Should everyone be trained before launching anything?

Training all staff before launching anything seems prudent. Sessions are organised on generative AI tools, people are shown how to write a prompt, and the hope is that use will follow.

It rarely does. A few weeks later, most of it has been forgotten, for lack of a chance to use it. Useful training starts from a concrete situation in the job: the case you handle every day, the report you write every week. It also teaches people to spot when the tool is wrong, which no generic training really teaches.

• Training without use fades fast.What is not practised in the following weeks is almost entirely lost.
• Needs differ from one job to another.A lawyer, a salesperson and an accountant do not expect the same thing from AI, and do not take the same risks.
• The best trainers are often colleagues.Those who have already built the tool into their work know how to show what works, in your company.

1. 01Start by training the team involved in the first use, on that specific use.
2. 02In each department, identify the people who already use AI with good judgement, and make them your relays.
3. 03Keep a short session for everyone, focused on the rules: what can be given to the tool, and what must be checked.

When tools are not adopted, we first look at how they were designed, before talking about training.

• How do we stop our teams rejecting AI?→
• Are our managers ready?→
• How do we keep human expertise when AI does the work?→
• We already have ChatGPT or Copilot licences. Is that an AI strategy?→