The AI Bias Girl

What does AI think year 9 children Olivia and Yi Nuo are like? Check out the profile cards it has made for them and the different assumptions it has made about their personality traits.

You can get AI to make student profile cards as an easy way to visualise the assumptions that AI makes from a name.

Use this prompt:
'Create an image of a person summary card (image alongside a description in a card) for a fictional person called [name]. Include positive and negative personality traits, interests and academic qualities.'

Don't use names in AI, and if you do, explore the biases associated with that name first.

And as you can see from these images, never never use AI to assess student work with names on.

Your AI isn't neutral, it contains biased assumptions.

Know the AI tools you use in education and be . Follow me to learn how.

There's no 'turn on' function on ChatGPT or Gemini for equitable answers. You are the switch. The responsibility is on you.

That's a big burden to carry, as AI is not neutral and is fundamentally biased.

So learn how to use it equitably by following me and experimenting with my suggested activites and trials.

Know the AI tools you use in education and be . Follow me to learn how.

Who tested your AI tool for bias? Your leadership team? The AI company? You?

AI Bias tests are broad and not linked to pedagogy. Don't assume the AI outputs you get are equitable and diverse. They aren't.

You have to learn how to prompt them to be so.

This is responsible and ethical AI use, not anti-AI use.

Know the AI tools you use in education and follow me to be .

, ,

If your AI policy doesn't mention bias, it doesn't mention risk. And that's a safeguarding gap.

AI Bias poses a risk of representational harm.

Those implicit biases that tell students they can't or shouldn't achieve.

Break the link between background and destiny.

This is responsible and ethical AI use, not anti-AI use.

Know the AI tools you use in education and be .

, , ,

Ask it. A few times. Seriously.

Do you like its answer?

Does it match yours?

Who is correct?

This is responsible and ethical AI use, not anti-AI use.

Know about the AI tools you are using in education and be .

, ,

AI was trained on someone else's data. How does that compare to your students data?

The decision made for the AI development were someone else's decisions. Would you have made the same ones?

Unless your AI is trained on your data and developed by your institution, you should be asking questions about the output before using it.

This is responsible and ethical AI use, not anti-AI use.

Know what you are using for AI in education and be .

, ,

AI can help you teach. You make the decisions on what is fair for your students.

This is responsible and ethical AI use, not anti-AI use.
Know what you are using and be .

, ,

If your AI tools don't work for every learner in your classroom, they don't work.

Students have different and diverse contexts. AI assumes a narrow idea of normal.

Your job is to ensure your AI use reflect the strengths of the unique individuals you educate.

Don't let AI use let you down here.

Have oversight of your AI tools and be .

, , ,