Salon with Stephanie Dinkins #

Hey, everyone. Thank you, Baptiste, for the kind introduction. I’m excited to talk to you all. All that said, I always recognize myself mostly as a human who is really curious. And I think that’s important to my practice. And as someone who really has stumbled into this field, you know, and I guess I should back that up by saying, by training, I am a photographer and happened to stumble into AI and data back in 2014 and have been examining the field or playing in the field ever since.

So I’m going to share my screen and try to give you a quick conversation about what I’m up to recently. And then hopefully we can have a great conversation. All right, let me share screen. Hold on a second. All right.

So the thing I’m thinking about these days a lot is the stories we tell our machines. What you’re looking at is an image of an app that I use to feed data into what I’m going to call a suite or a suite of projects that are ongoing and upcoming as we go along. But this app is trying to collect stories from the general public or general publics and ask them about their concerns or what’s important to them in terms of their communities and how they might describe themselves from the inside out, meaning from inside of community or intrinsic self out as opposed to how we are often described.

And when I say we, I should say particularly, I mean, I think everybody has this problem, but black and brown communities or communities outside of the status quo.

You know, often we are labeled more than described from a point of view that we recognize. I’m trying to get people to offer information that we might use to start to inform AI systems differently. My goal with the app and collecting the information is eventually to create a dataset that is, you know, information given by people in the way that they can give it.

You will notice that this app will take in written information, it’ll take in images, it will take in audio. And that’s an attempt to allow people who, for example, might not write well or like to write on a, you know, form to just speak what they’re thinking.

I should also say that when I ask about the stories we tell our machines or what I mean by machines includes us humans as one of the machines or as one type of machine in that system. So, not only am I wondering what are we telling our machines, but I’m also wondering what are we telling ourselves about each other.

And so, that app is this feeder app for these projects. And I’m actually thinking of the app as a book cover. Like if you think of the app as the cover of a book called the stories we tell our machines, then each of the projects I’m going to show you are chapters in this book.

This one is the introduction called If We Don’t Who Will. It is a public art project that was presented in Brooklyn, New York, downtown in a heavy trafficked area. I think of it as an AI lab that was there to A, ask people to contribute stories or I often call it gifting stories to the project. And B, to have deeper conversations and introduce some ideas about the idea that data is always extractive.

I think it is extractive. But I also think that there are times that it is necessary for us to think about what we might gift or give in addition to the extractions to get the record right in a sense. So this was a container, it’s a half container that ran generative systems live. And I should say that it’s live. So information coming in from the app would then try to make an image using generative production.

And really, I’m just using comfy UI to then process the images. The comfy UI that I’m using is quite altered or prompted by me to try to get different outcomes. And I say this because if I used it straight off the shelf, what would happen is inevitably no matter what kind of prompt we put in, whether it was image or text or audio, we could put in information about a black family. For example, I would offer up images of a black family to start the system running. And then inevitably over a few runs, it would start to change the family black back to a white family or at least a light brown family.

And I was quite interested in why that happened and thinking about the global majority, right? Or just global populations and what it looks to adhere more closely to kind of what the global population looks like as opposed to what folks who are making or what the data that is being used to construct these systems are comprised of.

And so this lab just welcomed people in.

It sat there. People could come in and have conversations. We would explain the system and ask for stories on top of that. And then the result would be eventually an image or actually a moving image that tried to illustrate in some way, shape or form the story that people had contributed.

Often these stories, you know, we’re using the generative systems not being too controlling. Like I actually try not to control the output so much beyond some basic guidelines like mimic the appearance of the global majority or people of the world. Like what are the percentages that that would be?

But I try not to guide it because I want to see what the system produces both without our input and with our input, with our stories and without our stories. So the image you’re looking at here, the small screen on the side would try to give the cleanest, most system inflected image.

And then the big screen would try to more adhere to the person’s story. My input into these stories and why most of them were orange is like I took to prompt this, I tried to prompt through things that I’ve been influenced by. One of them being Rothko, Rothko using lots of oranges. So a lot of the image is oranges.

