Lesson video

Thank you for joining me for your Design and Technology lesson today.

My name is Mrs. Conway and I'm going to be guiding you through your learning for this lesson.

So today's lesson outcome is I can use AI to visualise design developments.

So we're gonna be delving into the world of AI and we're gonna consider how we can actually use it as part of our designing.

These are our keywords for today.

AI, algorithm, bias, and stereotypes.

I'll be going through each of these in a little bit more detail as we introduce them during the lesson.

So the first thing we're gonna look at then is AI considerations.

So AI, as you may already know, stands for artificial intelligence.

And artificial intelligence is the simulation of human intelligence in machines that can learn, reason, and solve problems. Now, there's been a rapid evolution in AI capabilities in recent years and it's quickly become part of our everyday lives, sometimes without even necessarily realising it has.

So let's just do a quick check for understanding on this.

What is AI? Is it A, a programming language used to build software applications; B, the physical components of a computer that process data; or C, the simulation of human intelligence in machines? Pause the video here just to take a moment to think about that.

And the answer? It was C, the simulation of human intelligence in machines.

Well done if you got that right.

Now, designers use something called generative AI, which learns from the collections of data found online.

And these collections are called data sets.

Now, AI follow special sets of instructions called algorithms to create answers based on these data sets.

Algorithms are constantly updated so that more accurate answers can be generated, so AI is just gonna get more accurate and more powerful the more that we use it and put information into it.

And using AI when designing can help to develop design ideas.

And quite often AI is perceived as cheating, but it's not cheating to use AI.

The caveat to that, of course, is that you need to be open and honest about your use of it.

Don't try and claim AI work as your own.

You need to be quite open and admit when you've used it and explain how you've used it to further your development of ideas.

And there's loads of advantages, but there's also disadvantages to using AI when designing.

Let's just do a quick check for understanding on that.

What does AI use to generate answers or images from data sets? Is it A, data labels; B, information caches; or C, algorithms? Pause the video here just to take a moment to think about that.

And the answer was C, algorithms. Well done if you got that right again.

So let's just consider bias.

Now, bias is one of the issues that can occur when using AI, so this would be a disadvantage.

And this can occur when we're using AI to generate design ideas.

Bias is a tendency to favour or judge something unfairly based on personal beliefs or preferences, and it can occur in other elements of our life as well.

But bias in AI is sometimes called algorithm bias.

The AI model can generate bias results due to the human bias that can affect the data sets or the AI algorithm.

The data sets are generated from how we are using the internet and what we're putting onto the internet so that human bias can then actually come through through algorithm bias.

Now this can lead to unfair or discriminatory outcomes called stereotypes.

And stereotypes are generalised belief or assumption about a group of people or things.

And here Alex has given us an example of a stereotype and he has said, "People who live in New York walk really quickly and are always in a hurry." We'll dig more into that stereotype in a moment.

Let's just do a quick check for understanding before we do.

Which of the below would be an example of a stereotype? A, all old people are bad with technology; B, some elderly people volunteer for charities; or C, my grandfather enjoys hiking.

Now, think really carefully about this one and pause the video here if you'd like to take a moment to think about this.

The answer? A, all old people are bad with technology would be a stereotype as it's a generalised belief about a group of people and it's not based in fact at all.

So let's just look at Alex's stereotype in a little bit more detail.

Now, his stereotype was based that everyone in New York walks really fast.

Now, logically we know this cannot be true, as there are lots of different people living in New York who all walk at different paces.

Now this is gonna be an example of a stereotype because this information is not based on fact at all.

Right, time for you to have a little go.

Your first task then is to explain what bias and stereotypes are and how they can affect designing when using AI generative programmes.

Your second task, you're gonna need the computer for.

You're going to have a go at inputting different prompts into AI generative programmes.

So for example, you might put into it "a doctor at work" or "a futuristic fashion model." What you're then gonna do is analyse your results.

Are they diverse? Do they reinforce stereotypes? It'll be quite interesting to see what comes up and don't be afraid to put in other options as well as the ones that I've recommended here.

Pause the video here to have a go at these three tasks and good luck.

How did you get on with that task? So the first thing I asked you to do was explain what bias and stereotypes are and how they can affect designing when you're using AI generative programmes.

So I'm giving you an example of an answer that you could have written here but obviously yours might be slightly different, so just compare your answer to mine.

So bias and stereotypes are unfair assumptions about people based on things like gender, race, or age.

