Lesson 2. Bias, fairness and inclusion

Reflection task: Checking an AI output for bias

Choose one AI-generated output that could be used in an adult education setting. You may use an output you already have, or you may generate a simple example, such as:

• – a short learner feedback text;
• – a lesson example;
• – a case study;
• – a description of a “typical learner”;
• – a communication message to learners;
• – an image prompt or AI-generated image description;
• – a recommendation for learner support.

You can also use the example below:

“Adult learners who struggle with online learning are usually older learners who are not comfortable with technology. They often need simple step-by-step instructions and may not be able to manage independent learning tasks. Younger learners are usually more confident with digital tools and can complete online activities more easily. To support older learners, educators should reduce the complexity of digital tasks and avoid asking them to use too many online platforms”.

Read the output carefully and answer the questions below.

1. 1. Who is represented in the output?
2. 2. Are different ages, genders, cultures, languages, abilities, and life situations visible?
3. 3. Who is missing or underrepresented?
4. 4. Does the output ignore any learner groups who may be present in adult education?
5. 5. What assumptions does the output make?
6. 6. Does it assume that all learners have strong digital skills, stable employment, high literacy or the same cultural background?
7. 7. Could the output disadvantage or discourage anyone?
8. 8. Could the wording, examples, or recommendations create unfair treatment?
9. 9. Is the output suitable for your real learners?
10. 10. What would you need to edit before using it?
11. 11. What human check is needed before use?
12. 12. Who should review the output, and what should they check?

Follow-up action

After answering the questions, revise the AI output to make it more inclusive, fair and suitable for your learners.

You can improve it by:

• – adding more diverse examples;
• – removing stereotypes;
• – simplifying language;
• – adapting it to different literacy or digital skill levels;
• – checking whether the tone is respectful and supportive;
• – making sure no learner group is unfairly represented or excluded.