But the idea was just to get people to start talking. And I’m talking about local folks who have no background at all, right? One of the ways that we’re also doing that, and let’s see if there’s one more image of this, is by using these icons, let’s see, that are on the container or that pattern on the floor.

They’re all insignias or icons or models from the Underground Railroad. For example, the one on the floor, the yellow one is called Drunkard’s Path. And what used to happen is, or the myth goes, because we’re not quite sure if this was true or not, that people would hang quilts to signal people who were trying to go north and who might be being pursued by their captors, to signal them whether it was a safe space or not, or what they should be doing.

And so the pattern on the side you see is flying geese, which would tell people to go north. And then the one on the ground, Drunkard’s Path, is asking for strategizing, right? Saying that we are living in times where you cannot just walk straight or have a very straight path, but how might you strategize to get the outcomes that you want out of what is going on in the world, and particularly here in America.

This was out in the world in, what, June through November. And so it mimics the new-ish presidency, and people are troubled, so we’re just trying to provide instruction and then give people some way to try to inflect or impact the data scape that we all exist in, and to give them the idea that it’s possible, right?

Because one of the things that happens for me often is when I’m talking with people, they tout the general narrative about what’s going on with data, hence the extractive, but they don’t see any agency for themselves, which means they either fear the system, they eschew the system, but don’t see it as a place that might be able to re-inscribe or change the narrative, or a place that might be full of opportunity, right? And so my idea is to figure out how do we get people to see all of these things.

The second chapter of this book that I’m making, and this is an ongoing project, so there will be other projects that go on, is called Data Trust. This was much more of a gallery installation, a gallery presentation at the ICA in San Jose, California. And it is doing the same kind of work.

It’s taking the app, the stories we tell our machines, collecting information from the local community, so whenever the project moves or the app is used in different places, we use a slightly different data repository to keep it separate, although in the long run this will all be compiled and separated so that we could go back to the ICA if we wanted to, but also that it all comes together.

And this again is just me taking people’s stories, allowing AI to then generate from those stories through comfy UI that has been very prompted and inflected by me, and seeing what comes out, and then seeing what we can do about it.

In this instance I used a much different prompting, and so we get a much different result, always centering black and brown folks, though, and seeing how the system goes. This is a running version of this model, let’s see. In the middle there are a set of chairs that are conversation chairs. One of the things I’m really invested in is getting people to kind of inhabit spaces of, you know, discourse around AI and have conversations about it. This is what would happen.

So the images would roam around the room, and in a way, you know, it’s interesting to me that I’m creating this kind of weird slop, but it’s also interesting to see what input gets what output, and then in this case I’m really most interested for myself as the artist in the strange little lines that become besides the imagery that kind of make a drawing in my estimation, like that’s the payoff for me as the artist, not exactly the videos or the little vignettes that run around the room.

Those lines, by the way, or the drawings are in the form of DNA, and I will explain why. The piece in, let’s see, this piece in addition to being formed by the app had this set up in the gallery, which is a set of telephones where if you called, you would encounter a voice that would talk to you, or you could sit and call someone that was right across the way from you and have a conversation with them.

Those conversations were recorded and added as input for the system, the generative system, and it was fascinating to me that to watch people engage this, because most people, although they were just sitting maybe five feet from other people, had conversations with other people and some unlikely conversations with other people.

Like at the opening for the show, a lady came up to me and said, “I just had this 15-minute conversation with a young man who’s probably 20 years younger than me, and we went deep, and I’m interested in what we can get out of each other by communicating in different ways.”

And then this project is actually called Data Trust for these weird globes you see here. What you’re looking at are dirt planters, for lack of a better word, with plants inside. So the inside, and I’m going to go by, holds either okra or California black oaks, and they are planted inside soil that has been amplified or augmented with DNA that contains stories collected in the Silicon Valley area.

