From Ashes to Art: How AI is Reshaping Museums, and Culture & Heritage

Priya Patel: Welcome to the Arts and Everything in Between podcast. I’m your host, Priya Patel. Today we are talking to Lucia Cipollina Kuhn. She is a final year PhD student at the University of Bristol, working on AI and machine learning, specifically within the arts, culture, and heritage space. Lucia is going to be sharing her work on AI and machine learning projects within the museum, culture, and heritage space, specifically on how AI and machine learning is evolving. In restoration and preservation projects and beyond. For example, Lucia talks about AI’s use in mapping heritage and culture sites and assessing and cataloging collections. And even uncovering previously unaccessible information. For example, from Pompeii’s burned papyri, [00:01:00] to unfinished masterworks. Lucia also shares her experiences working with the Tate Britain’s AI project, Which helped increase audience engagement and interest.

Hello and welcome to the arts and everything in between. I’m your host Priya Patel. Today we are going to be talking about AI’s role in arts and heritage preservation, reconstruction and restoration. And we have a very special guest today, lucia Cipollina Kuhn, and she’s a final year PhD student at the University of Bristol and a member of the Lion AI research group.

Her research focuses on the application of computer vision for the restoration of cultural heritage, in particular damaged frescoes or paintings. Her work has been displayed at the Tate Britain in 2023 and has been published in academic conferences such as CVPR and ICML. She’s won several research awards such as the Helmholtz Research Grant in 2023 and a British Telecom PhD [00:02:00] grant in 2020.

So we’re delighted to welcome Lucia to the podcast. Thank you, Lucia, for taking the time to talk to us.

Lucia Cipolina Kuhn: Thank you very much for having me.

Priya Patel: So Lucia, I talked briefly about your bio, but I wondered if you can expand on that a little bit more.

Lucia Cipolina Kuhn: Yes. I think with the search of AI everyone is drawn into it and then the applications are infinite. We feel like now everything is AI and it’s a blooming. Feel now now AI is used into fields that we never even imagined. And you know what they say, like the job of the future doesn’t exist today.

And I think AI enables all that to create endless possibilities or things that we haven’t thought. So my research is a combination of AI from reinforcement learning that people may. remember AlphaGo achievement to then more apply things like how do we use AI [00:03:00] to apply problems and maybe for social good, what can AI do for us in practical matters.

So that’s how I became interested in art restoration, art preservation, And then as an extension of that cultural humanities beyond the restoration and the preservation of art, that will be maybe buildings or objects for example AI in a drone. that helps us measure something or take pictures.

You use computer vision to learn techniques to maybe say, for example, we use techniques to distinguish, whether a painting is real or fake, that’s something that we can do with AI or at least guide us. Is this a real Picasso or this is just a forgery?

Questions like that. . Yes, and recently some others have made them discover hidden paintings without damaging the original one. So you can have a painter [00:04:00] painting on top of another, and then you have two paintings in one. So that’s another interesting application of AI.

Priya Patel: Did you have a background in art or painting or art restoration before or is it is this new for you? The art restoration side of it?

Lucia Cipolina Kuhn: My background is in mathematics, so I started with Escher, which is a painter that combines both into completing some of the paintings that Escher left unfinished and see how we can restore it.

There’s a very famous one that was very famous painting. It’s actually a lithography by Escher called Prince Gallery that features Malta. It’s not very well known, but it has been a mystery since 1956, since he died. And then left that painting and finished. And because the mathematical structure behind all of Escher’s paintings, there’s a mystery of how come AI saw [00:05:00] something that has some mathematical pattern.

So it’s a combination of math and AI that translates into art. That was the first one that we started working as a challenge. And then from there we, we continue forward.

You can see some of my restore works in. A website called Art Restore AI, so people can go and then see our proposal for restorations. And then you can see the before and after. That’s a good way to grasp what AI can do for the restoration. And look, the idea, I think is to have several what AI can offer for us is to have alternatives of different visions on how something can be restored without being damaged.

So preserving what we already have and then filling. What has been restored. So it lets you [00:06:00] make sketches or blueprint on deciding on. It gives you ideas based on the knowledge that he has from the painter or the school, or sometimes it happens in frescoes in Italy that you don’t know the painter.

