Articles For leaders deciding about AI

What should training providers charge for in the AI era?

Price practice, expert feedback and observed assessment by what each gives the buyer.

Jacob Top Møller3 min read

Illustration: two paper workstations show a learner getting help and a learner presenting a finished bridge

A training provider can sell help getting better at a task, or evidence that someone can do it under stated conditions. When a course bundles the two, AI can be useful during practice while making it harder to tell whose work earned the pass.

In the AI credential experiment, providers issued 424 credentials to one account and those credentials were verified against public provider records. That tells us about issuance in this experiment. It leaves open what the account holder learned and what buyers would pay for if training and assessment were offered separately.

What should learners get to practise?

Imagine a course for people who have to explain a proposed process change to a customer. The learner could ask AI to play a sceptical customer, try a different explanation and compare the answers. A coach could then flag the point at which the learner promised something the team could not deliver. The learner gets another attempt while the feedback is still useful.

Illustration: two teal paper figures try parts for an unfinished bridge

That gives a provider several things to price. The explanation can be self-serve. Practice can come with scenarios learners can repeat. Expert feedback can be offered on attempts where someone needs to see why a decision failed. If buyers mostly need practice and feedback, ask which of those they would purchase on its own before adding more content to the package.

What does a pass need to show?

If a buyer needs evidence that a person can handle that customer conversation, give the learner a new scenario and ask them to make and explain a choice. Say which tools are allowed, which decisions the learner must explain personally and what will count in the score. If the real job allows AI, an assessment can allow it and ask the learner to identify what they checked and why they accepted or changed its suggestion.

Illustration: a teal paper figure presents a finished bridge to an observer

Carnegie Mellon’s Eberly Center advises choosing assessment tasks that reveal whether learners met the learning objective. For this course, the buyer might want to see someone identify a customer constraint and offer an alternative they can deliver. Scoring that decision takes a different kind of evidence from recording that a lesson was completed.

Where is human review worth paying for?

A provider can keep completion simple and charge separately for an observed assessment when a buyer has a decision that warrants it. That assessment needs reviewer time, a way to score consistently and a plan for repeat attempts. Spell out those costs before setting the price. A buyer who just wants staff to practise may have little reason to pay for that extra work. A buyer deciding who can lead a customer conversation may want it.

Before changing the product, write the sentence the buyer could rely on after each purchase. “The learner completed the course” and “the learner demonstrated the task with these tools and met these criteria” support different decisions. Name the decision first, then price the work needed to support it.

If you think I could help, write to me or give me a call.