Welcome everyone. Thanks for joining us today. My name is Amy and I'm from the Access and Learning team here at the National Portrait Gallery. We're currently joining you from the unseeded sovereign lands of the Ngunnawal people. And I'd like to pay my respects to the ongoing custodianship of the lands, waterways and skies of where I was born and currently stand. I'd also like to extend my welcome to all Aboriginal and Torres Strait Islanders joining us today and encourage everyone able to use the chat function of Zoom to let us know which country you are joining us from today. Some Zoom tips just before we start today's program. Being the National Portrait Gallery, we love to see your faces and classrooms. If you are happy to keep your cameras on, that would be great, but we do ask that you stay muted. We also recommend speak of you. So now I'd like to introduce Jess Herrington, our facilitator for today's brains versus computers science week workshop. Over to you, Jess. Hi everyone. My name is Jess. I am an artist and a neuroscientist. So the sort of art that I do is digital and neuroscience is a study of the human brain. So in this workshop today, we will explore the fascinating similarities and differences between our brains and computers. Then we'll dive into some fun and creative art activities, which will help us understand these concepts better. And finally, we'll end with an exciting challenge. We'll design an artwork inspired by what we've learned about brains and computers. And I can see that you've come from all around Australia today, which is really exciting. Thanks for writing in the chat where you're from. Okay, so our brains and computers are pretty incredible. So there's some things that are similar between brains and computers and some things that are different. And you might know some of these or you might not. So as you know, both the brain and computers use electrical signals to communicate. So as you can see in this image on the side here, there's a picture of a human brain, which looks like it has lights in it, but human brains don't really have lights in it. It's pretty dark inside there. If people always draw it as lighting up because they're just trying to illustrate that idea of electrical signals. But right next to that brain is a picture of a neuron, which is a special type of cell that exists in our brains. And it's what our brains are made up of. Neurons are really, really tiny. In humans, you can't see a neuron with just your eye, you need a microscope. So this is kind of what our brains look like. This is a plastic one. So the really cool thing about the human brain is that just like a computer, we have what's called memory. So memory in a human brain is about connections between those neurons. And there's a saying that neurons that fire together, wire together. So what that means is the more you think about something, the stronger that connection becomes. The more you do something, the stronger that connection becomes. So maybe you want to practice something like tennis or soccer. The more you practice it, the better you'll become. And that's a really cool human thing. So computers don't really work like that. Computers store data in chips. So while humans are capable of learning and adapting to new information, we can learn new skills, we can do new things, pick up new things in our environment. Computers use software updates. Another really cool thing that I want to tell you about the human brain, which is pretty unique, is you'll notice that the human brain has all these funny wrinkly, wriggly lines around it. So these are called sulci. And that's a really important feature of human brains, because it allows us to fit a lot of information into quite a small space. So if you were to look at another type of brain of another animal, like say a rat or a bee or any other animal that can't do more complex things, like drive cars, do maths, that kind of stuff, their brains are a lot smoother. So even though there are some animals that have much larger brains, like elephants and like whales, they don't actually have as many wrinkles, they're not packing as much information in. So that's a pretty cool and unique human thing that we've got going on. So humans are also different to computers in that computers, as you'll probably know, process information with ones and zeros. It's quite a different sort of way of processing information. It is kind of a little bit hard to understand and it's a little bit abstract, a bit different to how we usually experience stuff. But I'll just go to the next slide. We'll see what's there. So artificial intelligence you might have heard of before. It's kind of been in the news a lot lately. Lots of people are talking about it and talking about how it's going to change our world. And it's a pretty important scientific development because it's like giving computers their own mini brain. And that's really exciting because it means that all of a sudden computers and computer programs can learn and adapt a little bit like humans do, not as well as humans do, but they can start to make better decisions. They can do things like drive a car. So you might have heard of self-driving cars where the computer in the car is able to see the traffic in front of it and avoid danger and go around other cars without having a crash. You might have also used AI like Siri or Alexa where it can chat to you and communicate. But the really important thing is it's different to people because it can't understand the information like humans do. So humans are pretty unique and pretty important and can really understand what the information means. Computers just follow instructions and