You don’t know how it looked. Before so then the day I can tell you well, these are the partners to follow. Sometimes you have an expert. You can have a expert in Italian frescoes telling you this is the way you should have. It should be restored. This is the color pattern because in this region there was this route.

But sometimes you don’t have the expert or you might want to have several different candidates. So AI can bring you. candidates for restorations. It can help you on the, on your research to say, look based on this school of art, this should be the colors. Like for example, when we propose a completion of a Cezanne, that’s a painting that Cezanne left unfinished, and you can see the [00:07:00] cardboard.

Cezanne had a, very particular palette of colors. So it has to be within this range, so it gives you an informed candidates for you. So you may give you 10 and then you select the human may select 10. But I think the greatest achievement is with restoring without damaging. So it’s a combination of reading from a scan so you can scan A painting.

For example, if you want to see what was the painting that was behind that’s something that you can do, scan the painting and then reading with machine learning you, you are able to read behind without damaging the piece.

Can apply different race and then you, they, I, you read the scan. That’s something that is part of the papyrus challenge as well is to read what is what’s in there without destroying the original.

Priya Patel: So it sounds like there’s sort of two aspects to it.

One is preservation, I think, and the other is [00:08:00] restoration or at least giving humans a guidance as to what the restoration. Should look like ? Or is that?

Lucia Cipolina Kuhn: Yes. So to restore any object, it really depends what information you have about the object or surrounding it, because usually to train a model, you have to give the model information and think about the model is, I like to think about it as a brain with infinite memory. So if you have infinite memory, and I show you all the Picassos that they are in, you will be able to tell me a pattern.

Or sometimes you may be able to tell me Picasso had this many eras. So in this era, he was using blue, in this era, he was using red, but then you become an expert on Picasso, and then you would probably be able to identify Picasso from now on. And if I give you a broken Picasso, you will be able to tell me, oh, this is how I would [00:09:00] fix it based on my knowledge.

And if you have infinite memory and infinite time, you can become an expert on any author just by seeing it. And then you will have this capacity of synthesizing what it’s, The features that certain element has so with machine learning is the same thing. It’s just something that has infinite memory processing power, and it’s able to identify patterns and features.

So to synthesize. in the case of Cezanne, what are the colors? What, if it’s a Cezanne, which colors, what’s the palette, which color should be in, and things like that. And if it’s an Escher, it will be the same. It’s it has seen so many of Escher, the school of Escher expanding more.

It’s not only Escher, because sometimes there are not enough paintings, [00:10:00] just usually a painter wouldn’t. wouldn’t do like a thousand pieces as you only have a hundred. So you have to tell the model look, this is how an Escher looks like. This is how the Escher school looks like. This is how I want you to also memorize this because this is close enough to Escher.

So once he has seen several examples, then he can infer, ah, okay, I get the gist. I get the General pattern.

Priya Patel: Wow. So rather than having a Picasso expert, you could have the AI be multiple, whether it’s Cezanne or Escher or Picasso or whomever. That is incredible.

And. Within sort of a process like that I presume that, there’s a human involved in validating what the AI comes up with.

Lucia Cipolina Kuhn: Yes. First, I think there is humans annotating the data set because this is like what is called supervised learning. So you need to provide an example and tell, okay, this is a share and this is not an issue.

So that’s [00:11:00] one thing that we did to then convert this into a classification problem. Okay, I know what an Escher is, I know what is not an Escher. So for that you need what is called label data. That for us it was just wiki art, like publicly available free information. Because in wiki art you have lots of painters with labels saying, okay, this is a Rembrandt.

This is the Rembrandt body of work, and this is Rembrandt’s genre, for example. So that’s something that humans have to collect, and it’s a big issue in machine learning, the availability of data and label data. Yes, sometimes you don’t know, and it’s easier to say to the machine. This is an issue.

But sometimes, as I said, with frescoes, you may not know who’s the author. So you have to go to the school or how frescoes look like. And then. For example, our models, there is a human in the loop at the end. The model will give you candidates with certain [00:12:00] probability. I am 90 percent sure that this is how an Azure of a Picasso should be restored.

And then. You have also this other possibility with an 80%. And then the idea is to give a restored candidates with several degrees of confidence.

Priya Patel: Okay. Okay. Oh, so interesting. They’re getting a lot more input to be able to understand which direction to take it.

Lucia Cipolina Kuhn: Yes. So you get like a distribution of candidates because, the more damage also a painting is, the more degrees of freedom to restore it.

