26 June 2026
Dr Beth Montague-Hellen, Head of Library & Information Services, The Francis Crick Institute
When the UK government released its “AI Skills for Life and Work” report alongside an AI Skills Hub, it positioned the initiative as a major step toward nationally accessible AI literacy. Every person in the UK now has access to training designed to increase their comfort with, and understanding of, artificial intelligence. At face value, this feels like a good move. AI is here, people should understand it. But from my perspective as an information professional, a closer reading leaves me less confident of the initiative: what kind of AI literacy training is being provide, and who benefits?
The government’s stated aim is to strengthen understanding in order to build trust. Yet the assumption that trust naturally increases with understanding is fundamentally flawed. In many cases, deeper understanding leads to appropriate caution, and often the rejection of AI for many use cases, rather than increased confidence and use. As information professionals, we know that scepticism is not a barrier to engagement but an essential foundation for informed decision-making. To treat low trust as something to be corrected rather than contextualised is to misread the role healthy doubt plays in navigating any complex information ecosystem, let alone one transformed by AI technologies.
Training Designed by Those Who Stand to Gain
One of the most striking observations arising from the government’s new portal is that the training content is largely created by the technology industry itself. These are organisations with substantial commercial incentives: they want people to use AI tools widely, regularly, and with minimal friction. The result is teaching that privileges the how, while downplaying or omitting the why and the more uncomfortable should you?
For librarians, this pattern feels familiar. It echoes decades of vendor-led “training” around databases, platforms, or discovery tools where the aim is user uptake rather than user empowerment. We know from experience that training driven by commercial priorities can sometimes gloss over nuance, limitations, and ethical considerations. The AI training portal follows the same pattern: its goal is to help people overcome fear, not necessarily to evaluate the technology’s real utility or risk profile. That imbalance should concern anyone whose job centres on information integrity, user advocacy, and critical digital literacy.
Ease of Use Is Not the Same as Meaningful Use
The use of AI tools is becoming increasingly simple, so simple that complicated prompts are already unnecessary for most people. It doesn’t benefit AI companies if the technology is hard to use. To profit, AI providers need the technology to be embedded by organisations and used in an increasingly seamless way. The direction of travel is to make it harder not to use AI than to use it. This frictionless design risks amplifying a broader societal misconception: that ease equals value.
Many employees are being asked by their leaders and managers to start to use AI. But for what? There is growing evidence that when AI is used carelessly and without critical thought it may reduce cognitive abilities. For experts, the editing that is usually required to prevent ‘AI slop’ can take longer than drafting the work by hand the first time. That’s not to say that AI is useless, I do use it in my day-to-day work, but there needs to be a reason, and outputs need to be critically engaged with.
The way we treat AI now feel very much to me like the early years of the internet, a period when the technology was readily accessible yet not particularly useful for many day‑to‑day tasks. The hype outpaced utility for many people. AI is in a similar phase. Tools may be impressively capable in the abstract, but without relevant use‑cases, domain‑specific judgment, and critical interpretive skills, a “just use AI” mentality is not only unhelpful but potentially harmful.
A model generating text is not the same as a model generating knowledge, and without training that explicitly addresses that distinction, users may conflate convenience with correctness. I strongly believe that training that concentrates on how to get a shiny looking output out of an LLM is teaching about AI by starting at the wrong point.
Misinformation: The Real Literacy Gap
The UK’s AI strategy misses the most urgent educational need: training citizens to detect misinformation. Problematic AI-generated content is now ubiquitous, from fabricated news stories to hallucinated academic citations, but misinformation is not an AI‑only problem. People have always manipulated information to push angles, narratives, and ideological agendas. AI simply accelerates these manipulations.
Information professionals have long emphasised critical thinking and source evaluation. These remain the most powerful tools against misinformation, whether human‑crafted or machine‑generated. Yet AI training does not always centre these skills, and while concerns about misinformation are rightly highlighted in the government report, they are considered to be driven by fear and distrust of AI. I would propose that the truth is in the other direction. A healthy awareness of misinformation and the need to critically appraise sources is what can lead to distrust of AI tools, and there’s nothing inherantly wrong with that.
Librarians and other information professionals already teach information literacy. Rather than considering AI as an entirely new topic, it can simply be integrated into the way we already encourage students and researchers to consider their sources. AI outputs carry the appearance of authority, although without the helpful filters of peer review or editorial control. Yet the government’s approach of reducing distrust of AI risks encouraging users to treat AI content as trustworthy. Instead, we should leverage our longstanding traditions of critical appraisal to shape AI literacy, discouraging fear, but encouraging a healthy level of distrust.