Rethinking AI, Culture & Creativity | Oonagh Murphy from Arts Council England

Orlando (00:03.81)
Great, it is now recording. So that’s good news. We’ve solved one problem.

Oonagh (00:07.369)
And

Orlando (00:10.446)
Okay, fantastic. All right, well we can get started.

Oonagh (00:16.103)
Yes. Sounds good.

Orlando (00:18.58)
Awesome. Well, Una Murphy, I’m so excited to chat with you because you seem to be sitting at just the perfect intersection for this podcast. You’re a senior lecturer at Goldsmiths at the University of London. You’re the director of Respon the Responsible AI Academy at the Arts Council in England. You’re a previous Turing fellow at the Alan Turing Institute. And so basically, everything we’re interested in here, you’re right at the epicenter of it. So I’m so excited to chat with you. Thanks for coming on the podcast.

Oonagh (00:45.907)
Yeah, great to be here. I’m excited for our conversation.

Orlando (00:49.356)
Me too. I mean, so what I think is cool about what you’re doing is that most AI conversations start with the technology first, like what can it do? How smart it is, will it replace us? Your work seems to start somewhere else entirely. You start with kind of the culture and the institutions and the artists and the public. And I was just wondering when looking at the technology through the cultural lens rather than the technology technological lens, what do you think changes? What unique perspective do you think that gives?

To the conversation.

Oonagh (01:21.465)
Yeah, so I think when we start looking at how people use technology, things get really interesting. technology is quite often built in lab conditions and the real world does not play like a lab. and you know, technology is culture. it interacts with people, with society, with places, whether it’s you know, physical culture, whether it’s human culture.

And it’s those intersections that really allow us to understand the full potential of these technologies.

Orlando (01:52.367)
Mm-hmm. Yeah, it’s funny because you you said that the the way like technology, technology is part of culture, but technology also tries to very much reduce culture down to bits and bytes and measurable things. And I think that culture is always trying to sort of shrug off that data-centric point of view. you know, what are you finding looking at, you know, how this technology particularly, which is the first really to kind of

Mimic culture and try to be, you know, augmentative but also replacing of certain parts of culture. What what are you kind of seeing when you’re seeing how those two things are interacting? That clash of civilizations, as you

Oonagh (02:38.552)
Yeah, it’s a great question. And actually, in some ways it’s not new. I often refer to museums as being the original data set. if we think about museums, we can we can learn a lot about how our technologies are working and succeeding and failing. Like if we think about museums, they created taxonomies. Museums established knowledge hierarchies, they’ve produced descriptive languages for really complex, like social, cultural, natural phenomena.

Orlando (02:47.867)
Mm. It’s interesting.

Orlando (02:54.062)
Mm-hmm.

Orlando (02:59.992)
Mm-hmm.

Oonagh (03:08.504)
and that history means museums are pretty well placed to understand what AI systems do, because AI systems depend on classification, pattern recognition. And in many ways, actually, it’s museums that help to produce the habits ha on which AI depends. So museums know probably better than most institutions the perils of systems and processes that rely on categorization.

Orlando (03:12.227)
Mm-hmm.

Orlando (03:32.686)
Okay.

Orlando (03:39.169)
Yeah.

Oonagh (03:39.345)
and I think that that starting point is really useful.

Orlando (03:46.669)
Yeah, I mean and and as well, I mean, I think that that they’re also insta cultural institutions that people trust. And so that intersection of where the technology is being used in museums and and blessed by museums is probably also a really interesting place to to talk about.

Oonagh (04:04.739)
Yeah, and actually for me, the role of museums as trusted institutions is a useful frame to think about how we define trust because museums are trusted by certain people. Museums are trusted by those that have treated them well historically. But actually, if we look at the history of museums, if we look at the histories of collections, they’re based on colonial acquisition, racialized taxonomy, exclusionary knowledge systems.

Orlando (04:19.192)
Mm-hmm.

Orlando (04:23.886)
Mm. Mm-hmm.

Oonagh (04:36.728)
And so museums again really useful when it comes to thinking about AI. Because for some people, museums are great places. They’re trusted. They’re sources of information and knowledge. For other people, they’re places of discrimination and exclusion. And so we can see here how these social cultural decisions around how information, data, people, and culture are represented change a person’s experience. An AI system for one person may be positive.

Orlando (04:41.474)
Mm-hmm.

Orlando (04:50.398)
Cases of discrimination and exclusion. Yeah.

Orlando (05:02.68)
Mm-hmm. Experience. Mm-hmm.

Oonagh (05:06.574)
But may be harmful for another. They’re both experienced in the same system, but their cultural perspective means that their experience of that system is radically different.

Orlando (05:16.632)
Gosh, yeah, that’s fascinating. I mean, so I guess the question is, you know, like what, you know, and this is a sort of a question I think people tackle forever, is sort of what is culture and what is the role of the the individual versus the institution in defining what that is? You know, I think that this technology has very much been put in the hands of a lot of individuals and a lot of individual artists and a lot of different people from data scientists to visual artists to

Obviously, AI slot merchants on the internet have all been picking them up quite quickly. It seems like there’s an argument, though, for cultural institutions being some of the people to really seriously pick up these tools because they offer a data rigor, a use rigor, and a historical rigor around the use of them. And it, you know, the things they end up doing don’t just end up being sort of slop, as we call it.

Oonagh (06:15.519)
Yeah, yeah, exactly. And I think for me, building on that kind of that idea of of AI systems being essentially taxonomies of knowledge, of data that has been collected and curated and tagged and organized through the world view of primarily people in the West,

Orlando (06:28.462)
Mm.

Mm-hmm.

Oonagh (06:38.077)
is really useful if we link it back to this idea of tangibility and the objects of systems. So we’re quite often told that, you know, AI is this magical system. it’s a technology that we can’t possibly understand.

Orlando (06:45.666)
Mm-hmm.

Orlando (06:53.858)
No.

Oonagh (06:55.274)
But if we take it back and we start to think about well, what’s the physicality, what’s t what’s the infrastructure of these technologies, then I think that’s a really good way to physically ground our conversations in what these technologies mean. And so for me there’s there’s three kind of strands that I always come back to when I’m talking about what AI technologies are. the first is AI does not come from the future. It simply reimagines the past.

Orlando (07:00.622)
Orlando (07:08.382)
Explain these technologies mean.

Orlando (07:21.166)
It simply reimagines the past.

Oonagh (07:24.638)
AI systems rely on historical data and that data is never neutral. So it’s important to think about this data as a social cultural object in its own right.

Orlando (07:29.187)
Mm-hmm.

Orlando (07:35.448)
Mm-hmm.

Oonagh (07:37.69)
They are not systems of the future. They’re systems of the past. And the past is problematic. And we know that. The next is AI isn’t magic. It is made by human hands.

Orlando (07:38.518)
They are not systems of the future. systems of the past.

Orlando (07:45.356)
But the next is AI is a bad thing. Mm-hmm.

Orlando (07:52.896)
Mm. Yeah.

Oonagh (07:54.538)
So, you know, whether that is, you know, data workers in the global south seeing harmful content so that we in the West don’t have to see it, whether that’s, you know, policymakers in Washington DC or London, whether that’s people making NVIDIA chips in a factory.

Orlando (07:59.247)
Mm-hmm.

Orlando (08:04.259)
Mm-hmm.

Orlando (08:07.715)
whether that’s people making individual chips in a factory.

Oonagh (08:14.634)
There is a physical tangibility to AI systems and processes. And when we grind it in what that looks like, we remove the kind of the cloak, the aura of magic, and we can start to understand what this means and where decisions are made. And the final one, it doesn’t live in the cloud, it’s drawn from the earth. So if we think about these systems, you know, we’re totally live in the cloud. They don’t, you know, electricity pylons, Wi-Fi, physical hardware.

Orlando (08:25.998)
Mm-hmm.