And in our website, we have paintings that have been burned, say 20%. Or you have, in the case of Cezanne, big patches of cardboard where, and then, you can put a house there, you can put a horse. Both will be correct. So that’s the idea to give you options and candidates with certain confidence interval of what’s more, more likely to be there.

Priya Patel: Yeah,

Lucia Cipolina Kuhn: [00:13:00] because also think that AI also involves robotics and For example in museums when you think about the Natural History Museum, and they usually deal with very little species like they may have inventories of species of bees for example that sometimes they are unlabeled or they’re labeled by hand 50 years ago, 100 years ago, whatever.

So the handling of that the label all can, that can be automatized by robots. That’s what the Natural History Museum in Britain is doing. Including slowly introducing robots to handle Something that is very small that you need to handle with care and that you may have thousands of very little of them.

So you’re not going to put a human to label the thousand collection of bees that you have. The time that is better, better spent. So if that includes robotics or that includes in the Pompeii Park drones to survey the growing of [00:14:00] grass, for example. So that will be preserve what we already have from Pompeii, because once you dig into something, then you need to keep it as clean as possible.

So that’s also done in archaeological sites, monitoring with drones and 3D scanning, or it can be infographics as well for a museum, and sensories, virtual reality, that as well.

Priya Patel: That’s really interesting. I was thinking about that even just sharing between museums and being able to then have all this data about all the different pieces

If you looked at that kind of across the world, you can really start to put together this amazing picture so it’s. It’s it sounds very exciting in that. And in terms of the other piece that you were speaking of, there was the virtual reality piece.

And I was interested in that a little bit more if, could you talk a bit more around that? It can be

Lucia Cipolina Kuhn: used. So for example, if you want to see. There was one exhibition of for example, pieces that burned. It’s very common sometimes, [00:15:00] in the past it was common, that actually currently in Bristol, I think one or two years ago, there’s a museum that burns because it’s just the paintings are made of flammable material.

So you can restore it or another option that museums have done is to showcase the burnt image and then on top of it you put VR glasses and then you see how it looked like when it was or even you can make variations of it if you want with the VR glasses and choose your own restoration that’s something that we’ve done in the tape that is people’s intervention on the pieces like virtually.

Priya Patel: Oh, yeah, that’s interesting. And are you In this work. It sounds like that. There’s a lot of really good goals that you could achieve around audience engagement and engaging new audiences or even accessibility. Have you seen that with the work that you’ve been doing?

Yes,

Lucia Cipolina Kuhn: because it allows you to interact with the piece. And I think it [00:16:00] provides you tools that. Are for everyone. The you don’t need to be an expert in a certain tool like with very little you can interact with AI that I think is the accessibility of AI for everyone. Like today with a prompt, you can just expand.

a canvas. That’s what we did at the Tate in the exhibition was to imagine that a piece, so this was the Tate Britain, right? So it’s the part of the, of art that has been more established and is more traditional. But you can still interact virtually with the piece and create, expand it, expand the field of view or create more artifacts in the painting.

And that will be with just a prompt. You don’t even know how, don’t need to, for example, use Photoshop or anything like that. So it’s, you can interact with AI more and [00:17:00] more with. At the beginning, you can do at least something with very little knowledge, so it’s the accessibility part of it that is engaging.

So with the Tate was a particular exhibition that we did for the re hang of the Tate last year. So there was a re hang of the collection and that was I think it was the biggest thing that we did was with the Tate on that re hang day that they had.

Because, yeah, organizing everything when you have the support of an institution makes things easier because basically the idea was that letting people come in for free and then they will have the computers there and we use so through Lion, I was able to get the sponsorship from Stability AI, which kindly helped me.

gave us free access to their stable diffusion models. So it was a collaboration between Tate, Lion, and [00:18:00] Stability, let’s say. And so with these diffusion models, they were able to, people were able just to toy around and reimagine pieces. Reimagine how this old, 100 year old painting would look today.

Priya Patel: Wow, that is, that’s really amazing. So the audience could just come in and mess around with the computer, try out some prompts, and get an image back.

Lucia Cipolina Kuhn: Yes. And you know what? I realized from that experience. My takeaway is that the entry point is very low.

