Choose exam board for KS4 Computer Science (GCSE)
Choose exam board for KS4 English
Choose exam board for KS4 French
Choose exam board for KS4 Geography
Choose exam board for KS4 German
Choose exam board for KS4 History
Choose tier for KS4 Maths
Choose exam board for KS4 Music
Choose exam board for KS4 Physical education (GCSE)
Choose exam board for KS4 Religious education (GCSE)
Choose exam board for KS4 Spanish

Lesson details

Learning outcome

I can explain how data patterns lead to AI bias and describe why outputs may be unfair or unrepresentative.

Key learning points

  1. AI bias occurs when a system produces inaccurate results because it has learnt from training data that is flawed.
  2. Because AI systems learn patterns from human-created data, they can repeat and increase existing social stereotypes.
  3. AI tools trained on datasets where certain groups are missing or under-represented lead to tools that perform poorly.
  4. Critically evaluating AI outputs is essential for identifying bias.

Keywords

  • Bias - when something is unfair towards or against something or someone

  • Dataset - a collection of information used to train an AI system, such as images, text or numbers

  • Data bias - when data is unfair or unbalanced, which can lead to wrong or unfair results

  • Fairness - when all groups of people are treated equally and there are no outcomes that disadvantage or misrepresent some people more than others

Common misconception

AI systems are naturally neutral and objective because they are machines.

An AI system will replicate and amplify any human prejudices present in its training data, and its training data may be inadequate for the task it is being used for.

Teacher tip

Task A and Task C could be completed as a card sort activity with students working idependently or as part of a group.

Licence

This content is © Oak National Academy Limited (2026), licensed on Open Government Licence version 3.0
except where otherwise stated. See Oak's terms & conditions
(Collection 2).

Lesson video

Loading...

Prior knowledge starter quiz

6 Questions

Q1.
Why is it important to remember that a chatbot is not human?

It can access your private thoughts.
Correct answer: It predicts text but does not truly understand you.
It has real emotions like people do.
It can replace teachers in all situations.

Q2.
Which example best shows healthy use of a chatbot?

Correct answer: using it to draft ideas before writing your own work
sharing your secrets because it never judges
relying on it instead of speaking to friends
asking it to make personal decisions for you

Q3.
Why can anthropomorphism be risky when using chatbots?

It worsens technical performance.
It helps the chatbot learn emotions.
Correct answer: It may lead you to trust the chatbot too much.
It makes the chatbot use more energy.

Q4.
Which situation shows emotional reliance on a chatbot?

using it to check spelling in homework
asking it for film times at the cinema
using it to translate a short sentence
Correct answer: feeling you must talk to it whenever you feel upset

Q5.
What is a key privacy risk when using chatbots?

They automatically delete everything you type.
Correct answer: Your conversations may be stored or shared.
They cannot remember anything.
They only respond during school hours.

Q6.
Which statement best reflects responsible digital boundaries?

AI tools should replace most real-life conversations.
Correct answer: AI is a tool and should not replace real relationships.
Chatbots are safer to trust than people.
The more time spent chatting, the better.

6 Questions

Q1.
Which statement best explains why AI systems are not naturally neutral or objective?

Correct answer: They identify patterns in human-created datasets.
They are controlled fully by the internet.
They always calculate the most logical answer.
They do not rely on any stored information.

Q2.
If an AI system is trained with images of green apples and red tomatoes, why might the system wrongly label a red apple as a tomato?

The camera quality was too low.
The user selected the wrong setting.
The algorithm stopped working.
Correct answer: The training dataset contained biased labels.

Q3.
Which of the following scenarios is most likely an example of representation bias?

A calculator gives the wrong total.
A weather app predicts rain using temperature data
Correct answer: A speech tool struggles with regional accents
A password is entered incorrectly

Q4.
A predictive policing system sends more patrols to one postcode because past arrest data showed higher crime there. Why is this a concern?

It ignores all historical data.
Correct answer: It may reinforce past inequalities.
It updates too frequently.
It only works at night.

Q5.
Which questions would help you decide whether an AI output might be biased?

Correct answer: Who might be harmed by this output?
How quickly did the system respond?
Correct answer: What data was the system trained on?
Is the device fully charged?

Q6.
Which actions can help reduce bias in AI systems?

Correct answer: using more diverse and representative datasets
Correct answer: testing systems for different error rates across groups
ignoring small differences in performance
removing human oversight from decisions

To help you plan your 8 digital literacy lesson on: Understanding bias in AI, download all teaching resources for free and adapt to suit your pupils' needs...