Skip to content
Interface language: English
AI course / Signal

One-to-one AI course or video? How to choose and measure learning

Compare recorded and individual AI learning by testing a real task, checking results and course terms.

Article language: English

Three people look at a model of buildings and bridges while one points to a detail.
In this article

Watching a course and completing a task independently are two different entries in a calendar. To choose between learning with an instructor and recordings, start with the work you want to be able to do. We will prepare an assessment card for someone developing company training materials.

The scenario is fictional. The task is to turn a short operating instruction into an exercise card for a new employee. The card must contain an objective, actions and a checking question. It must not add rules absent from the instruction.

Trial task: turn the instruction into a one-page exercise card.
Material: “We receive submissions through a form. We check completeness.
We return missing items for completion. We pass complete submissions to the case owner.
The employee receiving a submission does not decide whether it is justified.”
Expected result: objective, four actions, a checking question with an explanation.
Do not add response deadlines or decision-making authority.
After completing it, identify which parts of the instruction support the card.

Keep your first independent attempt. It is a baseline for learning, without having to claim a percentage productivity gain for either format.

When video may be enough

A recording can introduce a concept or show a particular action. Ask a practical question, however: can you repeat the task on different material without copying the instructor's movements at the same time?

If you can find and correct your own mistakes, you mainly need an organised path, current materials and time to practise. There is no reason to pay extra for a consultation simply because it appears in a more expensive package. Check whether the course lets you work on your own example and shows criteria for a correct result.

In our exercise you can check the number of actions and the prohibition on deciding whether a submission is justified. If the card gives the employee that authority, you have a tangible error to correct. A certificate for watching a video does not change the card's content.

Assess a specific error. In the fictional card, the new employee was instructed: “Check that the submission is complete and reject it if you consider the claim unjustified. Respond to the customer within two days.” Neither that authority nor the deadline appears in the original instruction.

Assess the faulty card passage against the instruction at the start of the article.
Identify every added authority and every deadline not given.
Explain how these additions change the new employee's task.
Correct the card using only the instruction.
Finally, prepare a checking question to see whether the learner distinguishes
checking completeness from deciding whether a submission is justified.

The corrected material should leave the decision to the case owner. A checking question could describe a complete submission with questionable justification and ask for the next step. The model answer is to pass the case to the owner, because the card does not give the receiving employee authority to reject it.

Read your own explanation aloud. If all you can say is “the AI wrote it”, you still need to work with the source. The technique of checking answers and the limits of a conclusion (in Polish) also helps when teaching others: training material can sound clear while leading to the wrong action.

What you pay for in a one-to-one meeting

In a conversation with an instructor you can show a particular result, explain your assumptions and receive feedback on what to improve. When choosing this format, ask how work is assessed, who leads the meetings and what you will prepare between them.

A session need not consist of clicking together. Bring input material, a prompt, the result and the point where you had difficulty. For example: “I don't know whether my checking question tests understanding or just memory of one sentence.” The meeting can then address your problem.

PARP's 11 May 2026 communication about the BKL study points to the importance of learning, adaptation and communication using AI alongside technical skills. This is a reason to assess more than use of an interface. It is not a study comparing the effectiveness of specific courses.

Test the skill on new material

After learning, prepare a second exercise card from a different instruction of similar difficulty. Keep the same requirements and record:

Fidelity: is every instruction supported by the source?
Boundaries: does the card preserve the employee's responsibilities?
Usefulness: does a new person know what to do and how to check the result?
Independence: which problems did I solve without the instructor's hints?
Time: record preparation, checking and corrections separately.
Error for the next attempt: one specific item.

Good work with AI assistance does not yet prove mastery of the whole skill. OECD Digital Education Outlook 2026 analyses conditions for using generative AI in learning; it concerns education, so do not transfer it directly into a claim about a particular professional course's effectiveness.

The broader report “Skills in the AI age”, dated 8 July 2026, describes changing skills needs. Your assessment card has a narrower task: to show progress in one activity you actually perform.

Also test transfer to another task. This instruction is new, fictional material: “Publication materials go to the editor. The editor checks that the sources are complete. Missing items are returned to the author. The head of the department approves publication.” Keep the difficulty similar instead of immediately attempting an entire editorial strategy.

Prepare an exercise card from the new instruction about publishing material.
Keep the same format: objective, actions and a question with an explanation.
Do not give the editor authority to approve publication finally.
Afterwards compare your reasoning with the earlier submission exercise.
Identify the shared principle and the difference in roles.
Do not assess my progress without seeing both actual pieces of work.

Assess the new card using the same criteria. You can use a tool to prepare a draft and then independently defend your assignment of roles and boundaries. Record where you still need a hint. You can bring that question to an instructor and return to it after corrections.

If your learning goal is to create processes, also try the task on human approval of AI automation (in Polish). It tests a similar skill on different material: identify who prepares the proposal, who approves it and when an action has a real effect. A change of context like this can show whether you understand the principle beyond one memorised example.

Relate this to the terms of the AI Course

Before choosing the AI Course (Polish-language page), check whether it is available and read the description of how learning works (in Polish). Ask about the current scope of materials, independent work and any meetings.

Before deciding, check the current price, payment and cancellation terms, access period and conditions for external tools. Do not assume the course provides every subscription without limits.

Compare the programme (in Polish) with your own task, and find information about the project on Aeneia. Before choosing, you can do a first AI exercise for free (in Polish), identify the tools you need (in Polish) or use checks for spreadsheet analysis (in Polish). If a course conversation is available, you can bring your first card and one mistake you cannot yet fix. They are useful material for judging what support you need.

*Publication dates do not represent the dates of all research described. This example is not a participant's story or a measurement of course effectiveness.*

Aeneia / Aeneia editorial

Editorial article. Source links are included in the text.

About Aeneia

Brands we have worked with

  • L’Oréal
  • Garnier
  • Netflix
  • Carrefour
  • Warner Bros. Discovery
  • Biedronka
  • HBO Max
  • Wolt
  • Żabka
  • COSRX
  • Embassy of China in Poland
  • Mixa
Aeneia / Initium

From knowledge to a decision

Tell us what you want to put in order. We will agree on where to start.

Explore the AI course (in Polish)

We partner with and use solutions from:

  • Google
  • OpenAI
  • Microsoft
  • Stripe
  • Anthropic
  • Linear