So that’s one thing. So you just need a prompt to say Here, instead of a man, I just want a woman, and instead of a naked woman, I just want a woman with clothes doing field work, so something more modern or more updated. You can do that, but, so that’s a little bit of bringing traditional point of view to a more modern point of view.

That was the idea of interacting with. With the collection. Just [00:19:00] having an easy tool to express complex ideas.

Priya Patel: Yeah, that’s really great because I think sometimes, it can feel like here’s that old bit of whatever it is, either an old painting, a bit of pottery, or whatever it is, and it can feel distant.

Whereas this kind AI, it does feel like you get to get in there and. Play with it and understand it a bit more. It much more engaging. It sounds like, especially for, I’m just thinking of like kids getting dragged around a museum. This sounds way more exciting if I get to mess on a computer and kind of experiment a bit.

Lucia Cipolina Kuhn: Yes. And all the infographics now that can be done with AI and 3D printing, it’s technology put to the service of people and making things accessible with a lower cost.

Priya Patel: You mentioned about the Herculaneum papyri, and I wondered if you could expand on that a little bit more in the work that you’re doing with that.

Lucia Cipolina Kuhn: Yes, that’s also a collaboration between Lion and the Vesuvius Challenge. And there are several [00:20:00] people working on that. University work, this work from the University of Kentucky. So what the university did, the University of Kentucky, what they did is they scanned the papyrus. So those are, these are papyrus that were burnt at the same time that Pompeii was burned.

Aquilanium is a city that is next to Pompeii, but it couldn’t be fully digged because there are buildings. on top of Herculaneum. So it’s just very partially did. They recover many papyrus and they scan some of them, a handful of them, less than 10. And the idea is to read what’s inside through a scan without opening the papyrus because the papyrus are so burned that if you try to open them, you break them.

So it’s a technique that is called virtual unrolling and meaning without touching them. We can read something that is rolling to itself. And there’s [00:21:00] progress being made on that. So there’s. Some phrases so we know that like a library that was holding papyrus that were old for their own time.

So at the time that Eric was burned, those papyrus were already old. So that was a library. So this were papyrus from. That were stored in a sort of a library and a colonium. So they had poetry and interesting thoughts of their life at the time. And the idea is to try to read as many as we can without damaging or preserving the papyrus as much as we can.

So it’s a combination of scanning them with high precision scans and then using machine learning to. detect the burned ink.

The machine learning part comes when you have to detect characters mostly in Greek, even though it was a Roman village,

, they were mostly Greek because they [00:22:00] were old.

So trying to distinguish the characters from the burned debris.

So the idea is to first identify as much characters as we can. And then translate them. So there’s a whole team of so there’s a university and the Vesuvius challenge. They have people working full time in scanning and then doing the virtual enrolling. And then once you enroll, imagine it’s if you have something that is brought up to itself, and then you are really, but just. virtually just following

Priya Patel: the

Lucia Cipolina Kuhn: lines that, that they are roll out. And then, and the idea is that you read the characters, try to identify the characters first, because this, of course, a lot of just burn debris. Once you identify the characters, then there’s the team that translate them from Old Greek into readable language.

Priya Patel: Wow. I’d say there are so many excited people about that project.

Lucia Cipolina Kuhn: [00:23:00] Yeah, the challenge started last year, so there was one papyrus that was already read, and we know a little bit what it said so that was announced about, around December 2023, and now we are working in enrolling, reading others.

Priya Patel: Wow. That’s incredible. And you mentioned about the drones and mapping archaeological sites. Is that something that that you’ve seen

Lucia Cipolina Kuhn: Yes, I don’t use them particular, but that’s using in the archaeological park of Pompeii that’s used in Notre Dame as well, because the restoring Notre Dame is so high that then it’s safer to send the drone to.

Take pictures than to send a human. ,

Priya Patel: could you touch briefly, your experience have you seen have been some challenges around using technology like this?

Lucia Cipolina Kuhn: Yes these models are all what we call data hungry. They need to see. Examples, at least in the order of thousands.

Sometimes you can deal with less, [00:24:00] but these are data hungry and the availability of data, good curated data. Now with generative AI, data is a huge topic. In art, it’s easier because there’s a lot of free databases that we can use. that are not but then you have the copyright so big painters, they all have estates and their copyright.

So I think it will be good for the future if that’s also an effort that we have spoken in Lion, that it’s bring everyone together in the table and listen to every party and see how we can work together. Because There are painters, there are paintings that are in wiki art, but that doesn’t mean they’re not copyright.

