What AI literacy actually means at work

Who this is for

This is for the L&D, HR, or transformation lead who has been asked to "do AI literacy" and has been handed a demo booking, a vendor webinar, or a one-hour lunch-and-learn as if those were the same thing. They are not. A demonstration of a chatbot is a product tour. Literacy is whether a claims handler, a planner, or a line manager can use the systems you actually deployed without creating a mess you will later have to explain.

If your current plan is to show people ChatGPT and call the workforce literate, read this before you send the invite.

The business problem

Most organisations can point at licences. Fewer can point at a shared standard for what good use looks like. McKinsey reported in 2025 that 88% of organisations use AI in at least one function, while only about a third have scaled past pilots, and that only 39% see any EBIT impact. The Conference Board reported in 2026 that only 33% of workers received employer AI training in the last six months. TalentLMS reported in 2025 that 49% of workers say AI is moving faster than their company's training.

That is the pattern L&D inherits: tools arrived, a few people taught themselves, everyone else is guessing, and a sponsor now wants a literacy programme that will survive an audit question. The wrong response is a generic demo. The right response is a definition of literacy that is specific to the work, then training and a record that match that definition.

The licences are paid for. The team was not trained. is the finance-shaped version of the same gap. This note is the definition: what you should mean when you say literacy, so you do not buy a demo and file it as a measure.

What literacy is not

It is not having seen a model generate a paragraph. It is not a certificate from a vendor who has never seen your policy. It is not "we rolled out Copilot." It is not a lunchtime webinar with no register. Those things can be useful introductions. They do not show that staff can handle the work.

A demo answers "what can this product do in a clean example?" Literacy answers "what may this person put into the tool, what must they check, when must they refuse the output, and how do they record that AI was involved?" Those are different questions. L&D that collapses them will train people to be impressed and leave them unable to work.

What you should be able to show

A sponsor, an auditor, or a customer questionnaire will eventually ask what you did. Prepare the answer before you buy the training.

1. A role map, not a headcount. Literacy for a receptionist who summarises notes is not literacy for someone who uses a model output in a regulated file. World Economic Forum reporting in 2025 found that 63% of employers name skill gaps as the top barrier to transformation. Treating the whole workforce as one audience is how those gaps survive a budget. Write down which roles use which tools on which kinds of work.

2. A checking standard. People need a way to decide whether an output is fit to use. That includes sources, numbers, names, and the organisation's own rules. AI Foundations for Every Role is the session Experrt runs so a mixed cohort shares a vocabulary before they specialise. It is not a product tour.

3. Judgement under your policy. What may go in, what must stay out, when a person signs the work. Responsible AI Use at Work is delivered against the organisation's own policy, not a generic acceptable-use poster. If you do not have a policy yet, do not pretend a demo created one.

4. A record. Who attended, what they were taught, whether the work was assessed. A smile sheet is not a record. A dated export that names the cohort, the course, and the people is. Sponsoring an AI Literacy Programme is the session for the person who will have to describe those measures later.

McKinsey has also reported that leaders with highly AI-fluent teams are 3.9 times more likely to capture enterprise value. Fluency is not a vibe. It is repeated, checked use against a standard. Literacy is the floor that makes fluency possible.

How this sits next to Article 4

Article 4 of the EU AI Act requires providers and deployers to take measures supporting AI literacy among staff. It does not require a certificate and it does not name a course. If you are in scope, the legal overlay is real, and EU AI Act Article 4: what L&D actually has to do is the briefing for that duty.

Do not wait for counsel to define literacy as "whatever the vendor sold." Define it as the ability to use the deployed systems, with judgement, on the organisation's work. Then commission training that produces evidence of that ability. A ChatGPT demo will not.

What to insist on when you commission

• A written definition of literacy by role, agreed before anyone quotes.
• Live facilitation on your tools and your files, or a close analogue, not only a public chatbot.
• Practice that includes refusal: when not to use the tool.
• Attendance and an assessed piece of work, not only a quiz on the slides.
• A sentence you can reuse: "These roles were trained to this standard, on these dates, and here is the export."

Experrt will not tell you that watching a model write a poem made anyone literate. We will train the cohort against the work and give you the pack.

What to do this quarter

Write the role map on one page. Name the tools those roles already use. Decide what a competent person in each role must be able to check. Book a foundations cohort for the mixed group and a sponsor session for the person who will answer for the programme. If you need help turning that page into dates and a record, contact Experrt. Bring the role list. That is enough to start.