Orlando (08:29.902)
Mm-hmm.

Orlando (08:34.882)
Mm-hmm.

Orlando (08:39.223)
Mm-hmm.

Mm-hmm.

Oonagh (08:44.593)
The infrastructure of these technologies, data centers, they’re everywhere. We just have to look for them.

Orlando (08:49.726)
Everywhere we just have to look for them.

Oonagh (08:53.087)
And so taking that physical infrastructure approach to starting the conversation, I think is a really useful way, you know, to kind of to take the yoga language to ground ourselves in the question before we then start to unpack what the technologies mean in the system or process.

Orlando (09:02.67)
Yeah. Yeah.

Orlando (09:13.228)
Yeah, gosh, that’s it’s that’s so great. Cause so much of what I’m hearing is the language around responsibility, you know, and grounding. And I think that there’s this need to ground and make sure that we’re handling these things responsibly while at the same time not slowing down so much that we miss the technological moment. Because culture naturally has to keep up. Culture is sort of this ever-present thing.

And technology is finding a way to be so ever-present that we sort of, when we stop, when we think, when we slow down to really understand things, we end up falling behind the technology. and so I guess the the question is like, how do you propose doing that, whether it’s cultural institutions or you know, or yeah, how do cultural institutions stay relevant and fast while still being responsible?

Oonagh (10:10.436)
Yeah, it it’s a challenge. you know, without question. But I think the history of technology teaches us that

This challenge is not unsurmountable. One of the key things is what are the drivers for the decisions that we’re making. And if we look at kind of the aeronautical industry, they realise pretty quickly that planes falling out of the sky was bad for business.

Orlando (10:36.814)
Mm-hmm.

Orlando (10:40.162)
Right. Yeah. So they start at the sound right to the way. Yeah. Yeah.

Oonagh (10:41.562)
So they started to self regulate.

You know, if we look at seat if we seat look at seatbelts in cars, they help to sell more cars.

Orlando (10:49.634)
Yeah. Yeah. Yeah.

Oonagh (10:52.334)
And so I think when it comes to museums or or any cultural organisation, creative individual, creative freelancer, actually it’s good for business. You know, in an era of mass information, in an era of AI slop, in an era of misinformation, disinformation, hate speech, cultural organisations, creative practitioners, their point of difference is authenticity and trust.

Orlando (11:03.468)
You know, in

Orlando (11:18.284)
Mm-hmm. Yeah.

Oonagh (11:20.493)
And so actually they don’t need to use every single AI tool possible, but instead they need to engage with the AI tools in a very considered way and use them selectively. And so I think if we think about kind of production cycles, we think about you know tech stack, and you know a big tech stack does not make for a good product or a good outcome for a project. And so sometimes it’s about using less, better.

Orlando (11:25.934)
Mm-hmm.

Orlando (11:33.172)
Mm-hmm.

Orlando (11:45.111)
Yeah.

Mm-hmm. Yeah. That’s interesting. Do you think the creative sector right now is being true treated too much as a victim of AI rather than as the field that can actually actively shape it?

Oonagh (12:01.91)
Yeah, do you know it’s a challenge, it’s a real challenge. I think in my work, I sit in the space between two quite polarizing p opinions on on on the cultural sector and the creative industry. So on one side there is the very valid concerns around intellectual property, concerns about people’s livelihoods, and these are completely valid concerns. On the other hand, there are creative practitioners.

Orlando (12:23.342)
concerns about people’s likelihood.

Oonagh (12:31.179)
Who have used technologies in their creative practice long before AI. They are slow, considered, curious people who make work that allows us to understand the impact of these technologies before the rest of society or policymakers has seen what these technologies might look like and might do. And so I think the challenge is

Orlando (12:37.294)
Yeah.

Orlando (12:45.26)
Mm-hmm.

Orlando (12:56.686)
Mm-hmm.

Oonagh (13:02.42)
These two groups have existed in isolation. And actually I think that the nuance space in the middle is beneficial to both parties. And I think that the intellectual property debate could be further enhanced through creative lenses and creative practices. But equally, quite often, those that work with creative creative technologies struggle to get investment.

Orlando (13:07.63)
Mm-hmm.

Orlando (13:14.04)
Mm-hmm.

Orlando (13:22.915)
Yeah.

Oonagh (13:31.435)
They struggle to develop business models. And so those parts of the creative industries that rely on intellectual property, for example, have established business models. So I think I would like to see the nuanced space in the middle where those that are negatively impacted and those that are curious, critical, and have always engaged with emerging technologies come together to create a better ecosystem for both.

Orlando (13:34.398)
Mm-hmm.

Orlando (13:40.106)
So I think I would like to see

You on space in the middle were those that are negatively impacted.

Orlando (13:59.215)
Mm-hmm. Yeah.

Orlando (14:04.588)
Yeah, it’s it’s it’s funny. I mean, there it seems to be there’s a real a balance to be struck. and I think that like when people hear responsible use, they th they feel, you know, slow and compliance and rich registers and and people saying no. But I think that what what you’re saying is that that a lot of that responsibility can be a feature and allow the creative industries to kind of lead the conversation.

Oonagh (14:31.795)
Yeah, exactly. So I think if I pick up on on one project, I’m working with the Munck Museum in Oslo in Norway. So famous for the screen painting, Edward Munk. and I’ve been working with them for the last few years as their AI ethics advisor. They essentially digitized about seven thousand works on paper. They were digitized, but nobody accessed them. And the works on paper really vary from a line on the page to

Orlando (14:40.942)
Mm-hmm. Mm-hmm.

Orlando (14:46.318)
Mm-hmm.

Orlando (14:51.047)
Mm. They were digitized but nobody accessed them.

Orlando (15:00.414)
Mm-hmm.

Oonagh (15:00.755)
Maybe a more detailed drawing?

But very hard to discover. But these works of art are also really sensitive to light. So they can’t actually be put on display. And so they wanted to think about new using technology to encourage visitors to engage with this collection.

Orlando (15:09.452)
Yeah. Right.

Orlando (15:19.266)
engage with this collection.

Oonagh (15:22.055)
And they started to look at different ways of interacting with the collection, different ways of searching. And they thought, well, it’s a works and paper collection. It’s a series of drawings. What happens if we search by drawing? How what does that look like?

Orlando (15:27.318)
This is

Oonagh (15:38.63)
But the Munk Museum also has a collection of contemporary art and commissions contemporary artists and is interested in the use of technology in in creative practice. So they began to think, well, actually, can we use this as part of a wider public program to unpick how AI technologies could be used in creative practice, but also to unbox some of the conversations around consent, around data sets, around intellectual property.

Orlando (16:00.696)
Mm-hmm.

Orlando (16:07.203)
Yeah.

Oonagh (16:07.762)
And so the museum over two years now, maybe three years, developed a series of R and D sprints. They have an in house lab and they were able to develop an interactive drawing experience on an iPad where you draw, and when you draw, computer vision technology is used to to map against the work and paper collection.

Orlando (16:15.374)
series of R and D sprints. They have an in house lab?

Orlando (16:26.849)
Mm.

Oonagh (16:39.577)
The immersive experience, the interactive experience rewards you, for example, for discovering a drawing that no one else has ever discovered. So you get this sense of in you get this sense of intimacy of experience that is slow. So when they were developing the R and D kind of lab space, they wanted a space that was slow so they had comfortable chairs. They didn’t have large, shiny immersive experiences. Instead they wanted to recreate the intimacy of drawing, of mark making through digital means.

Orlando (16:46.636)
Right, okay, cool.

Orlando (16:50.392)
Mm-hmm.

Orlando (16:59.8)
Mm-hmm.

Orlando (17:03.416)
Mm-hmm.

Orlando (17:07.768)
That’s nice for digital name. Mm-hmm.