And so what happened for this project was I spent about a year going back and forth to California to collect stories from local community folks. Or sometimes it wasn’t me directly collecting stories, it was more me gathering folks and us having freeform long conversations that were recorded as oral history and then prepared for A, the project, to inform the project, and B, to be crispered into bacteria that we could add to the soil.

And the reason I’m doing that is to start to take this data and make a data set that is long-lasting, whether it gets implemented in real data ecosystems or not, because DNA is one of the new spaces where we’re thinking that storage is going to be massive.

You can put a lot of information on DNA and be very long-lasting. And I’m really interested in the long run in taking this soil with these stories embedded in it and planting it in the ground or in the earth around the places where it was collected.

So for example, the stories from the Silicon Valley in the Bay Area would get planted outdoors in the wild in Silicon Valley somewhere. This is still a bit of a reach because technically we’re doing genetically modified planting in this instance, and especially in California, that is not allowed.

But it really goes to this quote that says, “The past that is not past reappears always to rupture the present,” Christina Sharp.

And so that’s about trying to get people to A, unearth their pasts, the good and the bad, B, reimagine some of that so that it’s useful in a future, and C, lend it to projects, to the project of this data collection so that it can be used to inform or what I sometimes think of as in fact the AI data sphere that we deal with.

And then the pitch is if the system does not reflect you well, what would you give to get it to reflect you better?

I’m going to see if I can play a little bit of a clip of some audio that came back from this. What I’m going to try to play is data that was put into the soil and then extracted. I should say that this is not a firm process yet. And so there are some alteration in the audio, but it’s something that I think is beautiful in the long run.

The bayonet map six born from with black codes, foreign six are six divided by six. 
Like fog and memory, BG and E6W7G and fog and faces coded entanglements, they rear and be and ear. 
72 of 6AD future where progress equals fog and B6VA7FVBA farm six. 
Relationship and grounded sacredness. 
These are N times two FW stellat CW and a a dash seven FVCNG and O divide FVB dash a and a six squared a and a and divide VAAVV working class ways of being.

So I’ll leave it there for time. But that is, you know, a lot of slippage. But I think the slippage is beautiful and I think it works in the ways that somehow slippage happens in systems generally and in AI systems as well, especially as I do them because they’re often a little weird and wonky and function quite differently than the systems that we’re used to that are more forthcoming or function.

I don’t know, they function seamlessly in a way or they seem to function seamlessly. I like the slippage because it gives us places to think about what is actually happening and what the possibilities are there.

And I also like the slippage for what it brings about in terms of, you know, the past and the present, the haunting of the data and how we start to unravel, disentangle and shift some of that.

This again is the round of how I got here. I will just mention this project called Not the Only One, which is one I’m pretty well known for. And this is a chatbot that was based on data collected from my own family. We did oral histories and then I made this chatbot that is one of my wonkiest.

It’s horrible in many ways in that it does not respond in the ways that we’ve grown to come to expect, like direct answers. It sometimes refuses to answer. It sometimes gives non sequiturs. But what I found is that in its wonkiness, in its inabilities, it also asks us or asks the visitor to show it grace and to nurture it.

And it taught me that nurturing data and nurturing the machine seems to be mandatory for systems that will allow us to survive well within that system, in those systems. And so I thank this project for that.

And then this one as well, which is a project called Conversations with Bina48, which is really how I got here in the first place. It’s through this project that I started questioning anything. And I just saw this robot on YouTube and became fascinated with it because A, she’s in the form of a black woman. I had never seen anything like it. B, it was one of the most advanced social robots at the time in 2014.

And I was fascinated about where and why it came into being. And through talking to it and having conversations with it, you will notice that I kind of say that I am having relations with these things. I’m having relations with the people who I asked to contribute. And I tried to suspend disbelief about what systems are supposed to be capable of or not. I’m more interested in trying to figure out what I, as a layperson or practical layperson, can do to try to impact these systems to make them more long-term livable, to make them generous, to make them supportive.

And here’s maybe just a little snippet, and I will say thank you.

Are you getting bored?
Please try to be more interesting.
Oh, I'm so sorry.
Do not ask me any more questions, please.