So that’s a challenge, how we can make a win, we use your models, but you get compensated, or we are licensed, or we are licensed for the entire wiki art, something like that, and [00:25:00] we’ll be happy to reach an agreement, or pay, or mention, something like that. So that’s a business model that I think it requires a conversation of all parties that didn’t exist before.

Currently, for example, search engines like Google, they’re free. You see it. You read what you need to read and then you go on with your life. But when it’s to train a model, in our case, Our models are for free, they’re open source, Lion is a non profit, so there’s really no monetary mean behind it.

So that’s a different story, but I think that the challenge that generative AI has, it’s on the availability of the data

Priya Patel: and

Lucia Cipolina Kuhn: everything surrounding the data. Yeah. Because it’s a new field, so I think new conversations have to be started with the artists and the estates, because you have the artist around the corner that is trying to survive, and then you have the big estate who’s just [00:26:00] preserving the legacy and, but it will be good to have everyone on the table and listening.

Yeah, how we can work together.

Priya Patel: You mentioned copyright. What are the risks here that you’ve identified maybe damaging to artists or to estates identifying fakes, are there other more practical, risks that you see?

Lucia Cipolina Kuhn: Yes In the current conversation is when you develop a model for profit, that’s separate from Lion, but it’s the current conversation with other vendors. So if you develop a model for profit, then artists should be compensated because you’re using them as an input. This or the state or whoever represents the artist, because you’re basically profit off on some states don’t allow you to use their work.

If it is for profit. Now, when you use something non profit, for example, for academic research, the work is not licensed.

Priya Patel: And

Lucia Cipolina Kuhn: it’s not licensed because it’s for academic [00:27:00] research, and it’s strictly non profit. In terms of identifying then real or fakes, there’s always, I think, to take models as a companion, a suggestion.

The good thing is that models give you some certainty threshold, so it can give you the probability of, it will never tell you 100 percent that this is a fake. It will tell you my confidence interval is between 90 and 95%, and then it’s for you to make the call. And it can tell you, because I think this line there, or for example, in the case of Escher, Escher was not right.

Escher painted, it was not on Eric. Escher painted realistic things. In unrealistic setting, but I want to say, it’s not deformed. It had a mathematical structure as opposed to the lead. Everything was deformed. You see the distinction between both. So [00:28:00] AI can tell you that. AI can tell you with this, if you present AI with something that is totally unrealistic, it will tell you this cannot be an issue.

That, that was our experience, identifying deformation from space twisting, which is what or mental puzzles is more like, there are different arts but it’s up art than what is things that are deformed like that. So that can tell you. And I think that it’s a tool, and a tool can always be for good and bad, but I think a conversation and including everyone is the way to go.

Priya Patel: Yeah, for sure. For sure. In terms of identifying, the school of, or was that actually done by the artist’s hand, or by a student of the artist? But I imagine that AI as a tool could help with something like that as well. Maybe

Lucia Cipolina Kuhn: it can help you. It can help you to identify exactly where I can give you a confidence interval. So you can use it as a tool. Among other things, of course, [00:29:00] dating carbon dating of the that’s the first thing I will do is to do a dating of the painting. But then I can tell you, look, here is where I think it goes off.

Yeah, with this certain confident level. So it can give you another tool. Of course, this is something as big as Da Vinci. There will be every expert that you, I’m telling you, things are less obvious that, you may not know us indigenous art in Latin America that we have some of them that you may only have only conservation of five, and you may want to know if it’s this tribe or this other tribe.

They might not be a world expert in this particular indigenous tribe, and then you want, you recover, I don’t know, a vest or a pottery, and that can help you to say there is a pattern of this tribe, and this other tribe use this other pattern.

Priya Patel: And do you think that is really Where you could really see some big advancements [00:30:00] is where you have such limited amount of maybe knowledge or pieces,

Lucia Cipolina Kuhn: yes, or where things are tedious and small, like in museums.

When you have a catalog, or, for example, in the Pompeii Park, where there is millions of little debris hanging around that they’re like, enormous puzzles, and you have bags and bags of debris that you don’t know if it corresponds to a house, to pottery if it’s or what it is.

