AI Literacy: Getting Your Team Genuinely Competent with AI

Most UK businesses have already spent money on AI. Licences have been bought, seats assigned, tools switched on. The awkward truth is that buying access is not the same as building competence, and the gap between the two is where budgets quietly disappear. A licence on a desk does not equal a person who uses it well, and Recon Analytics has been blunt on this point: workers are adopting AI faster than their employers, but a paid seat is not the same as real adoption. If three quarters of your staff barely touch the tool, three quarters of that spend is doing nothing.

The missing ingredient is AI literacy: the everyday competence that lets people use AI confidently, responsibly and independently as a baseline job skill. The way SpotDev closes that gap is not a slide deck of features and prompt tricks. It is a deliberate method built on a model that has guided skills training for half a century, the conscious competence ladder, applied to AI and run by educators. If you also want to understand how the tools themselves get built into your operation, our guide to Claude AI agents for business covers the implementation side. Literacy and implementation are two halves of the same job.

The conscious competence ladder, applied to AI

Our method starts with a map. The most useful one for adult skills is the Four Stages of Competence, often called the conscious competence ladder. It was created by Noel Burch in the early 1970s at Gordon Training International, who called it the Four Stages of Learning Any New Skill. It is frequently misattributed to Abraham Maslow, but the model does not appear in his work, so the credit belongs to Burch and Gordon Training International. The ladder tracks two things: awareness and skill. Applied to AI, the four stages look like this.

Stage one: unconscious incompetence

People do not know what AI can do, and do not realise they are missing a skill. This is the colleague who still copy-types data between systems, or who assumes AI is "not for my role". There is no demand for training here, because the person sees no gap to close.

Stage two: conscious incompetence

People realise AI could help and that they cannot yet use it well. This is an uncomfortable but valuable stage, and it is often where fear and avoidance live. Handled badly, people retreat. Handled well, this is where learning begins.

Stage three: conscious competence

People can get good results from AI, but it takes deliberate effort: careful prompting, checking outputs, thinking about what to delegate. The skill is real, but it is not yet automatic.

Stage four: unconscious competence

AI is woven into how someone works without conscious thought. These are the power users, and their value to a team is significant. Most organisations have a handful of them already, often without naming them as such.

The goal we set with clients is straightforward: get everyone to at least conscious competence, then use the unconsciously competent power users to lift everyone else. That second move is the one most programmes miss. A power user's edge is only useful to the wider team if their tacit, automatic skill is made conscious and transferable. So the practical step is to identify your internal power users, surface what they actually do, and use them to coach colleagues, rather than relying on external training alone. The ladder then stops being a diagram and becomes an operating model. We apply the same framework outside AI too, as in our piece on the conscious competence ladder for CRM teams.

The AI-specific challenges generic training ignores

AI adoption carries problems that ordinary software training never has to confront. A course that only teaches feature clicks will leave every one of these untouched, which is why we design the method around them.

Fear of replacement. Anxiety makes people hide their AI use or avoid it entirely, which blocks honest learning. A Pew Research Center survey from early 2025 found that 52% of US workers were worried about the future use of AI in the workplace. You cannot coach people who are concealing what they do, so the method has to make it safe to be a beginner. That is exactly what the conscious incompetence stage is for: naming the gap without shame so people will actually engage.

Calibrated trust in outputs. Literacy means trusting AI the right amount, neither blindly nor not at all. A 2025 study by KPMG and the University of Melbourne, surveying more than 48,000 people across 47 countries, found that 66% of people rely on AI output without checking whether it is accurate. The opposite failure, refusing to use AI at all, wastes the investment just as surely. The skill we build is knowing when to verify, which only comes from supervised practice, not from a one-off briefing.

Governance and safe use. People need to know what is allowed, what data can go where, and how to verify. Governance is part of literacy, not a separate policy bolted on afterwards. Where there is no sanctioned, well-understood route, people improvise with personal accounts and the organisation loses sight of its own data. A safe, sanctioned route plus the skills to use it well is what keeps people out of the shadows.