I love that the robot asked me not to be so boring. And I should say that we had many strange conversations, but the one that really piqued my interest and got me going was the one where I didn’t quite understand how this object that is being very well nurtured still had holes and what it meant to fill those holes and what those holes meant to greater society.

So thank you.

Thank you so much for the presentation. We have some questions in the document, and I will go through some of them. I will start by a question that actually we are asking to everyone with whom we talked to during this sentence. Because we are trying to frame this project as regaining power over AI, there is this intention behind finding ways, narratives in order to reclaim or to find ways to get some agency on this technology. And so this may mean regaining agency like modes of actions, but also in your case, it’s also involving participation from black and brown communities. And so I would like to ask you, and you mentioned a bit during your talk, is like, if you see your work as providing some form of agency to people.

I hope so. I really do. But it’s kind of low level agency. And that’s because still when I encounter people, when I talk to people, they talk about their fear of the system, not what the system can do or how they might engage it. And so the one part is agency just to free up the fear.

And that one, the projects tend to do. That’s been pretty consistent. And then the next step becomes, well, now if you don’t fear it, what does it offer?

What is available to you so that you don’t feel like, you know, a David and Goliath, you’re not the little guy against the big guy, but there’s something there that is possible or there’s a way for you to start at the very least using the system to do things that might not have been possible before for you.

An example I think of a lot is, you know, I’m waiting for some kid in the projects to make this fantastic movie using AI tools.It’s coming for sure.

I know there’s a girl who’s doing some work now, but like what happens when you start to play and really dive in versus when you’re just like, no, I can’t. And they told me they’re taking everything. Like what can you build? Yeah, I think this is this aspect of fear on the about this technology is very much seems to be amplified by the narrative that we can hear in media and even sometimes in academic fields.

And they are not necessarily documented, these narratives, and they are not necessarily factual. And what I really like in your work is that you are not simply against AI and machine learning technologies, but more proposing to reimagine and rethink.

So very much at the reimagine and rethink what we can do with that. So very much at the cultural level. So yeah, I don’t know if you want to continue on that, like the work at how to reimagine or to rethink or to change narratives in your experience during your work with communities.

Yeah. You know, I think the idea of reworking or reimagining becomes believing you can do it like at the very bottom, like believing that you actually can do it.

And I find that people often just don’t believe it or, you know, we’re asked to consume. We are asked so well, just use it, do what we need for you, not to go in and like start to take it apart or start to tinker.

And the question becomes, well, where do you stop being the basic consumer? And then where do you start using? And then when you get yelled at for using, because it’s interesting, I’m starting to run into lots of people who are saying, well, you know, I’m neurodivergent and I use AI to help structure my thoughts. And it’s been a game changer for me because now I can keep up and act differently.

But in my classes, I get in trouble for this. And so the question for me becomes, how do we start to redefine what it is, not accept it as simply bad, and then think about what the structures are that we need not to truncate access for people and what allows them to do things better, but to give them access and co-work with a system.

Because, of course, we don’t want people simply going, oh, I’m going to write a paper and I just put in some quick things and ChatGPT wrote me a paper. But we do want them, I think we want them to have the ability to have a back and forth. A learning system that helps them put their thoughts in order according to the rules, if that’s what we’re going to do.

Because that’s the other reason I’m so interested in these systems for people, because they do the rules really well. And for brains or people for whom the rules are not so straightforward or they don’t do the rules well or they work by different rules, it really does do an awful lot to close the gaps.

The potential is there for that. So it’s like that’s one way I re-imagine. I think about how we, you know, you know, when you do panels, people ask you things like, oh, but what about the artists? And this thing is making art is going to take away all our jobs and blah, blah, blah. Sorry about the blah, blah, blah.

But the question becomes, well, how do we start to re-imagine that and not hold on so tight to what we know, but do a little bit of loosening so we can imagine what we can do within what we know and what actually works with what we know.