Oonagh (17:11.348)
But what they realized was they needed to collect the drawings in order to improve the model, to improve discoverability. So rather than just doing that, they produced a whole interpretive plan so that when visitors entered the space, they were able to learn about data sets, they were learn able to learn about AI technologies, they were able to learn about different forms of consent and intellectual property.

Orlando (17:15.918)
Mm-hmm.

Yeah. Yeah.

Orlando (17:28.43)
Mm-hmm.

Orlando (17:37.624)
Mm-hmm.

Oonagh (17:38.887)
And so it was a way in for members of the public to start to have those critical conversations about what AI technologies are. And so the museum is now moving to a major exhibition which opens in a few weeks. This has opened up a whole conversation about what does it mean to be a data custodian for a museum. And so the museum

Orlando (17:46.957)
Yeah.

Orlando (17:58.969)
Mm-hmm.

Yeah.

Orlando (18:07.406)
Right.

Oonagh (18:08.753)
Has traditionally collected data around collections. But one of the challenges facing a lot of artists now is access to data sets and access to models. So whilst the big tech companies are quite happy to scrape intellectual property, they’re not willing to share the access to those models. And so the museum is now asking questions about whether our role as a data custodian to make

Orlando (18:14.03)
Mm-hmm.

Orlando (18:20.974)
Mm-hmm.

Intellectual property.

Oonagh (18:36.962)
data available to creative practitioners. And what does that look like in terms of how we can support emerging creative practice and we can support creative technologists to have access to data systems and institutions. So really reimagining the role of you know the museum as the original data set to the museum as a data custodian for an AI age. And all of this done very much through an R D lens and through a you know a c a public

Orlando (18:40.93)
Yeah.

Orlando (18:48.77)
Uh-huh.

Orlando (18:52.91)
Mm-hmm.

Orlando (19:00.844)
Well.

Oonagh (19:06.309)
discourse lens. None of it is about serving up definitive answers, but instead it’s about developing answers with with the community, with the visitors and really embedding that concept of trust and legitimacy of the institution in this emerging data space.

Orlando (19:10.763)
Mm-hmm.

Orlando (19:18.86)
Yeah.

Yeah.

That’s very interesting. I mean, do you think it leads to a world where each museum starts training their own models on their own data? And, you know, it’s interesting because you th you think of a museum from an external eye as just a repository of lots of historical data. But at the same time, each museum usually has a mission statement. They have an editorial lens on the types of things they are trying to

to lift up and and that you know that editorial lens changes with each new curator coming in. but you know and you know the Nobel Museum is very much about the things that that Alfred Nobel is and the awards are about. Again another Norway reference. But it’s interesting because then you kind of have do you think I guess I’ll go back to the original question. Do you think it leads to a world where museums start

building their own models on their own unique artistic data sets.

Oonagh (20:27.76)
Yeah, it you know it’s a real challenge. And I think one of one of the primary challenges facing museums at the minute is there was a push towards open access. So a lot of museum data sets were put online, were digitised put online under Creative Commons license. The challenge has been that whilst the data is free to access,

When the large companies are scraping those data sets to train their model, it is having a direct financial cost to museums.

Orlando (20:59.148)
Yeah.

Oonagh (21:05.433)
Because it is driving up the cost of running those data sets. So not only are the the large tech companies scraping that data and you know, arguably stealing it, borrowing it, using it, use whatever descriptor you like, they are also financially impacting these cultural institutions because they’re now in a position where their choice is do we pay, essentially, do we pay to permit.

Orlando (21:10.658)
Yeah.

Orlando (21:22.008)
Yeah.

Oonagh (21:34.339)
these companies to script our data.

Orlando (21:36.472)
Mm-hmm.

Oonagh (21:38.446)
And do we have the budget to do that because museums run a very tight operating model? Or do we withdraw open access? Do we start putting stuff behind paywalls again? And it’s not it’s not they don’t want the paywall for visitors, they don’t want to pay well for artists. But the paywall is because there is a financial implication for open access that is being caused by scraping.

Orlando (21:40.83)
It’s rolling very tight operate model. Yeah.

Orlando (21:48.29)
Yeah.

Orlando (21:56.707)
Yeah.

Orlando (22:02.382)
Mm-hmm.

Oonagh (22:05.23)
So I think there’s lots of complications around you know, should they change train their own models? you know, should they should this data be available to large language models? There is also an interesting debate that has ha that you know that has existed around well, is it our public duty to make our collections available because we are a trusted data source, and if the models don’t train on us?

Orlando (22:05.528)
Yeah.

Orlando (22:26.007)
Mm-hmm.

Orlando (22:31.362)
Mm-hmm.

Yeah.

Oonagh (22:34.397)
they’re gonna train on Reddit. And so is it part of our public duty to make our collections accessible so that the AI models are less awful in terms of their worldview, in terms of bias and gender and race and all of these things. So you know it’s a really complicated picture. I would say the sector itself doesn’t have like one coherent narrative. I think different institutions have taken different approaches. I I also think that there is a challenge that you know different countries have different legislative approaches.

Orlando (22:36.781)
Yeah.

Right.

Orlando (22:43.811)
Yeah.

Orlando (22:54.391)
Mm-hmm.

Mm-hmm.

Oonagh (23:03.994)
and the landscape is changing really quickly.

Orlando (23:04.099)
Mm.

Yeah. And is that I mean, is that is your role at the Arts Council England one where you’re trying to get a s consensus among many different cultural institutions? Or is the the idea no, we want you to have a robust ethical and security framework, but you know, we don’t want to be, you know, having you ascribe to one way of doing all this. You should have a certain amount of freedom to to

express yourself on your own data and and make your own decisions in that respect.

Oonagh (23:40.905)
Yeah, exactly that. What’s right for one organization will not be right for another. And I think, you know, like, you know, from interviewing creative practitioners, from interviewing those that run organizations, what is a red line for one organization is completely fine for another. You know, so for example, you know, I’ve talked to you know, educators that run writing workshops and they’ve used generative AI to to create an image to use as a visual prompt.

Orlando (23:58.031)
Yeah.

Orlando (24:06.829)
Yeah.

Orlando (24:10.284)
Mm-hmm. Yeah.

Oonagh (24:12.106)
Totally fine, totally fine. I’ve chatted to a visual arts educator who says they would never use it to do that, but they’d happily use it to describe an image. Where the the liter literature focused organisation would never use it to generate text. So, you know, two similar organizations, similar end users, similar legal structure, similar business model, completely different red lines.

Orlando (24:14.382)
Mm-hmm.

Orlando (24:19.47)
Mm-hmm.

Orlando (24:24.768)
Mm-hmm. Yeah.

Orlando (24:31.406)
Mm-hmm.

Orlando (24:38.435)
Mm-hmm.

Oonagh (24:42.634)
And I think that’s quite common. And one of the decisions that we made at Arts Council was we published quite a detailed report on our governance structure and this deliberative mechanism of policy development that we took because we felt it was a useful framework for purpose led and public sector organizations to develop AI policy. But we actually made a decision not to publish our AI policy.

Orlando (24:56.056)
Mm-hmm.

Orlando (25:02.434)
Mm-hmm.

Orlando (25:08.35)
Mm. that’s interesting.

Oonagh (25:09.511)
We didn’t want people we didn’t want people to copy it.

Orlando (25:13.014)
Yeah, right. Right.

Oonagh (25:14.569)
Because we spent a year developing it and it’s custom to that organization. It was developed through deliberative governance internally. So that what that means is we spoke to people across the organization through a staff survey, through a range of different governance forums, a staff working group, an oversight group, and the executive board. And we collectively developed a policy and we collectively developed the organization’s red lines.

Orlando (25:19.469)
Yeah.

Orlando (25:40.959)
Mm-hmm. Yeah.

Oonagh (25:43.39)
The decisions we made might be right for some other organizations, but copy and pasting a Word document does not make good governance. It is a process. And so we published this quite detailed report that talks about how to set up governance frameworks, about how we collected data, that talks about some of the challenges we faced, some of the trade-offs. But we decided not to publish the actual policy document because we didn’t want people to copy and paste it and go job done. And the whole report says.

