How do you choose an AI course for your workplace? Seven questions

The supply of AI courses has multiplied, but the quality has not. MIT research (MIT NANDA) shows that 95% of corporate AI projects deliver no measurable results, usually because people get the tools but not the right training. These seven questions help you separate the wheat from the chaff before you sign.

Why most courses deliver nothing

A typical AI course looks like this: a lecturer shows slides about the technology's potential, participants try a few generic examples and everyone goes home with a certificate. Two weeks later nobody remembers anything and the ways of working are unchanged.

The problem is not that people are uninterested. The problem is that the course never connected to the participants' real tasks. A finance officer who practises on recipes and travel tips does not learn to use AI for reconciliations and reporting.

The research shows what is at stake. A Harvard/BCG field experiment with 758 consultants showed 25 to 56% faster completion of tasks with proper use (Harvard/BCG 2023). But the same study also showed that untrained use can make experienced people slower and more confident in the wrong answer. Choosing training is therefore not a formality. It decides which side of the line your company lands on.

The seven questions

Send these questions to anyone offering you a course. The answers tell you everything you need to know.

Question 1: Do staff practise on their own tasks? · good answer: Yes, the workshops are built on participants' real tasks · warning sign: Generic demonstrations that have nothing to do with your work

Question 2: Is the training tailored to roles? · good answer: Finance practises on finance tasks, marketing on marketing material, small groups · warning sign: Everyone in the same room, the same content for all

Question 3: Are results measured before and after? · good answer: A baseline before training, the same measurement repeated later, you get the numbers · warning sign: A satisfaction survey at the end of the day and nothing more

Question 4: Is safe use part of the curriculum? · good answer: Usage policy, data classification and verification of results are part of the teaching material · warning sign: Safety dealt with in a single slide or a disclaimer

Question 5: What happens after the course ends? · good answer: In-house specialists trained, a roadmap, the option of follow-up · warning sign: A certificate and goodbye

Question 6: Does the instructor build AI solutions themselves? · good answer: Instructors who build and use the technology daily in real projects · warning sign: Lecturers who talk about the technology but do not use it themselves

Question 7: Is a vocational training fund grant available? · good answer: A defined course with a curriculum that meets the funds' conditions, help with the application · warning sign: "You sort that out yourselves" or no answer

A provider that answers all seven credibly is rare. All the more reason to ask.

The three questions most people forget

Measurement (question 3) is the strongest filter. Someone who does not measure results never has to deliver on anything. Someone who measures a baseline before training and repeats the measurement on the same workflows later is betting their own reputation on the outcome. You want your own numbers from your own company, not statistics from foreign reports.

Follow-up (question 5) decides whether the skills survive. According to a Viska survey, 80% of professionals in the Icelandic labour market use AI at work but only 34% have received training from their employer (Viska 2025). A single one-off course does not close that gap. Look for training that leaves behind in-house specialists who keep the development going.

The instructors' background (question 6) is a simple quality signal. The AI landscape changes monthly. An instructor who builds solutions every day teaches what works today. An instructor who put together slides last year teaches what worked last year.

Funding changes the arithmetic

Question seven is not just about savings but about the provider's professionalism. Vocational training funds reimburse companies a large share of training costs, in some cases up to 90%. But the funds draw a clear line between a course and consulting: an eligible course needs a defined start and end, a published course description and a predetermined curriculum.

A provider who knows these rules, structures the offer accordingly and helps with the application saves you both money and effort. A provider who has not heard of the funds has probably not sold training to many Icelandic companies. We went through the funds and the amounts in detail in a separate article on vocational training funds.

Entitlements depend on the employees' unions, so the first step is always to map out which funds apply to your company.