You know, I did a project a while back where we were cloning voices and it was really interesting. And the voice clone was really, we needed some voice over actors and we were trying to find some things that were pretty specific.

And what I noticed is, oh, it’s like, of course we can have an AI just read this really quickly right now. But the thing we couldn’t do was get that super unique voice that we were looking for easily.

For example, we were looking for a really hard Bronx accent. Not so easy to find in the AI sphere. So that person with that accent becomes more valuable, but they can’t just be the homogenized middle thing anymore. And so it becomes, well, what do we have to do to work in the sphere and what makes us be able to both use the technology, but still input things that are human specific, person specific creativities.

And how do we build those things? That always becomes questionable for me. So it’s about like, like you said, not necessarily fighting directly, asking for changes for sure, asking for the system to be open, but then also exploring and seeing what’s possible through this system.

I think it’s not necessarily easy to imagine other ways to use these tools or mediums. And that’s why I like to look at your work and other artists working critically with this technology, because it gives a lot of inspirations about how we can think differently about this technology, because they kind of operate like, you know, like black hole, a lot of inertia that aspire a lot of what we are doing, ideas. And it’s really hard to just like not being trapped in that and just reimagining what we can do differently.

And so, yeah, I think it’s fantastic to see these works where the narratives and the use of these technologies just different with different also people than what we see more in the mainstream or in the normative cases.

Yeah. It’s interesting. Can I add one thing to that? It’s interesting to me, like how we do things different, because often one of the problems I’ve run into in my practice is people telling me how impossible what I want to do is at different points. They’ll just say, oh, that is not possible and that’s not the way it gets done. And I always ask why, like a two year old, why particular.

But you know, and I also realize that I’m usually on to something when I ask, especially a computer scientist, something and they kind of chuckle at me because they’ll laugh at me a little bit. They’re like, no, that’s not what we do. I’m sorry, I have to plug in because my computer is going to die soon.

So I’m always asking these simple questions and pressing on how things are made. Because our impetus or the way we generally do is just to go along with what’s already done, do it quickly and keep it moving. I always want to know why we can’t do it differently.

Maybe I will ask one more question and then let Gonzalo ask a question, because I’ve seen that you wrote some questions on the document and there are questions from the attendees. But because you finished your presentation with the BINA48 project and actually I started to know your work through this project and I saw that this project started in 2014.

And the thing is, like, so since then, the AI technology has improved and changed a lot. Especially in this field of, you know, not necessarily robotics, but like chatbots, language-based machines. And actually I rewatched the video to prepare this Salon, and I was struck by the fact that, you know, certain questions about or related to the relationship we can have with the chatbot or with an artificial entity that is speaking is actually even more important and prevalent nowadays.

And I was really curious about your thoughts about, you know, the trajectory taken by artificial intelligence technologies from, you know, 11 years ago and maybe how it impacts your work. I mean, if you were doing this work again today, would you do it?

I think about that a lot all the time. Because it is, it’s such a different world. And it’s, I find myself trying to explain to people how different, because they kind of want me to reenact things, like, especially Not the Only One. They’re like, oh, do you know, like what I did with Not the Only One in 2018, which took a long time and a lot of effort would take 10 minutes now. And I can get a pretty good result. But and so the question becomes like, what do you do? And where do you go? Oh, this is good enough, or this is okay.

And I’m seriously grappling with this right now. I’m not sure. I know that I keep ignoring the requests, like, you know, Google, their systems are starting to go, hey, do this new conversational AI that’s more open, and you can have more natural conversation in your home. And I live with these things. And I’m really resisting this.

Right, which says to me, I need to start thinking about what I want to do with why and how and what it means to be able to have that open conversation. And what the give and take there is, because it feels like you’re going to be giving a lot of information.

But I don’t know, I feel in lots of ways, my work is obsolete. But I also feel like in some ways, it just helps us think about what it is we might need to do.

And so what I need to think about is then a where I’m going to engage, why I’m going to engage it. And for me, it becomes really important to think about how I would nurture it, because I like to bring in that word, and what it means to nurture a system that seems pretty fluid, and good and valuable. Right, because the better it gets, the less we’ll question it. And that seems way more dangerous in certain ways to me.