that's what AI does as well. So yeah, AI helps computers adapt and change based on the information they receive, just like how our brains learn from our surroundings. And yeah, so someone in the chat has actually written that the octopus famously has quite a smooth brain which is correct. They do. And the octopus brain is pretty unique as well because it's actually spread quite a lot throughout their body like a starfish. It's quite unique in that sense as well. So like a starfish is pretty cool because they don't have a centralized brain in the middle. They'll have like sort of like a ring and then their brains actually extend into their like tentacles which is pretty amazing. All right. Cool. So a really important thing is to talk about like how computers see versus how humans see. So a lot of what science tries to do I think in terms of understanding how we move around our world and understand our environment is sometimes some scientists might try and replicate different human senses. So the sense of touch, the sense of sound and sight and smell. So sight is probably one of the easier ones to try and replicate and that's where people have gotten the furthest really in terms of the science. So with a human our brain interprets images and stuff that's visually around us based on our experiences but we also connect that to emotions, things that have happened to us before like you know if you might have been somewhere and you've been swooped by a magpie and that was a horrible experience for you. You might avoid that place later on. Computers don't really do that. We also have context. So we might know that the magpie is going to swoop us in spring time but not in winter. So it's okay to walk through this place in winter but not in spring. So that's one example. So when we look around our space we see more than just shapes and colors. We actually see stories and that's a really important part of human experience. So computers are different because they process images pixel by pixel which is what images are made up of on a computer. Tiny little dots and they analyze patterns and colors but without that emotional connection. So I'm going to start off with getting you guys to do a little activity. I'd like to usually use drawing materials to copy a special artwork from the Portrait Gallery collection. So I'll share the artwork details on the next slide but this is something we're going to do with real physical materials. So whatever you have to hand you can use paper, you can use cardboard, you can use pens, pencils, text, whatever you've got. That's perfectly fine. So what we're going to do is try and copy a drawing as best we can and then we're going to compare those drawings that we've done with our friends or with other people on the chat and talk about why our drawings might be similar and why they might be different and how our emotional connections and our stories might affect the way that we draw. So this is the artwork from the Portrait Gallery collection. It's called Self Portrait with Glove. It was done in 1939 by Herbert Badham. It's an oil painting and you can see it's a man smoking a pipe wearing a special scarf called a cravat and pulling on a glove. So interestingly the artist might have special connections of why they painted this particular collection of objects with this person so you might like to think about that. So if you can get your drawing materials ready and give it a go see if you can copy this image as well as you can and then we'll see the differences that everyone has. So I'm just going to be quiet for a little bit but if you have any questions in the meantime just pop it in the chat. Someone said that's so hard. Yes it totally is but that's also kind of the point right. So your artwork is going to be totally different to this artist even though you're trying to copy the same thing. So just do as well as you can and if you want to change it up and make it a little bit unique to you like drawing objects that you care about you can do that too. So someone in the chat has just asked can it be a person like that or the same person? The interpretation is open to you. I suppose what we're trying to do here is copy the same sort of subject so it's a person sitting in a chair pulling on a glove maybe smoking a pipe but the interpretation will be unique to you. You can put your own style and spin on it. If you want to draw a woman sitting in the chair you can do that too. You can draw yourself sitting in the chair. Someone's written in the chat is this hard but maybe they meant to write this is hard. Yeah it is but I guess the cool thing is that next we will do an activity where we play around with an AI system and it will be a little bit of a different experience. Do you have to do the background is something someone has also asked in the chat? Yes I think the background is pretty important because there's lots of objects in there that might have hidden meaning. I can see some books. I can see a woman sitting in maybe a mirror or a painting in the background. Someone said what happens if you don't do it? That's perfectly okay you can just sit and listen and have a think about it. How you would draw it. Interestingly someone has said hands are hard to draw. Yeah they are and that's a really interesting point because AI systems find hands really hard to draw too just like a person I suppose. The really interesting thing about drawing hands for a computer is that you have a left and a right hand but which looks similar but different and a computer doesn't really understand that so well. You know when we draw faces we sort of have two halves of the same face and it's sort of quite symmetrical but hands