So that has been around for ages what to do with archaeological sites where you have very little things that are tedious for humans to put together and to realize if it’s this. Part of a wall, it was a pottery or it has nothing to do. So that I think the combination of. Every tool that AI brings from robotics to drones to computer vision identification pattern.

I think that’s the beauty of it, that it brings a bag of tools.

Priya Patel: I love what you just said there [00:31:00] about all the different pieces that you can bring together, it really makes sense. a full circle, I think, especially when Ultimately, we want to share the knowledge.

I think most museums want to share the knowledge Their visitors with their audiences and so that people get excited . If I can shift a tiny bit to more practical stuff in terms of a lot of our listeners might be curators or managers at You Cultural managers, heritage managers, how should they start thinking about AI?

Like, where could they begin? What would your advice be to, to them?

Lucia Cipolina Kuhn: I would say engage in every talk that you can with AI and art. Look on LinkedIn or look on the websites. Look what museums are doing with AI to get ideas. And then if you usually, if you approach people and say tell me more about what you’re doing, or invite an academic to your lab to talk, people who are on the field, that’s how I, if I know [00:32:00] nothing about a field, I would just invite someone to give me a talk Digest this for me, break it down into pieces and tell me what I need to know.

So that would be like a more personalized talk into what you need to know. And then try it out, try something small, try an exhibition of one hour put together something As I said, that’s how I, we started with the tech, people don’t need to know all the models. You don’t need to go into the deep of becoming an expert, maybe something that you just want to become a user.

So by being in the conversation in circles or starting for example, in Bristol, you can go to every Friday. There are talks that the watershed organizes in technology and art. So get involved and then you get ideas and you. Become part of a circle. And then one thing leads to the other maybe collaboration with the university with students, they come and then do an exhibition with [00:33:00] virtual reality, or they do some little project and then little by little you get your foot into our field.

Priya Patel: I think that’s it. That’s a good way to start as well, I think for people to start to dip their toe into it as little by little and focusing maybe on more on the audience side rather than trying to dive straight in.

Lucia Cipolina Kuhn: Yeah, I think it’s interesting. The last part that you mentioned AI, it’s a tool. And we want to use it for be useful for what we need and not be overwhelmed. It’s, it happens so quickly and all of a sudden that we don’t know how to handle it. Is it good? Is it bad?

People like it. People hate it. And, when I was working with artists and with AI, one thing that I realized is this is like Photoshop on steroids, really. Is it’s the next step, but it’s not more than that. I don’t think it will replace artists or creative creativity. It’s a [00:34:00] tool in the same way.

Photoshop didn’t replace photographers or artists because the more craftsmanship that a person has to use these tools, the better the results. Like me as a computer scientist. I wouldn’t be able to produce the level of qualities that I saw real artists, like people who are training art, using these tools.

So it’s not a replacement, it’s an aid. And it’s an aid that can give you as much as you want. It can give you a second opinion, an informed opinion, an opinion with a probability. It can give you an experience to users. It can give you a nice interactive board. It can give you AI art there’s now a festival on cinema made by AI that’s in Spain, and, it’s on its own category as long as we are transparent look, this was made with AI, and I think that I’m excited for What else can be done on [00:35:00] top of what already exists, not in replacement of, but, think about all these possibilities also in archaeological sites that didn’t exist before.

I think these are really exciting times and we have to embrace new things.

Priya Patel: Yeah, I think some of the fear is in terms of fake news and all of that, but it does seem to me that maybe on the creative side, as you said, it’s a tool and it’s, it can be more supportive , than a hindrance provided that we’re transparent, because I think that is very important.

I’m excited for the future of AI and art and archaeology and history. It’s really interesting to hear you talk because it just, you could see how it could all just really be so cool eventually.

Once it can all, you can bring so many different things together.

Lucia Cipolina Kuhn: Yes, and I’m really looking forward for the Herculaneum Papyrus because the poetry that it’s in there, it’s a millennial poetry and it’s so beautiful and we want to know more, right? And we want to preserve that for [00:36:00] future generations and probably 10 years from now, there will be more that is read that we can read today.

And there you see really AI in action, like something that couldn’t be done before. The papyrus have been around, they were discovered in the 1500s. And yeah, they were discovered a long time ago. But no one could open them without breaking them. So there were a couple of attempts and we have.

