Lesson video

Hello.

My name is Mrs. Holbrough, and welcome to Computing.

I'm so pleased you've decided to join me for the lesson today.

In today's lesson, you'll be exploring AI tools that are used to support accessibility.

But what is an AI tool?

And how can bias affect accuracy?

Welcome to today's lesson from the unit, School blog: digital inclusion.

This lesson is called: AI and digital inclusion, and by the end of today's lesson, you'll be able to identify the benefits and risks of AI tools in supporting accessibility.

Shall we make a start?

We will be exploring these keywords throughout today's lesson: AI tool.

AI tool.

A software application or system that uses artificial intelligence techniques to perform tasks.

Bias.

Bias.

When something is unfair towards or against something or someone.

Look out for these keywords throughout today's lesson.

Today's lesson is broken down into two parts.

We'll start by identifying how AI tools support accessibility, and then we'll recognize risks of biased and incomplete data sets.

Let's make a start by identifying how AI tools support accessibility.

Digital accessibility means designing technology so it can be used by people with a variety of different needs.

For example, visual or hearing impairments, physical impairments, or neurodivergent conditions.

Digital accessibility is about ensuring equity of access in digital products.

An AI tool is a software application or system that uses artificial intelligence techniques to perform tasks.

Here are some common AI tools.

How many of these do you recognize?

ChatGPT, Gemini, Microsoft Copilot, some customer service chatbots which rely on AI technology, Snapchat My AI, TikTok AI filters or chat features, Siri, and Alexa.

Are there any others that you can think of?

AI tools can reduce barriers by adapting information into formats that more people can use.

AI tools and applications can be used to support individuals with visual impairments by: converting text to speech, screen readers, enlarging or simplifying text, recognizing and describing objects through a phone camera.

AI tools and applications can be used to support individuals with hearing impairments by: converting speech to live captions, generating subtitles for videos, and translating spoken language into text in real-time.

AI tools and applications can be used to support individuals with physical impairments.

Some examples include: enabling voice control so individuals don't have to type or use a mouse, using predictive text for common phrases, automating repetitive tasks, sharing maps and information about wheelchair-accessible spaces.

AI tools and applications can be used to support individuals with neurodivergent conditions by: breaking down and organizing information, simplifying text, providing step-by-step instructions, or making summary notes from large documents and presentations.

Okay, time to check your understanding.

I have a question for you.

Which AI tool may be useful for someone with a hearing impairment?

Is it, A, predictive text, B, automatic subtitle generation, or, C, a screen reader?

Pause the video whilst you have a think.

Did you spot it?

Well done.

B is correct.

Automatic subtitle generation could be useful for someone with a hearing impairment.

AI tools aren't just useful for people with additional needs.

AI tools can also be used by designers and content creators to check the accessibility of their products.

AI accessibility-checking tools include: digital PDF and web accessibility checkers that scan PDFs and websites to pinpoint accessibility issues, color-contrast checkers that detect low color contrast on websites and in applications, tools that identify missing image descriptions, or alt text.

AI tools can benefit everyone, even if you don't have an additional need.

For example, captions are useful in noisy environments or when you don't want to disrupt others.

Translation tools are useful if you're learning another language.

Voice assistants are useful if you're busy or need to be hands-free.

Okay, time to check your understanding.

This time, I have a true or false statement for you.

"AI accessibility tools are useful only for people with additional needs.

" Is this true or false?

Pause the video whilst you have a think.

Did you say false?

Well done.

AI tools aren't just useful for people with additional needs.

They can benefit everyone.

For example, AI tools can be used to assist content creators with checking that products meet accessibility standards.

Okay, we're moving on to our first task of today's lesson, Task A.

Read the scenarios and identify how AI tools could support the students.

So Lucas says, "I struggle to read long texts like essays and articles.

How could AI tools help me?

" Sofia says, "I have a hearing impairment.

How could I use AI tools to help me with my schoolwork?

" Pause the video whilst you have a think.

How did you get on?

Did you manage to identify how AI tools can support Lucas and Sofia?

Well done.

Let's have a look at some sample answers together.

So Lucas said, "I struggle to read long texts like essays and articles.

How could AI tools help me?

" AI tools can help by adapting long texts into formats that are easier for Lucas to process and understand.

For example, Lucas could use AI tools to write summaries for longer documents or highlight key points.

Lucas could even get an AI tool to make a set of revision flashcards based on a larger text.

Sofia says, "I have a hearing impairment.

How could I use AI tools to help with my schoolwork?

" AI tools can help Sofia by reducing the barriers of spoken or audio information.

For example, Sofia could use AI tools to automatically generate subtitles for educational videos or provide real-time captions during lessons.

Okay, we're now moving on to the second part of today's lesson where we're going to recognize the risks of biased or incomplete data sets.

Lucas says, "How do AI tools work?

" Do you know the answer to this?

Maybe pause the video whilst you have a think.

AI systems and tools are data-driven, meaning they are trained on vast amounts of previous data to create algorithms that predict and generate the most likely output for a prompt.