are sort of opposites. So computers get that wrong sometimes. Too interesting effects sometimes you might end up with more or less than five fingers. Someone said their hands look like bananas. That's wonderful I love it. If anyone has any drawings that they'd like to show up put up to the screen. Oh I can see an amazing one there. Pretty cool. Yeah wow. Amazing. Wow you've done so great all of you. So many. It's a pretty hard job though isn't it trying to learn to draw a person isn't it and especially copying something that already exists. There's a lot of detail to take in but you guys have all done amazing. Wow and I can see some of you chose to use color and some of you chose to just use pencils. That's amazing. Well done. That's great. Okay well we might move on. Okay now we're going to spend a little bit longer on this activity because I think you'll need it. So what we're going to do is we're going to now create an artwork using AI which is going to be pretty fun. So this the idea of this activity is to really think about what you just drew and what the original artwork is and you're going to try and prompt an AI system to create the painting that you've just seen. So I'll show it again after this slide so you don't have to just try and remember it. So yeah we're going to create a unique artwork. We're going to use an AI system like Darlie but if you have another one that you prefer to use that's totally fine. There are other options available. So we're going to try and recreate an image like self-portrait with glove. So if you want you could create your own special version of the self-portrait with glove. You could think about what you'd like to include. Maybe there are some special objects that are meaningful to you. Maybe you'd like to have the person represent you. Or you could try taking a photo of your hand-drawn artwork using a phone if your parent or teacher is able to help you and you can upload that to Darlie and generate variations and see what that looks like using the AI system. So what we're trying to remember is to try and blend our own unique human creativity with the AI to create interesting and inspiring results. So if you don't want to use an AI system or you have extra time you can instead create a unique drawing yourself using pens and paper that highlights the differences that you've learned about brains and computers or come up with an interesting design. So someone in the chat has just mentioned that they don't have Darlie because it is no longer free. It's free when you first sign up. There are options. You can either sign up with another email address. That'll give you another free account so you'll have credits. If you don't have Darlie and you don't want to use that one there's also Bing. There's an image generator in Bing on Microsoft Edge but you do have to install Microsoft Edge. And if you don't have access to these AI systems you can do the alternative activity that I just mentioned. Cheers. [no audio] Okay, I'm going to demonstrate one of the examples. Can I share my screen, Hector? Cool. So this is what Darlie looks like. So you can have a little look at how I'm using this system if you'd like. So what I'm going to do is actually... I'm going to... Sorry, can I just go back to the slides? For a second, I do need to show them the image again. Great. So this is the sort of image that we're trying to recreate. And up here, I've put an example prompt. So the prompt is the thing that you put into an AI system. It's a string of words. Sometimes it's crafted as a sentence, but by no means do you have to have it as a grammatically complete sentence because computers actually understand language quite differently as well. So here I've written realistic painting 1930s self-portrait with gloves sitting in chair with pipe. So now I'll switch to the AI generator and I'll show you what we get when we put that in. So we'll generate. [silence] All right. So this is the kind of results that we're getting. So we're getting something that's in the style of a painting. It looks like a man sitting in a chair. He's smoking sometimes something that looks like a pipe. But as you can see in this second image, it's kind of a little bit wibbly wobbly. Like a computer doesn't really know what a pipe looks like. It's kind of doing an approximation of what one is. And it does that by analyzing all the different images of a pipe that it's ever seen, that are on the internet, and it synthesizes, it creates a pipe that it thinks exists. But sometimes it gets things a little bit wrong. So we might want to change it up. We might want to have it look like a woman or have it look like someone else. [silence] There we go. So we've got a few other options. We might also want to do something like change the style to like a surrealist kind of style and maybe make it hyper-color. [silence] Let's make it hyper-color rainbow just because that's kind of fun. See what that looks like. [silence] Yeah, so this is pretty cool. I think I especially like the one on the end here. That's pretty nice. I really like how the rainbow's been broken down like this. Let's make it a little bit darker. So I'm adding the word dark here. So that will change the overall aesthetic as well. [silence] That's a bit nicer. I think a bit more aesthetically pleasing. So like I really like this one on the left for example. That's really awesome. But her face has lost quite a bit of detail. So we might want to add some more detail in there. So you can even add things like highly detailed face. [silence] [silence] Yeah, so we've brought back a bit more detail in the face here. Like especially in this particular one. Like we've moved away quite a