The seats-bought versus skills-built gap. All of the above feed the core problem we opened on. Buying the tool is the easy part. Building the judgement and safe behaviour to use it well is the work, and it is the work generic training skips.

How to build literacy that sticks

Behaviour change does not happen in a one-day bootcamp. The shape of the training matters as much as the content. A single intensive day is the wrong shape for a skill people need to apply at their desks, because most of it is forgotten before anyone gets to use it.

This is why our training runs as spaced learning: short sessions, with deliberate practice gaps in between, so people process and apply skills between sessions rather than forgetting them by the following week. Spacing study over intervals is one of the oldest and best evidenced findings in learning science, and it is precisely the right shape for a workplace skill that has to be practised on real tasks. The conscious competence framework is the progression model underneath, moving everyone to at least conscious competence and then leveraging power users to lift the rest. Sessions are delivered remotely with recordings, so the learning is repeatable and new starters can catch up.

The commercial discipline comes from measurement. We run a before-and-after competence survey, so the shift up the ladder is visible and the return is provable rather than assumed. We are measuring movement from unconscious incompetence towards conscious and eventually unconscious competence, person by person and team by team. Tie that improved adoption back to the wasted-seat problem and the case writes itself. Fewer dead licences, more people actually using what you bought, and a number you can show the board.

Why SpotDev is built for this

This is training led by educators, not a slide deck assembled by a vendor. SpotDev's founder, John Kelleher, is a qualified educator, a former secondary teacher who progressed to assistant headteacher, which is where the pedagogy-led approach comes from. The training pedigree runs back more than 10 years through the acquired agencies Klood and ESM Inbound. We are a HubSpot Diamond Partner with an in-house team and more than 300 technology projects delivered.

On the tooling side, we specialise in Anthropic's Claude as an area of expertise. To be precise, that is deep expertise in the tool, not an Anthropic partnership. The combination matters: educators who understand how adults actually learn, paired with specialists who build the AI into your workflows. AI literacy is the human layer that makes AI implementation actually pay off, which is why our training and our Claude implementation are two halves of one offer. We work with UK businesses of roughly 30 to 250 staff, the size where adoption either spreads or stalls.

Frequently asked questions

What is AI literacy in a business context?

AI literacy is the everyday competence to understand, evaluate and responsibly apply AI tools in real work. It means knowing where AI fits a workflow, where its limits are, and how to combine it with human judgement. It is not coding or model building. For a business it covers staff who can use AI safely and independently, plus leaders who can judge its potential and risks.

What is the conscious competence ladder and how does it apply to AI?

The conscious competence ladder, created by Noel Burch at Gordon Training International, maps four stages of learning a skill: unconscious incompetence, conscious incompetence, conscious competence and unconscious competence. Applied to AI, the goal is to move everyone to at least conscious competence, where they can get reliable results with deliberate effort, and then use the power users at the top of the ladder to coach colleagues so the skill spreads.

Why does spaced learning work better than a one-day course?

Spacing study and practice over intervals is one of the best evidenced findings in learning science, and it produces far stronger long-term retention than cramming the same material into a single day. Short sessions with practice gaps let people apply skills on real tasks between sessions, which is what makes a behaviour-change skill like AI use actually stick.

How do you measure whether the training worked?

We run a before-and-after competence survey, so the shift up the conscious competence ladder is visible and the return on the training is provable rather than assumed. That lets us tie improved adoption directly to the wasted-seat problem: fewer dead licences and more people genuinely using what the business has already paid for.

Work with a Claude specialist

SpotDev designs, builds and deploys custom Claude agents and enterprise Claude rollouts for UK businesses, with fixed packages from £8,000 to £45,000 and training and adoption built in. Explore our Claude implementation packages or talk to one of our engineers.