AI literacy at work

Lesson 1 of 5

What AI is, and what it is not

Language models such as ChatGPT and Claude write the most likely answer based on patterns in data. They can be wrong with great confidence, make up sources and miss things. Use AI as help with drafts and overviews, and always check the result before you use it.

Lesson 2 of 5

Personal data and confidential data

Never type or paste names, national ID numbers, account numbers, customer data, contracts or other confidential data directly into an AI tool. Use GoalCube Personal data: the AI sees pseudonyms such as [PERSON_1] instead of names and numbers, and the real personal data stays in the EU (Art. 5 and 44 GDPR).

Lesson 3 of 5

The person decides

AI must not make decisions about people, for example about hiring, promotion or terms of employment. The AI recommends; you assess the facts of the case and decide yourself in GoalCube, with a reason (Art. 22 GDPR, Art. 14 AI Act). Be aware that it is tempting to just follow the AI.

Lesson 4 of 5

Prohibited and high-risk

Some things are prohibited, for example measuring employees' emotions with AI (Art. 5 AI Act). Recruitment and HR cases are high-risk and require human oversight and a log (Annex III). So record every new use of AI in GoalCube's AI inventory before you start it.

Lesson 5 of 5

When something goes wrong

If the AI gives a wrong, discriminatory or strange result, or if personal data was sent to the AI by mistake, stop and report it to the person responsible for the AI policy.

Question 1 of 6

A colleague wants to paste an applicant's CV into ChatGPT. What do you do?

It is fine if the chat is deleted afterwards Use GoalCube Personal data so the AI only sees pseudonyms It is fine if the name is removed, but the rest stays

Question 2 of 6

The AI recommends refusing an employee's request for reduced hours. Who decides the case?

The AI, if it is sure A person who assesses the facts of the case and gives the reason for the decision The employee themselves

Question 3 of 6

What must you do before you start a new use of AI?

Record it in the AI inventory Nothing, if the tool is approved Ask the AI whether it is lawful

Question 4 of 6

Which of these is prohibited under the AI Act?

Letting AI write a draft of a newsletter Using AI to read employees' emotions through a camera Using AI to summarise a report

Question 5 of 6

The AI refers to a source you do not know. What do you do?

Uses it – the AI knows enough Checks the source before using it Asks the AI to confirm that the source exists

Question 6 of 6

You discover that customer data was sent to an AI by mistake. What do you do?

Nothing, it happens a lot Deletes the chat and forgets about it Reports it to the person responsible straight away

Finally

Write your name and finish

Your name is stored only at GoalCube in the EU and is not shared with the AI. It appears on your proof of completion.

Your name

Your role (optional)