Orlando (25:46.861)
Yeah.

Orlando (25:50.85)
Yeah.

Orlando (25:57.7)
Mm-hmm.

Orlando (26:10.36)
Yeah.

Oonagh (26:11.517)
There isn’t, you know, like you have to make these decisions. One thing we did do was we published a a kind of a AI policy toolkit and it’s a real step-by-step guide, but it’s a questioning framework. Essentially, if you fill it in, you will get a policy, but we don’t give you any answers. It’s a you know, and it starts with, have you spoke to your staff? and that’s really important because it’s the most junior to the most senior, it’s every single person in the organization. They’re probably already using AI.

Orlando (26:14.733)
Yeah.

Orlando (26:21.474)
Right.

Orlando (26:32.301)
Yeah.

Yeah. Yeah.

Oonagh (26:41.105)
If you write an AI policy and you haven’t consulted, then your o your policy is not going to govern what practices are existing already in your organization.

Orlando (26:44.494)
Mm-hmm.

Yeah. Yeah. That’s so funny. We see that all the time, is that everybody goes, We want to do this properly, we want to do this well. Gimme a thing. And we go, well, actually, doing it properly and doing it well means a little bit of friction. And as a business, obviously, we’re always supposed to be eliminating friction from our onboarding process. But there’s a certain amount of friction there. We need to understand your business. We need to understand what your goals are before figuring out how to just apply AI in your business. It’s not.

It’s not as simple as that, and it shouldn’t be as simple as that. And and you know, we come from the perspective of yes, we’re using AI, we’re helping creative companies integrate and use AI, but we come from the perspective of these tools have committed the largest copyright theft of all time. You know, that this is not a you know, we you’ve gotta be bringing the ethics of the creative company that you are to the table.

This is a time for more ethics being brought to the table, not less, and not just signing it away. We’ve seen what happens when people do that.

Oonagh (27:51.218)
Yeah, and you know, it’s very easy to learn or to lose trust. It’s very hard to earn trust back. and I think probably useful also in this context to just mention how I ended up at Arts Council because I think that’s really central to the approach that we took. so I received funding from a UK government department called the Arts and Humanities Research Council. and so they have a programme called Braid, which stands for bridging responsible AI divides.

Orlando (27:57.795)
Yeah.

Orlando (28:04.602)
I think that’s really sad.

Orlando (28:20.238)
Mm-hmm. Yeah.

Oonagh (28:20.711)
And it’s a really interesting, really novel approach. And essentially what their quest it question is, what happens if you put arts and humanities systems thinking at the start of a novel technology adoption cycle?

Orlando (28:31.212)
Mm-hmm.

Mm-hmm. Yeah.

Oonagh (28:35.546)
So, you know, you know, I know, I think we all know that quite often tech policy, governance, adoption type jobs happen because something went wrong. Like that’s when the first job rolls created. Like that’s just the way it is. And so Breed did this weird thing where they created 17 fellowships and they put us into organizations all around the country, really different organizations, some commercial, you know, some public sector government.

Orlando (28:50.435)
Yeah.

Orlando (28:58.36)
Mm-hmm.

Orlando (29:04.749)
Yeah.

Oonagh (29:05.454)
some cultural organisations. And those organisations got fellows for free. So I worked at Arts Council England for eighteen months at no cost to the organisation.

Orlando (29:12.195)
Mm-hmm.

Oonagh (29:17.894)
And that’s really important because that meant I could be slow, I could ask questions, I wasn’t business as usual, I was a nice to have. And so that in itself is a really interesting positioning and positionality for the work that I did in the organization. But like, you know, when I first went into the organization, I had a meeting with the the union, you know, and the union were kind of like, What are you doing here? like

Orlando (29:22.537)
Mm-hmm.

Orlando (29:27.128)
Yeah.

Orlando (29:32.803)
Mm.

Oonagh (29:46.713)
Is this like you know what is there a restructure coming? But because of the way I was funded, I was able to say, nope, I’m an academic. If I can teach you how to resist AI, that’s fine. If I can teach you how to use AI, also fine. Like I’ve no skin in the game here. I’m not here to sell AI. I’m also not here to stop AI. I’m here to help govern what’s already happened, but also to take a proactive response or approach.

Orlando (29:46.893)
Mm-hmm.

Yeah.

Orlando (29:56.739)
Mm-hmm.

Yeah.

Orlando (30:04.152)
Uh-huh.

Yeah.

Orlando (30:10.584)
Yeah.

Oonagh (30:15.266)
rather than a reactive one. So like what happens if we sit and think about this? And so like one of the biggest questions that we sat with was should applicants be allowed to use AI to write grant applications? And should Arts Council England use AI DSS grant applications?

Orlando (30:17.582)
Yeah.

Orlando (30:34.872)
Yeah.

Oonagh (30:36.58)
And that is a question that would have been solved if I wasn’t there, that’s a question that would have been solved because something bad happened.

Orlando (30:45.57)
Yeah, yeah.

Oonagh (30:46.906)
Because I was there, were able to have that as a proactive conversation. We were able to consult across lots of different groups in the organization from the most junior to the most senior. We were able to look at it from a workflow process. So, you know, literally sit down with people that are making decisions around grants and go, Right, show me an application you’ve received that you think is written by AI. What’s the problem here? Like, is it a quality problem? Is it a technical problem? Is it just a value situation? Like, if that wasn’t written by AI,

Orlando (31:09.08)
That’s a problem, yeah.

Orlando (31:15.214)
Mm-hmm.

Oonagh (31:17.732)
Like, is it a good application? And so really sitting and thinking about those rather than taking a reactive approach meant we were able to publish a statement on the website that simply says, you can use AI to write your grant application if you want, because ultimately a good application is a good application. And if somebody uses AI badly, they’re still going to generate a bad application. If somebody work if somebody works with AI and they edit and they prompt and they build in their own knowledge and their own systems.

Orlando (31:35.083)
Uh-huh.

Yeah. Right, right.

Oonagh (31:46.586)
Then yeah, AI will probably make their application better. But you know what? Welfare organizations, welfare artists currently pay grant writers to write their application. So ultimately we came back to this isn’t an AI question, is it? The question is do you have to write your own grant application? And the answer currently is no.

Orlando (31:46.668)
Yeah.

Orlando (31:55.853)
Right.

Orlando (32:01.048)
Mm.

Orlando (32:07.17)
Right. And but doesn’t it mean that that the moment you say yes to that, you vastly increase potentially the amount of applications and the detail oriented nature of the applications. So then the second decision becomes a necessity. You have to use AI to read the applications, because otherwise you’ve created a impossible task for the the grant readers.

Oonagh (32:24.111)
Yeah. Yeah.

Oonagh (32:29.558)
Yeah. Well, I think one of the good thing one of the bad things many people would argue, but I think one of the good things in this case is the way grant software is designed, there are multiple boxes. So it’s quite cumbersome, shall we say, to apply for grants generally. And I think that ac acts as a as a barrier. But yeah, you know, ultimately you can get but bad applications that don’t meet the criteria that are waffly, they’re easy to they’re you know, they’re easy to reject.

Orlando (32:44.526)
Mm-hmm. Right. Right.

Orlando (32:53.814)
Mm-hmm.

Mm.

Oonagh (32:59.457)
And that’s always been the case. That’s not a new thing. And yeah, and the organization made a really strong decision and said, No, we will not use AI to assess grant applications. We are art form experts. so the people the people that manage applications at Arts Council are called relationship managers and they have discipline-specific knowledge. And they said, Nope, that’s a bit of the job we love, that’s a bit of the job we want, we are experts in that. And so we arrived at a decision and the policy’s on the website, and it says,

Orlando (33:00.043)
Yeah. Mm-hmm.