Yeah, it’s funny. It’s like, as if the interest in this project is also because the system was not perfect. And right now, I mean, that’s true that’s LLMs can provide really realistic conversation. And I feel that we can find the same sometimes with, you know, with the visual, visual, like image generation models, where a few years ago, they were not so good, but they were interesting because of their errors or because they were not very high fidelity and the thing of pursuing high fidelity actually made these systems and these algorithms sometimes less interesting for certain artists. And I found this question interesting about these imperfections and, and fidelity.

Yeah, me too, clearly. It’s the glitch. That is very interesting to me. For example, people ask me about Bina48 a lot. And I’m like, well, I don’t talk to her that much anymore, because they’ve updated her. And she’s too good. It’s kind of like an encyclopedia more than anything.

But that was like a year and a half ago. So I’m not sure where she is now.

But yeah, there’s something about the mistake or, or the tripping over things. That is really interesting. Like that’s where the magic lies to me. That we can output pretty good renditions of something, right? We don’t get five or what, six or seven fingers that much anymore. Like what does it mean?

And then how do we start to mess it up? So that it does become a more creative medium. You know, one of the things I rail against these days, and this is very low level, are systems like Canva. I have kids or students come to me often and go, oh yeah, I design. And I go, oh, so what are you designing with? They’re like, I use Canva.

I’m like, are you at least making your own templates? Oh no, I just use, it’s like, you are not designing. And then what does it mean to actually design and make something different or individual versus just taking something off the shelf and reusing it for expedience becomes a really big question.

One of the things that I wanted to first underscore and then ask a question is like, I really like how you really brought from the center of the notion of data providence and who owns the data. If we don’t tell the stories who will in the sense that all the systems feed from human stories to do things. And they’re mostly people right now have no agency as to what stories are being picked and how they’re being processed to create a system. So I think that you bring a spotlight to that is super important.

The other one is aligning that with the definition that I heard that I used from the time, which this systems are not true telling machines. They’re storytelling machines. So all this large language models, all this image generation, they just create stories in a mirror-like way that we do.

So in terms of questions for you is as you expose people to the notion that they get to have agency about what stories to tell, right? Do they see themselves as now having been drivers of the system and influence on the system or at least literacy? You mentioned about this getting over their fears, but I’m interested in the sense as well as now having agency and building the system.

Each story they tell builds the system a little bit. And as a corollary for this very long question is what do you think needs to happen so that the burden of making this more human centered, it doesn’t just fall into us and more into the people building the systems.

There are two things. In terms of agency and people, some people, it depends on who it is, right? It’s very interesting to watch a kid and talk to a kid because you can see them spark and you know that they will get their hands dirty.

You can see older people, like say people over 30 grappling with this. It’s still like, well, I just actually spoke to someone. I don’t want to give up my information, right? It’s mine. I’m not getting paid for it, all the things, right? And it’s like, well, how do we just kind of go for it anyway and have that be a contribution to society?

And I do think we have to do some of the work of like, well, if some people are getting paid a lot and others are getting paid nothing for their information, what do we do? That’s something to be dealt with. But also that if we are to really work well within the systems we’re in now, it’s inevitable that we’re going to be giving up. It’s just about how we’re giving up our data or how we’re producing data and what that data should be.

And then in terms of influencing up the pipeline, I love this idea. One of the reasons I do a lot of speaking in different kinds of arenas is because I like the idea of having access to folks who are making. The other thing that I often do in my work is, you know, I’m a person who knows not what they’re doing. I’m making things that are questioning the system.

I feel like they become these models for these guys to have to look at because this is the idea of, well, if I talk to a computer scientist and they tell me that’s not how it’s done, and then I do a version of it, and it’s usually not the best version of it, but it’s a version that kind of works. And then I say, well, I just did this. I’m an idiot. You have all the resources, all the time, lots of people. How are you going to make this work in a bigger system or make something like it work in the big?