bit from the original image and what we were trying to achieve. But I suppose what we're trying to achieve is maybe looking at the composition of the original image, breaking that down into how a computer might understand those individual components. And then adding our own spin. Adding back in the emotion, the interesting parts, and making our own images. [coughing] Excuse me. [silence] So I'll stop talking for just a little bit and leave you to experiment with this for five or ten minutes or so. And then I'll jump back in and tell you a little bit more about prompt engineering. I can see some of you are holding up your drawings. They look amazing. Some of the people who did physical ones. And I hope that the people who are using AI generators are having a good time too, working together to come up with some interesting artworks and interesting ideas. Amazing. These look great. Well done. Very cool. So I will tell you while you're working on this a little bit more that using AI generators is a bit of a mix of art and science. You can understand about how human brains work. You can understand about how computers understand language. But sometimes even when you know those things, it's a little bit of an art to try and get what you want out of a system like this. So the job of doing this sort of work is now called prompt engineering. So there are people who are professional prompt engineers who just work on understanding AI systems and trying to get the best out of them, working out what the problems are. And yeah, some people have jobs working for companies doing this kind of thing, either with language generators or art generators. Sometimes another tip that I have when I am working with these sorts of AI tools is I focus on texture quite a lot. That's a good one. So something that computers do know how to do is represent textures. So I would talk about the texture of wood grain or the texture of a linen shirt or those sorts of things, texture of wallpaper in the prompt. And that can sometimes get you better results as well. So we'll just go for a couple more minutes, I think. And I will just mention as well, if you can't use Dali, you can use Bing Image Generator, which is free and it's based on Dali. The only thing is you have to use Microsoft Edge, which is a specific browser. But it is free. So how are you all going? Did you come up with some interesting results? Yeah, does anyone have anything? I guess you can show me your physical drawings on the screen. Well, does anyone want to write in the chat or post in the chat what their image looked like? Can you post images in the chat? Yeah. Amazing. I can see some great drawings there. I also saw someone dancing like a robot. I thought that was pretty good. Getting into the vibe of AI. You all look like you're having a lot of fun. [silence] Oh, I see there's an interesting one there. Lots of interesting work going on. Alright, well maybe we'll move to the next slide. Okay, so do you think you can maybe, can the groups drop in the chat, tell us a bit of a story about what you created and what inspired your choices? Maybe what sort of words you used in your prompt if you were using AI? I can see a very cool picture of a brain up there. And I think what I want you to know is all the different styles and interpretations are quite interesting. No two pieces will be exactly alike because even though they're created by machines if you're using AI, we're using human input as well to drive that. So we all have unique viewpoints and perceptions that make our world a rich tapestry of experiences. Would anyone like to write the sorts of words that they put into prompts? I like to put words that represent different painting styles sometimes. Sometimes that's a good way, sometimes you can write oil painting, sometimes you might write charcoal drawing. Oh wow, I can see that someone did a physical drawing with a mixture of animals. That's a good one. Yeah, because we did talk about how different animals have different types of brains and what they look like as well. So human brains are quite wrinkly and then other brains can be quite smooth. Good choice of animals there. Anyone else? Ah, I can see someone's done a great drawing there of a person sitting down next to some books. That looks a lot like the person with a glove. Oh, I can see you there, yeah. And isn't it interesting how all our drawings look a little bit different? So we bring our unique human perspectives there. Amazing. I see that a lot of you have drawn the bookcase next to the person, so you've picked that out as being pretty important. So that's an interesting feature of humans too. So we're really good at looking at an image and realising what's important because we have that idea of storytelling and we tell ourselves stories to know what's important in our environment and what we need to pay attention to. You know, a computer might not know those sorts of things and so we have to tell it what to focus on. I see there's a lot of detail in some of these. Amazing. You've all done really well. Someone didn't put hands on theirs. That's okay, hands are really hard. I don't know if I can draw hands either. Amazing. Oh, I see a nice close-up of a face smoking a pipe there. I like that one. I love the glasses, love the hair. Oh, another one. Great shirt on that one. Great shirt. Oh, and is that a sun up in the top there? Yeah. I love it. You know, what's really interesting as well is when you look at all these drawings, right, we all focus on this idea of a very central figure in the face. So that's really important to us as humans because what we're doing is, when we look at a painting of a person, we're trying to