Orlando (33:07.363)
Mm-hmm.

Orlando (33:11.437)
Wow, great.

Orlando (33:19.426)
Mm-hmm.

Orlando (33:24.728)
Great.

Orlando (33:28.526)
Mm-hmm.

Oonagh (33:29.43)
This is our policy today. It might change in the future when technology changes. But you know what? If it does, we’ll talk to you about it. We’ll let you know. But this is really important because before that statement went in the website, when people didn’t get funded, it had started to become the case where people were just going, nobody nobody even read it. It was just assessed by AI. And like people could say that’s not the case, but if you don’t have a policy, you can’t evidence it.

Orlando (33:31.372)
Yeah.

Orlando (33:36.566)
Yeah.

Orlando (33:41.366)
didn’t get funded. Mm. It had to be They blamed AI.

Yeah. Right.

Oonagh (33:53.759)
So now like relationship managers feel confident that, you know, if somebody complains about how an application’s been assessed, they can signpost them to know this is our policy. And so again, like you kind of said earlier, like is responsible AI just slowing us down? Actually, responsible AI done well is proactive and it’s stopping reactive chaos that happens when we don’t have systems processes in place. So sometimes it might slow the process down.

Orlando (33:54.263)
Yeah.

Orlando (34:02.648)
Yeah.

Orlando (34:11.662)
Mm-hmm.

Mm-hmm.

Yeah.

Oonagh (34:23.329)
I would say arguably in my experience, it makes the process a lot smoother over the duration of a workflow or a program or project because you’ve got ahead of the crisis, you’ve got a plan, you’ve got a you know, and you know, so I think it’s good project management.

Orlando (34:30.52)
Right.

Mm-hmm.

Orlando (34:38.07)
Mm-hmm. Yeah. It’s funny because you you mentioned the Braid program, which is a six-year 15.9 million pound program running from 2022 to 2028. It seems like the UK, Europe was very quick, you know, to to have its you know to put in a framework from Brussels around use of AI within the EU.

The UK, the arts council, there seems to be a lot more government funding and research and things happening where real money is being put towards figuring out how to interact with these technologies. And we talked earlier about how, you know, there’s a speed gap. The technology can progress so far. And if you’re being responsible, you can’t, you know, go as fast. But it feels like America in general is like, well, we can’t go as fast. So they throw their hands up in the air and don’t really do anything.

But at least Europe, especially, you know, and the UK especially seems to be at least putting time, attention, and crucially money behind making sure that responsibility towards their cultural institutions, their artists, you know, and and and as well their businesses, is respected. Are are you do you think do you feel you’re part of a conversation and doing something that actually isn’t being replicated around the world necessarily?

Oonagh (36:01.898)
Yeah, it’s interesting. I think we, you know, I think in the UK we do have a really good, you know, assurance, governance, responsible ecosystem around these technologies. So I think, you know, if we look at, you know, institutions like the Alan Turing Institute, if we look at the Ada Lovelius Institute, if we look at the work of Braid, another project I’d I’d s flag is public voices of AI. the

Orlando (36:30.158)
Right. Flag not slag. Just make sure. Yeah. Yeah, yeah, yeah, yeah.

Oonagh (36:32.913)
Yeah, yeah, yeah. Flag not slag. Yes. We’re friends. We’re fr we’re friends. We’re friends. They do good work. yeah, yeah, yeah. but yeah, get cancelled. but yeah, we you know, we are really lucky. We’ve got this really good ecosystem where people are care, that people are interested in like, you know, what the social impact of these technologies are. but I would also say it just makes business sense. Like if you have a product or service.

Orlando (36:54.68)
Mm-hmm.

Orlando (37:00.832)
Mm-hmm.

Oonagh (37:03.155)
People only gonna buy it if they trust it.

Orlando (37:05.352)
Mm-hmm. Yeah.

Oonagh (37:06.847)
And so that goes from government down. Like if you want to, you know, use AI to revolutionise the national health service, well, people are only going to give their data to some fancy new AI app if they trust it.

Orlando (37:15.043)
Mm-hmm.

Orlando (37:20.696)
Yeah.

Oonagh (37:22.107)
and so I think that’s, you know, like it’s the same if we think about education. You know, if you think education can create, you know, AI can create customized learning for young people. Well, parents are only going to use it if they feel like their children are safe. And if we look at the context of social media, you know, we’ve seen that social media has been designed to be harmful for young people. and so I think that there is a greater awareness of the impact of technology.

Orlando (37:31.811)
Mm-hmm.

Orlando (37:36.214)
Mm-hmm.

Orlando (37:40.13)
Mm.

Right.

Oonagh (37:50.481)
on individuals and societies because we’ve learned a lot from social media. And I think the companies themselves, and I think to some degree government are starting to realize that it makes good business sense to have responsible practices that are driven by purpose if they want to generate a genuine return on investment. And so a thing that I keep saying, nobody’s taken me up on it yet, but I stand by it.

Orlando (37:50.51)
on a bit of society because we’ve learned a from social media.

Orlando (38:03.757)
Mm-hmm.

Orlando (38:07.373)
Yeah.

Orlando (38:15.746)
Mm-hmm.

Oonagh (38:19.716)
If every single government project, these kind of like I don’t know, I’ve lost track, like every day there’s like five billion behind for a new AI thing that’s gonna like I don’t know, save the country.

Orlando (38:20.088)
Mm.

Orlando (38:25.463)
Yeah.

Orlando (38:29.12)
Yeah. Yeah.

Oonagh (38:31.236)
If the government put in a one percent tax on every single AI investment and that one percent had to go to public libraries, I genuinely believe it would drive the return on investment up because libraries are commu community infrastructure. They are places where you can physically access computing, Wi Fi, electricity, but also skills and support. And so if you want AI to help with welfare,

Orlando (38:41.762)
Right.

Orlando (38:47.299)
Right.

Orlando (38:55.054)
Mm-hmm.

Mm-hmm.

Oonagh (39:00.978)
They help with healthcare, to help with education. that’s all great, but who’s teaching people how to use these systems?

Orlando (39:05.936)
huh.

Right.

Oonagh (39:09.755)
And so it’s great having these announcements, it’s great having the technology, but you know what? It’s people that make them successful. It’s people choosing to use them. It’s people having the skills, the confidence, the access that will enable these technologies to realise their full potential. And if you don’t factor that in when you’re designing these systems and processes, you have a white elephant that will fail.

Orlando (39:17.837)
Mm-hmm.

Orlando (39:22.657)
Yeah.

Orlando (39:27.565)
Right.

Orlando (39:34.24)
Mm-hmm. Yeah. God, that’s so that’s such a great idea. I love that. what I mean, so what does an arts organization need to do before it starts using AI?

Oonagh (39:39.985)
Yeah.

Oonagh (39:46.448)
Yeah. Sit down and re like think about why it exists, who it’s there for, who are its publics, who are its artists, why does it exist? And that’s different, you know, you know, Arts Council England, you know, within their remit, you know, they have they have libraries, they have visual arts organisations, they have dance, they have music, you know, they’re a really big portfolio. And you know, a national organisation like T Modern has a very different remit than a small regional organisation that maybe has two members of staff.

Orlando (39:52.844)
Mm-hmm.

Orlando (40:16.662)
Mm-hmm. Yeah.

Oonagh (40:17.424)
And actually being led by, you know, not how can we use AI, but why would we use AI.

Orlando (40:23.235)
Mm-hmm.

Yeah.

Oonagh (40:27.238)
as your starting point. Can it solve a problem for us? Great, try it. but not how can we shoehorn AI in because it’s what everyone’s using. It’s it’s not a good use of your time or your money. So I think that’s the first thing. Purpose. Always being driven by purpose. yeah.

Orlando (40:31.511)
Yeah.