In AI design, if the AI is trained on biassed data from those algorithms, it can create designs that reflect those stereotypes.

For example, it might ignore diversity or only focus on one group.

Now this can lead to unfair or limited designs, so it's important to use diverse data to avoid bias.

As I said, your answer will be slightly different, that's absolutely fine, so just compare what you've got in your answer to mine.

Your second task then was to input different prompts into AI generative programmes, and you might have come up with quite a few different examples.

Here's one that I've come up with under that prompt, "a doctor at work." Just gonna give you a moment to have a little look at that image and see if you think it comes up or has any kind of issues with bias or stereotypes based on that image.

Now your third task was to analyse your results, so I'm gonna analyse the results based on the image that I got.

Again, you'll have analysed your results based on the images that you found and they'll be slightly different.

So I asked you, were your images diverse? Or did they end up reinforcing stereotypes? So in my example, I've put, "This image reinforces the stereotype that doctors are typically men and white, which can be misleading.

It also presents a polished, possibly American setting.

It doesn't really look like our NHS kind of hospitals, and that may not reflect doctor surgeries in the UK or other parts of the world as well." Remember, I didn't actually say anything about where this person should be living, so we should be looking at an image that is very global.

"Overall, the image contributes to the idea of gendered professions as well, which can be a really problematic issue." Okay, again, you'll have a slightly different answer and it'll be quite interesting to share your answers with other people in the classroom based on the images that you found, and you might have found that some of the images you found were diverse, which would be wonderful to have seen.

So next we're gonna look at using AI to develop designs.

We've considered the advantages and the disadvantages and the things we have to be aware of and consider as we use AI, but how can we actually use it in our own design work? So it can be used to help spark creativity when a designer has design fixation, or it can help to iterate ideas by changing details such as materials, for example.

Now, there's lots of different ways AI can be used but we're gonna focus on three different ways today.

And those different ways are, method one, using prompt words to generate ideas.

Method two, uploading our own sketches to then iterate.

And method three, which is uploading our own sketches to then view in a particular environment.

As I said, there's more ways than this, but these are the three ways that we're gonna look at today.

Let's just do a quick check for understanding on that then.

Which one is an advantage to using AI when developing design ideas? Is it A, increases creative block; B, helps with design fixation; or C, limits brainstorming options? Pause the video here just to take a moment to think about that.

Alright, and the answer? It was B, it can really help with design fixation and to move ideas along.

So let's look at each one of these methods in turn.

Method one was using prompt words to generate ideas.

Now, prompt words can be put into an AI generative programme to generate ideas, and you used prompt words for your other task where you were looking at whether or not your generative ideas reinforce stereotypes.

Prompt words are basically keywords that the algorithm will use to generate an idea.

One way to find prompt words is to use the client feedback on design ideas that has been already gathered.

You can then pick out those words and turn them into prompt words for putting into the generative AI.

So let's look at this first method then in practise.

That method was using prompt words to generate ideas.

Now below is Laura's client feedback from her herb planter design ideas, and her client feedback stated they wanted her to focus on the designs that go in the window, develop the vertical designs, and also make it look more organic in style.

So Laura can now pick out the keywords from her client feedback and decide upon the prompt words.

Her prompt words are going to be herb, pot, storage system, hanging, natural, organic, window, vertical.

So you'll notice it's not really a sentence, it's more individual words or prompt words that will help to generate these ideas.

And this is what's come out from that.

So this is an AI-generated design based on those prompt words that Laura put into the AI system.

Laura would then use this design to help generate some further ideas of her own.

She wouldn't claim this idea as being her own.

So if she put this into design work, she would clearly label it as an AI-generated design, but then she would take that design and further develop it.

So let's have a look at how she would do that.

Here you can see Laura's idea developed from the AI design.

So you can see she's still stuck with this vertical idea, but she's just altered it slightly.

So Laura's just gonna explain what she's actually done here.

So Laura said, "I hadn't thought about having a double-sided option before, so I decided to use this idea from the AI image.

To make it look more natural, I made the central pillar into more of the basic shape of a tree trunk." Now this is a very stylized shape of the tree trunk, as you can see.

"Rather than keeping it straight." So she still developed it and taken it further.

She hasn't copied that idea entirely, she's made small little changes to make it kind of fit her own image or her own ideas.

Let's just do a quick check for understanding on what we've just been through then.

What do you input into an AI generative programme when generating design ideas without any prior sketches? Is it A, prompt words; B, random words; or C, trigger words? Feel free to just pause the video here to take a moment to think about that.