A prompt is the input question, instruction, or message you give to a large language model, or LLM.

For example, an AI application that describes images for visually impaired individuals works by comparing the image provided to a vast amount of images stored in training data.

The AI application then makes a prediction about what the image contains.

So the AI doesn't understand that this image contains a picture of a llama.

It's just making a prediction based on the other images it has stored in its dataset.

Lucas says, "But what happens if the AI tool hasn't been trained on the right data?

" AI systems learn patterns from data.

If the data reflects bias, missing voices, or historical inequality, the AI system may repeat or even amplify those inequalities.

The term bias means when something is unfair towards or against something or someone.

Algorithmic bias occurs when an algorithm produces unfair or discriminatory outcomes that favor some groups over others.

AI systems can reinforce inequality when: certain groups are underrepresented or missing in training data, historical bias is included in data sets, systems are tested on limited populations, or designers do not consider diverse user needs.

Okay, time to check your understanding.

I have a true or false question for you.

"AI systems and tools understand fairness.

" Is this true or false?

Pause the video whilst you have a think.

Did you select false?

Well done.

AI systems do not understand fairness.

They detect patterns.

If unfair patterns exist in the data, the AI tool may wrongly treat them as normal.

Sofia says, "So AI tools and applications could actually add barriers and not improve accessibility?

" If accessibility tools do not work equally well for everyone, they may exclude the very people they are meant to support.

Let's take the example of a voice-to-text application.

A voice-to-text application has been designed to support people with physical impairments, such as fine motor difficulties.

A student using the application has a speech impairment and the application frequently misinterprets their words.

The application has not been trained using enough data, which means it does not represent a wide range of users.

Let's look at another example: an AI image description tool.

An application has been designed to describe images to users with visual impairments.

A blind learner uses the tool to describe a geography diagram, but the tool only says, "A map," instead of explaining the key features.

The application has not been trained using enough data to provide the same detail that someone without a visual impairment would see.

Another example: Using an LLM to summarize text.

A student who has dyslexia is using an LLM to summarize some content on a revision website to prepare for an assessment.

The LLM summarizes the information, but it leaves out some key information needed for the assessment.

The student misses this key information and is under-prepared for the assessment.

AI tools can also widen inequality.

Not everyone has equal access to AI tools.

This could be because of a lack of: access to devices, reliable internet connectivity, financial ability to afford premium AI tools, or skills and confidence needed to use AI tools effectively.

The difference between those who have access to digital devices and tools and those who do not is often called the digital divide.

AI tools and applications can widen the digital divide.

Let's have a look at some examples here.

A student with a hearing impairment may benefit from live captions, but this will only happen if their school provides the technology.

A student who struggles with reading may use AI tools to summarize text, but they'll be only able to do this if they have access to a suitable device.

Time to check your understanding.

I have a question for you.

Why can AI tools increase inequality?

Is it, A, because AI tools replace teachers, B, because AI tools do not require internet access, or, C, because not everyone has equal access to AI tools?

Pause the video whilst you have a think.

Did you select C?

Well done.

AI tools can increase inequality because not everyone has equal access to AI tools.

Okay, we're moving on to our final task of today's lesson, and you've done a fantastic job so far, so well done.

For Task B, we're going to be recognizing the risks of biased and incomplete data sets.

AI tools can significantly improve accessibility for people with additional needs.

However, if they are not trained correctly, they can reinforce existing inequalities.

Using a scenario, explain a risk of biased or incomplete data sets in AI tools.

Pause the video whilst you have a go at the task.

How did you get on?

Did you manage to come up with an example?

Well done.

Let's have a look at a sample answer together.

Remember, this is just a sample answer and yours may be different.

AI tools can help people with different needs, but they depend on the data sets they are trained on.

If that data is biased or incomplete, the tool might not work equally well for everyone.

For example, a deaf student relies on AI live captions during lessons.

The captioning system was mostly trained on clear speech from adults with similar accents.

In a science lesson, the teacher has a strong regional accent and uses complex subject vocabulary.

The AI frequently misinterprets keywords, like photosynthesis and evaporation, and replaces them with incorrect words.

Because the dataset did not include a wide variety of accents and technical language, the captions are inaccurate.

This creates a barrier instead of removing one.

The deaf student may misunderstand important information and fall behind compared to other students.

This scenario shows that if AI systems are trained on biased or incomplete data sets, they can reinforce existing inequalities rather than improve accessibility.

Remember, if you'd like to pause the video here and add any extra detail to your answer, you can do that now.

Okay, we've come to the end of today's lesson: AI and digital inclusion.

Let's summarize what we've learned together.

AI tools can improve accessibility.

Artificial intelligence, or AI systems, learn patterns from data to make predictions.

AI can reduce barriers, but it can also reinforce inequalities if not carefully designed.

Access to AI tools can widen inequality.

I hope you've enjoyed today's lesson, and I hope you'll join me again soon.

Bye.