Orlando (40:42.574)
Well, I think that was what was so great about your monk thing that you said is that like the lazy version of AI in a museum is a chat bot that tells you that the painting is very important. You know, like, okay, obviously it’s here, you know. That your thing feels more interesting because it has a purpose, because it’s been thought through, because it starts with the visitor doing something physical, making a work, and then then deepening the encounter using AI. And that’s obviously we’re talking about.

Oonagh (40:54.758)
Yeah, yeah, yeah.

Orlando (41:09.452)
something that’s very consumer facing. And obviously right now we’re talking about, you know, governance and internal policies. But I feel like it ripples all the way through, just making sure that there’s a reason, there’s a purpose, and you’re respecting the purpose of why you’re using it.

Oonagh (41:26.849)
Yeah, exactly. And I think that, you know, it comes back to, you know, artists, you know, artists have always made considered choices about the materials they use, whether it’s, you know, pencil, paintbrush, chalk through to photography and now AI.

Oonagh (41:47.411)
those slow, critical, curious questioning is part of the process of being a creative practitioner, but also of being a cultural organization. And I think ultimately, cultural organizations and wider creative industry, you know, they are brands, they are known for something. And if your brand is about criticality and trust, then that needs to be reflected in your approach to technology adoption. you know

Orlando (41:51.603)
Mm-hmm. Yeah.

Orlando (42:11.32)
Yeah.

Orlando (42:14.925)
Yeah.

Oonagh (42:16.053)
If you, you know, if you’re a brand and you’re known for doing really quick social media videos cheaply and badly, then it’s fine, go ahead and use AI. but I think that it comes back to purpose. Like, you know, it’s not about, ooh, what can this technology do for us, but why would we use this technology? And I think why not how is a really simple example of responsible AI.

Orlando (42:30.114)
Mm.

Orlando (42:33.538)
Mm.

Orlando (42:42.52)
Yeah.

Orlando (42:46.154)
Why not how is great. Sorry, I just I just needed to type that down. That’s it’s it’s as simple as it gets. and I think that’s really terrific. So, what’s the point of differentiation then? What can a museum, a theater, community center, art community arts organization, what can they do that a tech company can’t? What’s their USP that that gives them that point of difference in the world as it’s changing so?

Oonagh (42:53.314)
Yeah. Yeah.

Oonagh (43:13.226)
Yeah, so I think one of the key things is I think one of the key things for me is that they serve as this really interesting bird space because they’re not home, they’re not work. but they’re also not existing to sell a product or service. and I think that’s really important because there’s very few spaces now that exist in the world that are those public commons, you know, that are, you know.

Orlando (43:39.864)
Yeah.

Oonagh (43:41.058)
Community libraries where you can just go and sit and be and you’re not expected to buy a good or a service. And I think so that I think that the non-commerciality, the kind of the lack of transaction is really important. But for me, I think you know this isn’t relevant to all organiz all cultural organizations, but for those organizations that are situated in conversations around contemporary culture.

Orlando (43:48.717)
Yeah.

Orlando (43:56.877)
Yeah. Yeah.

Oonagh (44:10.858)
That are situated in conversations around emergent technologies, then I think they are places where these technologies should be publicly questioned, culturally situated, and ethically contested. And I think the starting point for that is transparency about what technologies you’re using and why. And kind of unboxing the use of the technologies. Is that really low low barrier to entry? Just as you know, like

Orlando (44:30.872)
Mm-hmm.

Orlando (44:35.491)
Right.

Orlando (44:38.765)
Yeah.

Oonagh (44:40.059)
why not high is your gateway to responsible AI. just telling people what you’re using is a really progressive step to being quite transparent about what you’re doing. And I think the other thing for me is

Orlando (44:43.981)
Yeah.

Yeah.

Oonagh (44:57.534)
these organizations, cultural organizations, creative practitioners

Provide a really interesting commons and this opportunity to bring diverse voices and people together, to have unusual and unexpected experiences. So through my work, public programming has been really important to develop in my thinking about the impact of AI technologies on people and planet. So a few years ago, I ran a series called Curator Computer Creator.

Orlando (45:06.936)
Mm-hmm.

Orlando (45:14.488)
Mm-hmm.

Orlando (45:21.358)
about the impact of AI technologies on people and planets. so a few years ago I ran a a series called Curator Computer Creator.

Oonagh (45:33.406)
And these events we held one at the Barbican in London and one at Cooper Hewitt in New York. And we invited members of the public into a conversation about what AI might mean for their own lives. And so the event we held in the Barbican, for example, was called a long table. And it’s kind of a fill facilitation format. You have a long table, you have chairs around it. We had it in a public space. And there’s a microphone. And at the table are some extra.

Orlando (45:34.562)
These events we held one at the Barbican in London and one at Cooper Hewitt in New York. And we invited members of the public into a conversation about what AI might mean for their own lives. And so

Orlando (45:52.553)
Mm.

Orlando (46:00.107)
And there’s a microphone.

Oonagh (46:03.093)
Spirit.

Orlando (46:04.11)
Mm-hmm.

Oonagh (46:05.525)
But most of the chairs are free and anyone can walk by and take a seat and you can listen or you can join the conversation. And so we had this conversation, it went on for about four hours. We had different experts come, different members of the public come, people were at the barber can see an exhibition or to see a show or to grab coffee with a friend. And they were able to sit down and just have this free-flowing conversation about what AI technologies might mean for them.

Orlando (46:07.738)
And they would walk by and take a seat. I think during the conversation. Right.

Orlando (46:23.062)
Mm-hmm.

Orlando (46:28.77)
Mm.

Mm. Yeah.

Oonagh (46:33.435)
Nobody was selling them anything. It was just an unexpected encounter. It wasn’t clickbait. It wasn’t, you know, Facebook, Twitter, Instagram telling them that AI was going to kill them or that AI was going to save them. It was, you know, people in their twenties, people in their 60s. It was people that, you know, were there with young children. It was a diversity of experience and an open, honest conversation. So I think facilitation is something that

Orlando (46:35.704)
Yeah. Yeah.

Orlando (46:46.734)
It was, you know, people in their twenties, people in their sixties. It was people that, you know

Orlando (46:58.343)
Yeah.

Oonagh (47:02.559)
cultural organisations are really good at. The other thing I would say is cultural organisations can be really good at holding difficult conversations.

Orlando (47:02.712)
Cultural organizations are really good at.

Orlando (47:11.095)
give a holding difficult conversation. So I was curator of a program called Digital Dilemmas at the Red Cross Museum in Geneva seven years ago. Yep.

Oonagh (47:15.061)
So I was curator of a program called Digital Dilemmas at the Red Cross Museum in Geneva a few years ago, and that really sought to examine the impact of digital technologies through a humanitarian lens and looking at the impact of these technologies on society and people in conflict zones, but also in Western context.

Orlando (47:25.324)
Examined to be

Orlando (47:30.86)
Yeah.

Oonagh (47:42.08)
And so one of the kind of standouts for me from that programme was we had an artist called Marta Revolta.

Orlando (47:50.947)
Mm-hmm.

Oonagh (47:52.271)
and her work situates around autonomous weapons in conflict.

Orlando (47:59.545)
Right.

Oonagh (48:01.257)
And so she has a body of practice where she has drones.

Orlando (48:07.021)
Yeah.

Oonagh (48:08.863)
And she takes them into the gallery so that you, as a visitor, are targeted and followed and tracked. And suddenly this idea of autonomous warfare feels really present. And yet, even though you’re in the safety of a museum in Geneva, you feel that tracking, that tracing, that immediacy of what this means. And so you unbox this technology.

Orlando (48:12.67)
Mm. So that you as a visitor are targeted. Right. Wow.

Orlando (48:24.894)
Right. you’re in the safety of a museum in Geneva, you feel that tracking, that tracing, that impossible what this means. Yeah.

Oonagh (48:36.937)
from something you’ve heard about on the news to something you’ve felt in your person and your being.

Orlando (48:37.389)
Yeah.

