You’ve probably seen it by now. A job ad that looks like every other job ad, until the last line of the “about you” bit: confident with AI tools.
If you work in an office, that line can land badly. What counts as confident? Do you need a course? A certificate? Is everyone else already miles ahead?
A big survey of employers came out in September, and it answers most of those questions. The short version: the bar is lower than the ads make it sound, and most employers would rather teach you than replace you. Here are four things people believe, and what the evidence actually says.
Myth one: everyone else is already using it all day
The Learning and Work Institute asked just over 1,000 people who make hiring and training decisions how their organisations use AI. Most of them (85%) are using it somewhere. But fewer than one in five use it “extensively”. In smaller firms it’s closer to one in eight.
So in most workplaces, AI is something a few people use for a few jobs. It isn’t the whole building. If you feel behind, you’re probably level with most of the people around you.
That doesn’t mean nothing is changing. More than half the employers expect AI to change the way they work a great deal in the next two or three years. It’s early, which is exactly why a little confidence now goes a long way.
Myth two: you need a qualification in it
Only 15% of the employers said AI training that ends in a recognised qualification was a priority. Half said the main way their staff pick up AI skills is on the job, by doing it.
The thing on the ad isn’t a certificate. It’s whether you can use these tools sensibly in the job you’d be doing.
Myth three: they’ll hire an AI person instead of training you
Government figures point the other way. Take the businesses that say a lack of know-how is holding them back on AI. More than six in ten said their answer was to train the staff they already have. Just 3% planned to recruit someone new for it.
For most jobs, the plan isn’t to find a specialist. It’s to get the people who already know the work comfortable with a new tool.
Myth four: “confident with AI” means something technical
This is where job ads mislead. About one UK job advert in thirteen now mentions AI, according to figures from the jobs site Indeed. But Indeed’s own researchers point out that a mention can mean almost anything. It might be real technical know-how. It might be a note that AI will sort the applications.
When Skills England set out the basic AI skills most jobs need, its list was short and practical:
- knowing the risks of using AI, and when not to use it
- writing a clear request, then adjusting it until the answer is useful
- reading the trends in a simple dashboard
- handing a routine task to an AI tool
Nothing on that list needs a degree. Most of it is judgement.
The report is blunt about where the real risk lies. “Over-confidence is perhaps the biggest risk here,” it says. Anna Dent and her colleagues at the Institute warn that learning on the job can skip the parts that matter most. That means protecting people’s data, and checking that what the tool gave you is right.
What confident looks like on a Monday
So “confident with AI tools” means something far more ordinary than it sounds. You can show three things.
- You use it for real tasks. Not “I’ve tried ChatGPT”. Something like: “I use Copilot to turn meeting notes into a first list of actions, then I edit it.”
- You check what it gives you. You know it can be confidently wrong, and you have a habit of catching it. In September we looked at arguing with it rather than taking its first answer. That’s the habit.
- You know what not to put in. Customer names, colleagues’ details, anything confidential. Next month we’ll go through exactly where that line sits.
Try this: before your next application or appraisal, write down two tasks you’ve used an AI tool for at work. Add one time you caught it getting something wrong. That’s your answer to “are you confident with AI tools?” It’s a better answer than most people give.
Never used one at work? Start with the dullest job on your list. Summarising a long email chain is a good first go. Then check the summary against the chain. You’ll learn more from the checking than from the summary.
The bit the headlines miss
The honest picture from this survey is a slow change, not a cliff edge. Employers want people who can use these tools in the job they already know how to do, and mostly they expect to train them.
The survey has limits. It asked 1,012 HR decision-makers online in June, with more London and tech firms than the country as a whole. But it points the same way as everything else we’ve read this year.
Being able to say what you did with it, and how you checked it, already puts you ahead of most applicants.
What’s one task at work you’d happily never do by hand again?
