ibrahim described governance as “active work at all levels”, with senior leaders accountable and external voices brought in when internal expertise is insufficient.

Another interesting trend picked up by the research is that of micro-credentials - short, assessed qualifications which are really gaining traction. Micro-credential enrolments rose 33% year on year, with AI micro-credentials up 122%.

The research report and yesterday’s discussion point to the same conclusion: the UK has real strength in the policy, technology development and the financial backing necessary to realise its ambitions to compete in the global AI marketplace.

Converting these foundations into the necessary productivity will require sustained, job-specific learning that combines technical capability with human judgement and accountability.