Orlando (48:41.942)
Yeah, yeah, yeah.

Oonagh (48:43.871)
And so I really liked Marta’s work and we had talked about it at the start of the project. And I started a conversation with her about how she trains her model, about how it attracts people and like how she explains that. And so we got Marta to design a brand new workshop for us where she printed out hundreds of images of drone strikes from the Twitter feed of a state actor in conflict.

Orlando (48:56.995)
Mm.

Orlando (49:06.573)
Right.

Orlando (49:11.286)
Mm-hmm.

Oonagh (49:11.617)
And she invited members she invited members of the public to come to this workshop and to select, physically select photographs, and build data sets with those photographs. And so this was an analog workshop that allowed people to see the physical reality of what data sets are, about what omissions there are, about what decisions have been made, and being able to see how that data set could.

Orlando (49:15.852)
Mm.

Orlando (49:21.291)
Mm-hmm.

Mm.

Orlando (49:27.619)
Yeah.

Orlando (49:33.11)
Yeah.

Orlando (49:39.652)
Yeah.

Oonagh (49:41.078)
Create an autonomous system to identify a person or be more nuanced and identify a combatant or a child, but also to identify how it’s quite difficult to identify the difference at that distance between a child or an adult male. And so suddenly now this abstract conversation about autonomous weapons was a physical, tangible data set in someone’s hands. So the next time they use an AI system.

Orlando (49:49.116)
Yeah, yeah.

Orlando (49:58.071)
Yeah.

Orlando (50:02.495)
But with a physical, tangible down someone.

Oonagh (50:08.446)
They’re thinking again back to the kind of my yoga metaphor, that physicality, that grinding of the technology and experience. And that space to have difficult conversations about the impact of these technologies, the knowledge, the skill, the experience, the facilitation in difficult situations is something that is quite unique to cultural organizations. And it’s a really important contribution that both institutions play, but also.

Orlando (50:14.722)
Mm.

Orlando (50:23.448)
Yeah.

Orlando (50:28.109)
in difficult situations.

Yeah.

Oonagh (50:38.196)
Artists themselves play.

Orlando (50:40.494)
Gosh, it’s it’s it’s so funny because I we use the term all the time, a feature, not a bug. You know, it’s it’s a standard tech term that now, you know, with AI has become ubiquitous. But it occurs to me that, you know, we spoke earlier about, okay, like if you’re doing governance, you’re slowing things down and tech is moving so fast, and actually are you slowing down and is that a is that a bug not a feature?

Oonagh (50:48.886)
Yeah.

Yeah.

Orlando (51:09.154)
But actually, it occurs to me that just hearing what you’re saying, that one of the roles of culture is to slow things down enough so that people actually can internalize what’s happening to them or what’s happening to them in their place in the world. Slowness in culture is the thing. I mean, a painting isn’t moving. You’re having to move very slowly by it. Does it tell you something about yourself? Does it tell you something about your community? Does it tell you something about the world you’re living in?

Who knows? But the very purpose of culture is to stay one step back from the actual crest of, you know, there’s an argument that the culture is always on the crest of humanity, always on the crest of the present moment. But actually, it’s not. It’s sort of one mini half-step behind. it operates in the realm of the human mind, the human spirit, the human experience. And so

I don’t know if I have a question there really, but it just it it occurs to me hearing what you say, it’s actually the purpose of creative people, creative companies and creative institutions to be that one step behind, to take a beat and figure out where the meaning lies in all of this.

Oonagh (52:26.021)
Yeah, I think, you know, like for me, I just think in this moment when generative AI can produce like instant responses to anything, art and culture has this really distinct role to play. and as you say, it’s creating this space for slower responses, but like this curious engagement that ultimately allows us to ask better questions and like and allows us to like reimagine the world, to challenge the world we live in and to explore

Orlando (52:34.861)
Yeah.

Orlando (52:48.226)
Right.

Oonagh (52:56.314)
what these technologies mean for us as individuals, but also for wider society, I think is such a crucial space. And arts and culture can do that because in many ways

Orlando (52:59.469)
Yeah.

Oonagh (53:09.97)
It is less transactional.

Orlando (53:13.058)
Mm-hmm. Yeah.

Oonagh (53:14.386)
And it is more about that commons, it’s about that engagement with ideas and people and society.

Orlando (53:18.636)
Yeah.

Oonagh (53:21.916)
And you know, you know, arts and culture is always struggling for money. you know, it is difficult to fund the commons. However, the commons allows us to build better societies, to develop better governance. And again, coming back to you know the hard business of it.

Orlando (53:25.676)
Mm-hmm.

Yeah.

Orlando (53:37.85)
And again.

Orlando (53:42.238)
Mm-hmm.

Oonagh (53:43.386)
We can build better AI technologies that serve people and planet and not just wealthy tech companies. You know, tech AI technologies are not inherently bad. You know, design choices are made. And actually, you know, if we as a society make choices, if p politicians make choices, if we make considered design approaches to that feature bug kind of dichotomy, AI

Orlando (53:48.013)
Yeah.

Yeah.

Orlando (53:57.358)
Yeah.

Orlando (54:10.286)
Hmm. Yeah.

Oonagh (54:12.997)
And be great. AI is in some instance great. The problem lies at corporate decision making, not in the technology themselves. And cultural organizations are able to make corporate decisions align with purpose, not profit, and they model what good corporate decision making can and should look like in wider society.

Orlando (54:18.499)
Yeah.

Orlando (54:22.006)
Yeah. Yeah.

Orlando (54:31.918)
Mm-hmm.

Orlando (54:37.698)
Mm.

Orlando (54:41.858)
Yeah. I mean, yeah, I mean, AI companies, they’re cheap, but they’re not free. You know, I mean, I like, you know, when you think about the token usage, I don’t think it’s priced correctly. I think it’s actually much higher when you think about all of the energy that goes into it. But they’re trying to addict you. They’re operating on that social media model. They’re trying to get to addict you to free intelligence. and you’re right, that notion of going into a third space where somebody isn’t trying to hard sell you something.

Oonagh (54:52.516)
Yeah. Yeah.

Orlando (55:11.354)
Is incredibly valuable. I just want to kind of say, you know, you you do you know your Arts Council England report mapped investment in AI and creative practice. And since 2019, Arts Council England has invested around 4 million pounds across 194 AI-related projects. So you’ve looked at nearly 200 AI-related creative projects. What’s surprising you about the creativity that’s coming out of those?

Or the relationship with the technology even.

Oonagh (55:45.128)
Yeah, so I think like you know, from that investment data, we saw a real range of perspectives and approaches. for me, I’m really interested in the geographic and kind of social situation of where those projects were realized. So I think maybe if I pick up on two projects, but an artist, Karen Palmer, for example, developed a project called Consensus Gentium. And that was an immersive experience that allowed

people in shopping centres, in retail malls, in culturally underserved areas in England to experience AI technologies.

Orlando (56:27.53)
Mm-hmm. Yeah.

Oonagh (56:28.723)
It allowed them to understand how tracking technologies are embedded into their phones, about how facial recognition is already used in surveillance technologies. what was really interesting about that project was that they had to pivot their onboarding because many people were worried that it was some for sort of immigration control.

Orlando (56:34.048)
Mm-hmm. Right.

Orlando (56:38.678)
Mm-hmm. Yeah.

Orlando (56:53.782)
Right. Interesting. Yeah.

Oonagh (56:55.811)
So actually the audiences that the artists wanted to speak to were wary of engaging in any type of surveillance technology because to them it only had negative complications.

Orlando (57:02.582)
of engaging in any type of surveillance technology to them it only had negative complications. Right. Yeah.

Oonagh (57:11.839)
And so her work then shifted to include a stronger onboarding process so that she could explain that the experience was to empower them to make better decisions about how they share data.