understand their body language, what they might be thinking, what they might be feeling. We're really actually trying to do, we're trying to get inside the mind of the other person and understand what they might be thinking and feeling. Even if we don't know it, that's actually what we're doing. All right, let's continue on. Okay. So I guess like a really important point of this is that AI systems, as impressive as they seem to be and as cool as some of the images are that come out, they really are unable to create art by themselves. So a human, a person is required to drive the AI, to put in the prompts and to get interesting results out. And even after that, you know, you saw that when I use those AI systems, four results would come out at a time. And it's up to me as a human person to choose which artworks are the best and then work on that. And it's called iterating. So you might choose an image that you think is good and try and build on those results. And as you feed more prompts into the AI system, it learns from what you're wanting. So it kind of changes that model of an AI system and changes the sort of output that it creates and drives it in a certain direction. Cool. So someone created a rainbow mag pie in spring. I love it. It sounds beautiful and dangerous at the same time. Someone didn't put feet on their person. That's okay. I like to just cover feet in shoes because feet are really hard to, not as hard as hands, but they're pretty hard. Hmm. Amazing. I see that some of you did a lot of drawings on green paper. I love it. I love green. Oh, and I can see someone's holding up work that they did with an image generator. That actually looks like a really beautiful texture. Is that right? It looks amazing. I like creating textures with AI generators too. That's actually one of my favorite things to do. You can get really amazing looking textures. Some of them come out really detailed. It's really cool. So some of the, one of the interesting questions that's just come out in the chat is what could be some of the ethical issues to consider around AI generated art? So that's a really interesting question and I'm involved in a lot of that kind of work. So the thing is, it's probably a little hard for the students to understand, but copyright issues is one. So in Australia, you can't copyright AI generated work. I mean, that's okay. It just means that it's all open commons, but it depends on what you're using the output for. Another issue is some artists feel like AI generators aren't creative or are stealing from other artists. I don't actually feel that way, but there's a few different issues there. If there's anything specific you'd like me to address, just pop it in the chat and I'm happy to chat about it. But one of the reasons why I suggest to Dali or Microsoft Bing, even though you have to pay for it, is actually because it's one of the safest image generators to use. It has a very high filter on it so that all the images that you get out are going to be very child friendly. So that's why I've chosen it for this reason, for this class. So for the kids, something that might be really cool to think about next is, what do you think an artist of the future might be like? I mean, what do you imagine if you're going to be an artist when you grow up, what sort of tools do you think you'd be using? Do you think you'd be using computers or do you think you'd be using pencils and paper? And you can write those answers in the chat too. I know for sure when I was a kid, I didn't think I'd be using computers to make art, and now I do it every day. So maybe an artist of the future might be making art in space. Maybe an artist of the future might be making art in virtual reality, or maybe they do something else that's completely different that we don't even know about yet. If you have any ideas of what you think artists in the future will do, pop it in the chat. I'd love to talk about it. So in one group, the majority of the class voted for pencils and paper. Good one. I mean, they'll always be easily available, won't they? And another group thinks that robots will do it for us. That'd be pretty cool, pretty cool. And another group thinks that it will depend on how you're feeling and what you want to achieve, which is a very good response. Do you think that robots will ever be able to paint, like create real physical paintings? Yeah, somebody thinks that robots paint and paint brushes will be the way. Do you think that we'll ever see paintings in an art gallery that are just painted by robots? Do you think there'll be a whole different portrait gallery, which is just portraits of robots and portraits painted by robots? Some people think that robots are already serving us food in some restaurants, so they'll be definitely doing paintings as well. Someone has said it wouldn't be the same. Now, I wonder why that might be. It wouldn't be the same if a robot did paintings for us. I wonder why. Do you think it's because we bring our own stories and connections to an artwork, and it's really important to see the human point of view? Or do you think it's something else? We put feelings into our paintings. Yeah, I agree. Someone's also said robot artwork will go into galleries because it is still artwork and it's still important, even if you're not human. That's a very interesting point of view as well. Some people believe that robots will be creating paintings and we will care about them and the robots just as much in the future. Yeah, people say robots don't really have any feelings. Someone's mentioned that in the chat. Someone else has said in the chat that it would be interesting to see ideas through robots' eyes. I agree. I