Broken ones. So just think about that. Like we were able to open or read things that are thousands years old without damaging preserving. So that’s a step forward into a new way of thinking. I think preserving intervention without damaging. Yeah,

Priya Patel: Yeah, that’s incredible because there’s no possibility of looking at it before now, right?

Lucia Cipolina Kuhn: So there were some people that they were opening reading and then throwing them away. [00:37:00] Because archaeology has evolved and before it was different. And you could never imagine that x ray was possible before, so that was the only way. And I think maybe curiosity killed the cat.

The EU has lots of funds for places who are humanities this humanity

Priya Patel: The UNESCO World Heritage Site type things.

Yeah. Okay. Pompeii

Lucia Cipolina Kuhn: is one of those. So it’s on the preservation project. There’s the repair project that actually deals with this little debris. So there’s a lot of academic work done. Surrounding a for this more digging going on in Pompeii, and there’s also preserving what we already have with sensors and drones and, embellishment of the park as well to give a better user experience and also trying to collect as much as we can of all that debris that is floating around and putting it together like this little puzzles.

Yes, [00:38:00] and there is then the independent archeologists using machine learning or AI for non-invasive treatment of archeological sites. And then hopefully museums will implement more and more AI from 3D scanning to digitalizing collections sharing information archival or sometimes, what is good with art is to put together schools of thoughts of different, even different similarly different cultures like Japanese art and trying to get patterns of influence from European art, things like that, that AI can snap very quickly. Oh, look this Japanese art is here that probably had influence from other regions.

So that’s something that AI can bring together like clusters of, And features that from different, you can quickly analyze bodies of work and put together clusters of features of seemingly unrelated things.

, that’s just so interesting and so exciting. So [00:39:00] exciting.

. Yeah, the future, future looks looks great. I think in this aspect, anyway, I don’t want to talk about video or voice clones or anything like that. But but just a small thing, like as somebody who is interested , I love the recreations that they’ve done of from paintings to show you the 3d face that they’ve done.

Priya Patel: I’ve seen that online, I must admit. But they’re pretty cool. It really does get you a little bit more engaged , what did Julius Caesar look like or what, in kind of modern terms, I think it’s very engaging.

Lucia Cipolina Kuhn: Yes. One of the projects that I have is completing the Venus of Milo.

What are those hands, right? And if I can even give you something that says, 90 percent confidence. These are the hands. Wow. So that can do, you can do with AI, you can do a 3D prototype, and then based on the posture and how other similar works, you can say and it’s probably again, because you have so many degrees of freedom, no?

Because it’s the forearm it’s the hands, the forearm. So there are many degrees of freedoms, but this. I can give you 10 with 90 [00:40:00] percent confidence. It’s something scientific, it’s not an opinion. It’s more mostly, more or less based on science, let’s say.

Priya Patel: I get so excited about it because I think it really brings it to life.

Lucia Cipolina Kuhn: Also the fascinating thing with AI, that it combines with traditional art. like scanning is a traditional field

Priya Patel: yeah. It’s fascinating. I love it’s very exciting. I just wanted to say thank you so much for joining us. That was a really fascinating topic.

Lucia Cipolina Kuhn: Thank you so much for this interview. Thank you. And anytime. We are open to and welcome. Conversation with everyone. So you have my details of feel free to drop me a line and we’ll be happy to engage and to listen to everyone and come and speak with everyone and just, be engaged and be in the community.

Priya Patel: Brilliant, brilliant. Yes, just to, just as a side note we will put all the contact details and the websites that Lucia has mentioned during the interview, we’ll put that in the show notes so everybody will be able to get access to them. But thank you [00:41:00] again. That was brilliant. Excellent.

Thank you so much. Bye bye. Thanks a million.

Lucy Costelloe: I’d like to one more time, just, just thank you, and a reminder to our listeners to like, subscribe and share this episode of the Arts and Everything in Between podcast.

Charlie: Recharge, the Ticket Solve event of the year, is back for 2024 and it’s bigger, better and more energising than ever. This autumn we’re bringing together delegates from across the arts and cultural sector for two incredible events. Join us in Dublin on the 26th of September and in London on the 3rd of October for a day of immersive learning with workshops, panels and talks from top arts and culture leaders.

You’ll have the chance to connect with professionals from across the UK and Ireland, share experiences, build lasting relationships, and get inspired by success stories and innovative projects. Register now at recharge. ticketsolve. com.

‍