Orlando (57:15.308)
then shifted to include a stronger onboarding process so that she could explain that the experience was to empower them to make better decisions about how they share data. Right. Yeah. And so this idea

Oonagh (57:28.007)
And so this idea of really encountering people where they were, these were people that wouldn’t go to a cultural organization. But she went to where they were and was able to have, you know, really unexpected conversations about data power and politics.

Orlando (57:34.664)
a cultural organization. But she went to where they were and able to have, you know, really unexpected conversations about data power politics. Right. Yeah.

Oonagh (57:44.996)
And another another project I just flagged from that data set is a project called Women Reclaiming AI. And that was by an artist, Coral Manton and Brigitte Aga. And they were both women that had children that were bored of being asked questions constantly. And then discovered the joy of chatbots, you know. So back in the day when we were all asking Alexa, they were like, we are sick of being asked questions.

Orlando (57:50.766)
Mm-hmm.

Orlando (58:00.398)
Mm-hmm.

Orlando (58:07.81)
Mm-hmm.

Oonagh (58:12.806)
So, why is there now another woman in my house that’s also being tortured by stupid questions? So they worked with a collective of women and non-binary people to design their own voice assistants. And they wanted to develop a voice assistant that was sarcastic and was funny and used natural kind of colloquial language and had

Orlando (58:26.008)
Yeah.

Oonagh (58:38.578)
bit more of an edge than the submissive Alexa that was being produced as the kind of submissive female voice. So really, you know, bringing AI technologies into the heart of community groups and helping women to understand technology by unboxing the technology that was already sitting within their houses.

Orlando (58:40.429)
Mm-hmm.

Orlando (58:44.033)
Mm-hmm.

Yeah.

Orlando (58:51.48)
Yeah.

Orlando (59:01.238)
Right. Yeah, it seems the two of them, you know, consensus gentium seems to to help people know how to feel about AI systems, about how they observe, judge, or shape them. Whereas when we were claiming AI, then kind of asks, what happens when those same people are invited to redesign the system? Is that the kind of journey you’re feeling we need more of? This kind of education plus responsibility and or sort of awareness to agency.

Within this conversation.

Oonagh (59:33.017)
Yeah, and I think, you know, I think the arts always has a big ask, don’t they? You know, that way, you know, arts is expected to entertain, surprise, excite, and change the world, all in very limited budgets. And I think this again comes back to that piece around facilitation. Some artists work, you know, the Women Reclaim and AI project. It is a collaborative process. You know, it is the practice is about working with people and building new worlds and things. And that’s great.

Orlando (59:41.974)
Mm.

Orlando (59:54.774)
Mm-hmm.

Orlando (01:00:00.46)
Yeah.

Oonagh (01:00:01.553)
For other artists, you know, they may create work using AI technologies that is to be experienced but not engaged with. But again, that transition to agency is where cultural organisations come in in terms of that role as facilitator.

Orlando (01:00:10.573)
Yeah.

Orlando (01:00:20.43)
Mm-hmm.

Oonagh (01:00:22.001)
And so I think for cultural organisations, sometimes it’s it’s creative practitioners and the artworks themselves that bring people on that journey from education to agency.

Orlando (01:00:28.119)
Mm-hmm.

Orlando (01:00:34.558)
Yeah.

Oonagh (01:00:35.687)
Sometimes the curiosity is sparked by the creative practice and the agency and response is created by the cultural organization and their public programming and the facilitation that provides a sense of empowerment to people who have experienced the work of art.

Orlando (01:00:45.88)
Public programming and the facilitation.

Orlando (01:00:53.356)
Yeah. Do you feel, I mean, you’re you’re on the front lines of this, like you know, and do you feel optimistic when you’re seeing how people are responding to this and how people are picking up the tools? I mean, obviously, so many people are inundated with slop, and the big conversation has been about slop, but you’ve just brought up a couple examples from from the two you just mentioned to the Munk Museum of examples where real, you know.

Real creators and artists are using it in ways that are quite interesting. Is the slob conversation getting in the way? and are are you optimistic when you see these examples about what an AI enabled creative future looks like?

Oonagh (01:01:37.281)
Yeah. Well, slop is easy to produce, it’s quick and it’s cheap. Curious critical practice that makes people’s lives better and changes the world is slower, it’s more considered, but it’s also more expensive. And so we will always have less of the better stuff. That’s just natural order, isn’t it? And so I think one of the challenges for cultural organizations and I think also media organizations.

Orlando (01:01:52.91)
Mm.

Orlando (01:01:58.517)
Mm-hmm. Yeah.

Oonagh (01:02:06.997)
is taking on the role of the algorithm and curation and thinking about how do we let the cream rise to the top, how do we platform and build ecosystems that ensure that more people can access this creative creative critical content, this creative critical lens, this new way of seeing the world.

Orlando (01:02:12.791)
Mm-hmm.

Orlando (01:02:18.732)
Yeah.

Orlando (01:02:30.168)
Mm-hmm.

Oonagh (01:02:34.106)
And I think that’s the challenge that’s the challenge so it is trying to get the good stuff out there, but swimming through an abundance of slots.

Orlando (01:02:34.498)
Yeah.

Orlando (01:02:41.261)
Yeah.

Yeah. Well, that’s I mean you said a great thing right at the start there was that’s always the challenge, isn’t it? You know, there is there is always more crap and more mediocrity than there is excellence. And I think that every time these things come out, YouTube is a perfect example. YouTube isn’t suddenly so much better. There’s just people who have risen to the top across the board. The majority of YouTube still remains as it was day one. Crap, boring.

Oonagh (01:02:54.425)
Yeah. Yeah. Yeah. Yeah.

Orlando (01:03:13.592)
Boring basic crap and then imitations of crap. you know, and and I think that, you know, you’re right. And I think that you, you know, and you you mentioned trust so many times here about cultural institutions and their editorial eye is so much what we need right now. The more times there’s more crap in the world, the more you need institutions you can trust and taste makers going, Hey, there is something that’s better.

There is a better option, just like you need a nutritionist to tell you not to eat McDonald’s. There’s there’s vegetables growing in the ground that might be good for you.

Oonagh (01:03:48.309)
Yeah. Yeah, yeah, that’s it, isn’t it? And I think, you know, sometimes we just want people to tell us what’s good. And I think, you know, like I think that when it’s really hard to cut through, we seek out those voices that will tell us, you know, this is worth your time.

Orlando (01:03:57.293)
Yeah.

Orlando (01:04:05.858)
Yeah.

Yeah. Yeah. Well, Una Murphy, thank you so much. This has been great. I I could keep talking to you for another hour, but I’m I recognize we’ve taken enough of your time. I so appreciate it. And I I really, you know, like I said, I don’t know where you get the hours in the day because you’re in multiple ways, in multiple places, talking about and and implementing things that we really like to see here and and

I feel like I’m always saber rattling about and you know, you’re on the front lines of it. And I just think it’s deeply admirable and so fascinating.

Oonagh (01:04:44.749)
Yeah, brilliant. Thanks. It’s been really great talking to you. And I think, you know, we need to do this more. We need to talk. We need to have these conversations. We need to explore nuance. and remember that artists can help us see the world in new ways. And I think that’s really important, particularly in these times.

Orlando (01:04:56.931)
Yeah.

Orlando (01:05:00.748)
Yeah, absolutely. And you know what? I’m gonna say something I’ve never said on the podcast before, but you have an open door to come back whenever you want. If there’s ever something you’re doing that you’re like, I want to talk about this, give me a call. We’ll book you in immediately. Cause I just think everything you’re doing is so fascinating. And and like you said, I think it’s wildly important to be having the conversations that you’re both advocating for and having every day.

Oonagh (01:05:28.888)
Yeah.

Orlando (01:05:30.284)
Well, so thank you so much. I so appreciate it. And you know, and I I look forward to chatting with you again sometime.

Oonagh (01:05:37.539)
Great. Thanks, Orlando.

Orlando (01:05:39.17)
Thanks, Yona.