The UK Government's AI skills platform fundamentally misunderstands how people actually develop AI competence. It confuses content with capability, curation with learning design, and comprehensiveness with usefulness. The courses themselves are fine. The way they are organised is the problem.
Four lessons from a £4m mistake
For HR and transformation leaders, before you build anything similar:
- Content is not capability. A course library without sequencing and learning design does not build competence.
- Theory without practice is wasted. You cannot build AI fluency by watching. Real upskilling needs a sandbox to experiment in straight away.
- "Comprehensive" can kill engagement. Breadth over relevance forces intermediate users back through beginner content.
- The "Netflix of learning" model has limits. Twenty-four modules with no entry point paralyses people rather than guiding them.
Fatal flaw one: too many courses, no first step
After the assessment, you land on 24 courses shown with equal prominence and no hierarchy. Repetitive stock photography means you cannot tell them apart at a glance, so you read every title, repeatedly. Within minutes you hit the paralysing question: the cloud course first, the Python fundamentals, or the ethics training? There is no sequencing, no prerequisite logic, no "start here". Progress tracking exists but is nearly invisible, and the time commitments are vague. One course lists "two to seven days" with no sense of the total pathway.
Fatal flaw two: no sandbox, no practice
The platform offers no integrated place to actually use AI. Learning AI without using AI is like learning to swim in the desert. The good AI education platforms, the ones from Anthropic, OpenAI and Google, give you working examples and a playground. The Skills Hub sends you off to external courses, many of which also have no hands-on practice. Adults learn technical skills by doing, getting immediate feedback, adjusting and repeating. Without a practice loop measured in seconds rather than days, proficiency never arrives.
Fatal flaw three: the "intermediate" lie
The assessment asks good questions about industry, role, experience and tech stack. Then, for an intermediate user, it generates a pathway where 71% of the courses assume zero prior knowledge. Tech-stack filtering works, so Google-stack users see Google courses. Experience-level filtering does not, so intermediate and beginner users get the identical pathway. Following it takes 50 to 73 hours. Strip out what an intermediate user already knows and it drops to 25 to 35. Comprehensive and personalised are opposites in learning design.
How you would actually fix it
- Make the pathway sequential. Show Module 1 only, with a "start here" button, lock the rest until the prerequisites are done, show "Module 1 of 4" and a real progress bar, and state the total time up front.
- Trust the assessment data you collect. If someone picks intermediate, hide the introductions. Add a quick-start option: six courses, 8 to 12 hours, focused on immediate application.
- Embed practice, progress and feedback. Put a chat interface on every course page. Show example prompts, and good versus poor outputs. Explain why each module exists and what you will be able to do after it. Test with five real intermediate users and watch where they get stuck.
The real lesson
The platform earns about a 3 out of 10. Good content organised badly is a wasted opportunity. A personalised assessment that does not personalise anything is just data collection. A skills platform that will not let you practise is not teaching, it is lecturing.
The hardest part of AI adoption was never the technology. It is the change management, the behaviour shift, the gap between knowing what AI can do and being able to do it yourself. Capability is built by making learning easy, safe, and part of the workflow, not by a glossy portal wrapped around a curated library.
That gap is exactly what we built the free AI Resource Hub and the AI Fluency course to close: short, practical, nothing gated, with the sequencing and the doing the Skills Hub left out.