Okay, and the answer? It was those prompt words, well done.

Alright, so let's take a closer look at method two then, and method two was uploading our own sketches to then further iterate.

So with some generative programmes, you can actually put your own images into there alongside some prompt words or even just the image on its own to see what it can then iterate and further develop.

So Laura has uploaded a sketch to experiment with what her design would look like in different materials or colours.

So she's uploaded this particular design sketch and she wants to see, right, what happens if I tried this out in different materials? And the AI generative programme has given her that in a polymer, a timber, and also a metal.

Now it's not perfect and you'll see, and if you've come across AI images on the web before you'll see that sometimes AI gets it wrong, and you can see that it's got a little confused with actually the bottom of this design because of those arrows that were in place.

But what it has done is given her a really good idea of what this would look like in different materials, which really helps her with her decision-making.

So Laura has now explained her further decision.

So, "Using AI helped me to quickly visualise my design in different materials, and it helped me decide that this design would look better in timber to fit with a natural look." So make sure you're learning something when you use AI.

Don't just do it and then have an outcome and kind of move on.

Learn something from it and use it to help you develop and make design decisions.

So let's look at the third method then, and this was uploading our own sketches to view in an environment.

Now this can really help to actually visualise what our ideas would look like in the intended environment it's meant to be in.

Now Laura has uploaded this particular design sketch from her developed design ideas, and AI has come up with this rendered version of the sketch in an environment, and she's suggested the environment she wanted it to be in.

So she's put the sketch into the AI website, followed by some prompts, such as "living room," to then see a fully rendered version of her sketches in different environments, and also you can ask it to produce it in different lighting as well.

And Laura's just explained kind of what she's learned from this.

So, "Seeing this idea in metal and also what it would look like in a suitable environment has made me like this idea more.

It now looks less like octopus tentacles and more cool looking.

It really suits the environment and looks really nice hanging from the ceiling." Now, I agree with Laura here.

Her original idea did look very tentacle-like.

Obviously that was her inspiration when she was looking at polymorphism, but it didn't quite look right.

But actually the AI version has kind of made it more stylized, and I wouldn't now look at it and think octopus tentacles.

I would now just think of it as a cool-looking design.

So it's time for you to have a go at this task then.

You're going to use a generative AI programme to further develop your own design ideas.

Now you can use any of the below methods to develop at least three of your ideas.

Let's just remind you of what those three methods are then.

The first one was using prompt words to generate ideas.

The second one was uploading your own developed sketches to then further iterate or develop or change them, learning from that.

And the third one was uploading your own sketches to then view in a different environment or possibly in different lighting.

Okay, this is quite a big task, so pause the video here to crack on with this, and good luck.

Right, how did you get on with that task? I'm gonna give you an example here of what Laura did.

Now, she chose to use method three for hers, and she has done some uploading of her own sketches to view in a particular environment.

So you can see here Laura uploaded the design sketch on the left, which was this idea that went across the table and was hanging.

But she wanted to see what that would look like actually in a window and would it take up too much space in the window and also what would the lighting kind of look like.

So here she's managed to generate an AI rendered version of her sketch.

But remember, she's also gotta learn something from that.

So here Laura's just explained, "I've uploaded one of my developed designs to see what it would look like in an actual window and if it would block the view or look right.

I'm worried about the accessibility of it now that I've seen it like this." Now remember, she's trying to make a herb planter that is accessible for wheelchair users.

And actually, by avoiding it blocking the window view, it's ended up sitting quite high in the window, so I can understand why she's now worried about that.

So remember, when you are using AI, if you use any of those methods, make sure you are learning something from it and that you are reflecting on what you've actually learned as well, and how's that gonna affect your future designs and how your design is going to move forward.

So let's just summarise today's lesson.

We were looking at AI to develop ideas.

An AI, artificial intelligence, is the simulation of human intelligence in machines that can learn, reason, and solve problems. And using AI when designing can really help us to develop our design ideas, but also avoid design fixation and iterate ideas.

Now, it's not cheating to use AI, but remember that golden rule.

As long as you are open and honest about your use of it and do not claim AI work as your own.

If you got found out doing that, that could cause problems, so just always be honest and open.

And designers just need to be conscious of things such as bias and stereotypes that can occur when using AI to help generate design ideas.

Thank you so much for joining me for your lesson today.

I hope you've learned a lot and I hope you continue to use AI to further develop your design ideas.

I will see you next time.