think that would be really cool. We have robots around in our everyday life as well. Some people have those little round vacuum cleaners called Roombas that vacuum up all by themselves. That's pretty fancy if you have one of those. They're pretty fun. I saw an artwork once where someone taped a paintbrush and painted with a Roomba just by having it drive around on a big piece of paper. That was pretty amazing. They had a few Roombas in the gallery and they were all driving around and painting, doing amazing squiggly work. I'm glad that some of the students in the chat are highly engaged in drawing their own creative pictures with pencil and paper and not considering future technological advances right now. That is okay. [laughs] Okay, well, you've all done such amazing work today. I might open up the chat to just general question and answers if you like. Well, so one student in the chat has noted that once someone taped a banana to a wall with some duct tape and that was considered a masterpiece, surely robots can do better than that. I'm not sure. Maybe they can't. Maybe they'll just continuously draw humans with banana hands. I don't know. Some students in the chat think that robots might take over the world. It's possible. I hope not. I don't think so. I think robots will always need people to do the big thinking, the big picture stuff. Someone thinks that humans will make invisible cars. Yeah, invisible cars do sound kind of dangerous, don't they? Yeah, does anyone have any general brain or computer or general science questions for me? You can type them in the chat if you do. Or any questions about AI systems or just anything that's come to your mind from doing this workshop. What is 2 divided by 0? I didn't say maths questions. Ah, so someone has mentioned in the chat that maybe robots will infringe the copyright of humans. Okay, and someone's asked a really great question. So how do the computers know what to do? So computers work by running programs and programs can be thought of as running a set of instructions. AI is a little bit like running a type of program, except instead of it being a really specific list of instructions where we know what happens, it's dynamic. It changes depending on the scenario. So the instructions could change from second to second. And that's how we get generative AI, which is how we get these really interesting images that are made with AI systems. So there's a general base sort of instruction, but there's also a lot of changing elements, variables that change the output. So someone's also asked, is AI controlled by humans or computers? And it is controlled by humans. It doesn't really do anything by itself. It requires humans to press the go button, essentially, and to write the instruction or the prompt that gets interpreted by the program. And someone's also asked, would AI be able to understand human feelings if we program them to? That's a really great question. And no one really knows. Computer brains, well computer brains, computers think so differently from human brains. It's hard to know whether they would ever be able to understand human emotions. So the way I think about it is like it's kind of like the difference between an analogue clock with hands and a digital clock. They actually sort of do the same thing, but the way they work is completely different. And they're not really related in any way, except that they can tell you the correct time when you want it, essentially. So I think, you know, whether AI systems will ever be able to understand emotions, I don't think so because I don't think they have the right hardware. So let's see. Someone has said, will AIs be able to rule the world? Some people are a little bit worried about that. It depends on who you ask. I don't think so. Oh wow, I can see another AI generated work there with a rainbow. It looks like a rainbow carpet or a rainbow road. Looks amazing. So some students say that robots won't feel human emotions because they don't have our experiences and they don't have hearts. So it's really not, it's true. It's not really about the heart, I suppose. It's really more about the experiences. They can't put stories together. It's not really the same sort of thing. Yeah. They're smart, but they're not like human smart. Humans are very creative. AIs seem like they're being creative, but they're really just putting out ideas that they've already seen before. Okay, I think I'll do one last question. And couldn't AI help humans do everything? I don't think they can help humans do everything, but I think they can help with a lot of tasks that we might find boring or repetitive. I think that AI could do those sorts of things really well. Like filling in a form for us. They could maybe tidy up for us. They can do things that are easy to write instructions for. Cool. Well, I might bring Amy back to wrap up. Hi, everyone. Hello again. I've been watching your amazing creativity. How you all go in, my little scientists. Wonderful staff. I'd like to say thank you so much for everybody for joining us today. And if you can, wave your virtual hands and thank Jess for doing a big thanks and a wave. That would be wonderful. Those people that haven't been able to put on your cameras, you're more than welcome to do their reaction buttons and give Jess a lovely wave and a thank you. On behalf of the gallery, I'd like to thank everyone for their time in particular Jess. So we'll leave you now. But if teachers, if you have any comments or need any resources or anything, please contact us here at the gallery and yeah, have a really good rest of your afternoon. Bye everyone. Big thanks for Jess again. Happy science week. Bye everybody.
