In this article
- What do you need before training starts?
- How should AI literacy training be tiered by role?
- What does a 90-day AI literacy plan look like?
- How do you turn your AI policy into training material?
- Which training formats work best for AI literacy?
- How do you keep AI literacy from fading?
- AI literacy programmes: questions people ask
The short answer
To build an employee AI literacy programme, write or refresh your AI policy first, baseline current skills, then train in tiers: a hands-on foundation for everyone, role-based practice for regular users, extra depth for managers and high-risk roles, and AI champions to keep it going. Run it over about 90 days, keep training records and refresh when tools or policies change.
The most common mistake I see is training before deciding the rules. People leave the session enthusiastic, then ask "but are we allowed to use it for this?" and nobody knows. So the plan below starts with policy and ends with habits, with training in the middle where it belongs.
What do you need before training starts?
Three things, all quick:
- An AI policy people can read in five minutes. Approved tools, data rules, checking, disclosure and who to ask. See what AI policy you need before training staff.
- Approved tools that are actually available. If staff cannot log in to the business version, they will use the free one.
- A sponsor. A director who opens the first session and uses AI visibly. Adoption follows the boss, as we explain in how to get your team to adopt AI.
How should AI literacy training be tiered by role?
Not everyone needs the same depth. Four tiers cover most organisations:
| Tier | Who | What they learn | Typical time |
|---|---|---|---|
| 1. Foundation | Every employee | How AI works, approved tools, safe data use, checking output, the policy | Half day hands-on, plus a short refresher |
| 2. Practitioner | Regular AI users in each team | Role-specific workflows, prompt library, context and documents | Half day per team, on their own tasks |
| 3. Manager | People who approve or supervise AI-assisted work | Reviewing AI work, setting team rules, measuring benefit and risk | 2 to 3 hours |
| 4. High-risk and champions | HR, recruitment, finance, legal, customer decisions, AI champions | Bias, explainability, escalation, records, supporting colleagues | Half day to a day, plus ongoing support |
For sizing and timing guidance, see how long AI training for a team takes and our worked example of AI training for a 20-person team.
What does a 90-day AI literacy plan look like?
Weeks 1 to 2: decide and map
Confirm the sponsor, list the AI tools in use, approve the policy and pick the teams to start with.Weeks 3 to 4: baseline
Run a short skills check. Our AI capability assessment and skills matrix template make this a half-hour job per person.Weeks 5 to 6: foundation for everyone
Hands-on sessions in small groups on the approved tools, using real but safe examples from your business.Weeks 7 to 9: practitioner and manager tiers
Team-by-team sessions on each team's own workflows. Managers learn how to review AI-assisted work.Weeks 10 to 11: champions and high-risk roles
Deeper sessions for HR, recruitment, finance and anyone using AI near decisions about people. Appoint one champion per team.Week 12 to 13: check and record
Short scenario assessment, update training records, collect the time saved and the problems found. Plan the first refresher.
How do you turn your AI policy into training material?
Policies are forgettable; scenarios are not. A quick way to build realistic practice material is to ask an AI tool to turn your policy into short situations your staff will recognise. Remove anything confidential first.
You are helping me build AI literacy training for staff at a [type of business] with [number] employees. Below is our AI acceptable use policy. Write eight short workplace scenarios, each 3 to 4 sentences, where an employee has to decide how to use AI. Cover: personal data, client confidential information, checking facts, using an unapproved tool, disclosing AI use, AI in a decision about a person, a task where AI should not be used, and asking for help. For each, give: the scenario, the right action under our policy, the policy section it relates to, and one common wrong answer. Use plain UK English. Do not invent policy rules that are not in the text; if the policy is silent, say so. POLICY: [paste policy text]
Check it: Read every scenario against the policy yourself. Where the AI says the policy is silent, that is a gap worth fixing before training.
Which training formats work best for AI literacy?
For AI literacy, the strongest format is live and hands-on, because the skill is judgement and judgement needs feedback. E-learning works well for refreshers and policy checks. Most programmes mix both.
Formats that build judgement
Live workshops
Small groups for the foundation and practitioner tiers, working in the approved tools on real tasks. See in-person vs online AI training.
Champion drop-ins
Fifteen minutes a week where people bring real problems and leave with a fix.
Formats that keep it going
Short e-learning or videos
Refreshers, new starters and policy updates at scale.
A shared prompt library
Each team owns and improves its own set of tested prompts.
“One of the best courses in 30 years of corporate life”
Review of The Oxford AI SchoolHow do you keep AI literacy from fading?
Skills fade if they are not used. Three habits keep them alive: managers asking "how did you check that?" in normal reviews, champions sharing one useful prompt a week, and a light refresher every six to twelve months or whenever a major tool or policy change lands. To show the programme is working, track the measures in how to measure AI literacy and the business benefit in how to measure the ROI of AI training.
AI literacy programmes: questions people ask
How long does it take to roll out AI literacy training?
A small or mid-sized business can design and deliver a full first round in about 90 days: two weeks to decide and map, two to baseline, six to train in tiers and two to assess and record.
Should we write an AI policy before training staff?
Yes. Training without clear rules leaves staff unsure what they are allowed to do. A short policy on approved tools, data, checking and disclosure should come first, and the training should teach it.
What is an AI champion?
An AI champion is a member of a team who gets extra training and time to help colleagues, share useful prompts and spot problems early. One champion per team or per 15 to 20 staff is a common ratio.
Is e-learning enough for AI literacy?
E-learning is useful for refreshers, new starters and policy checks, but on its own rarely builds the judgement AI literacy needs. Hands-on practice with feedback on real tasks is far more effective.
How often should AI literacy training be refreshed?
At least once a year, and whenever your approved tools, AI policy or the roles using AI change significantly.
Keep reading
We will run your 90-day AI literacy programme with you
Policy, baseline, tiered hands-on training and records, delivered around your teams and your tools. You keep the materials and the champions.
- Basic AI Training Hands-on AI literacy for every employee.
- AI Policy Workshop Write the rules your training will teach.
- AI Capability Assessment Baseline your team before you train.
Harry Lang
Harry Lang founded The Oxford AI School in Witney, Oxfordshire. He trains UK businesses and teams to use AI well, writes the Marketing Unfiltered newsletter, and is the author of Learn To Fly With AI.