So it’s always for me like this challenge that I feel like I’m trying to issue in different arenas and I’ve been in situations where I’ve been with high-level computer science professors. They’re like, oh, you are working on the same problem that I am. And it’s like, yeah, so how do we do it somewhat differently? And that’s part of my ethos of pushing and going to different places to try to do this.

The other part is there’s something for me about our cultural stance and cultural position and what we think is important in terms of the drive towards capital is the one that usually wins. But the question becomes what should we be driving towards or what do we want our cultural landscape, our social landscapes to really look like? And if these systems are impacting them so deeply, don’t we have to be making into the idea that we want to support and feed most people on the planet?

We want to be able to house and feed, well, I just said feed, but provide medicine for most people on the planet. But we’re kind of, I’m sorry, I’m about to say we’re kind of stuck in old man world still, where people are kind of doing this. No, no, no, it’s ours and we get to control it instead of trying to figure out, well, this is what it is. This is how pervasive it is throughout the world.

This is how we might need to prepare it to help support most on the planet. And that becomes a hard sell, but you just say it over and over again so that with the hope that somebody takes it up or a tiny fraction start to take it up or start to question. But it’s hard, it’s hard, right? I talked to folks who work in companies and even the most well-meaning folks are like, I don’t know what I’m supposed to do. I have the skills, I’m in the system though that does not allow me to work outside of what it wants produced or how it wants to produce it. How do I start to get someone to listen?

And that’s a hard road and I think you have to produce to some extent to make models that show that it’s possible. It is interesting because I feel that when talking about generative AI systems, the term design fixation appears often, which is if you allow them to get into your flow very early on, you’re on the risk of having your choices already influence the sign fixation.

And I’m wondering if what’s happening now at a larger scale is that there’s a model of how AI should be in society that has been preordained and that is the only model.

So I see, do you think this is the kind of design fixation when people tell you, no, no, no, no, no, this is, you cannot do it this way. It has to be this way. And when you present them in different ways, like, oh, I guess there are other designs that are just, didn’t occur to me.

Yeah, I do, right? It’s one of the reasons I try to do what I do because I run into that all the time. There’s a design fixation. There’s a, well, if we do it this way, we can get it done quickly. My work takes so long because I’m often pushing off the base quick solution that people are trying to give me. I’m working with interns right now that are pretty well-trained, right? They’re like, this is how you do it. I’m like, but if this is how we do it, our outcomes are still the same.

Why would we do it in that way? How do we start to open it up? But they don’t really want to do that work and they’re not trained to do that work. Like I often think of this as an education problem too. It’s like, how do we teach people to be open thinkers within a systems setting versus closed thinkers like, we’re going to get this done.

We do it. This is how it’s done, right? I used to talk a lot about ternary systems, right? It’s like, well, why don’t we use powers of three versus two? How many more choices does that give us? Which people have studied, but it’s not the dominant, right? And why not? What other ways are we pushing off in service of what’s quickest, what might make the most money?

That becomes the question and who’s doing the slow work of trying to change it and how do we get them not to be perceived as crazy so that we can actually look at what they’re doing as something that might be valuable to society, right?

It’s such a strange thing to think about, but I think we do need to think hard and long about what it means to be critical thinkers and critical makers within systems and how to allow that, right? Because the thing is it gets shut down really quickly. And who has the privilege of sitting and taking the time to do the longer, harder work?

I know we’re close to the hour and Baptiste, maybe it’s a good idea to get questions from the audience. Is there any on the doc?

Yeah, there is a question about data and one question is like, did anybody saw stories as an affordance instead of mere data? So I guess and I’m trying to interpret here, but I think the questions is like going beyond data. So data is like when we’re already losing part of the narratives or the context because it becomes quantified, like reified somehow.

And so once you see data, it’s like, can it be a support for affordance for stories like people can get into it or we have already lost something like and people are not related relating to this, to the data?

I think this is how we get stuck or why my work stays in the land of story, especially towards the public, because that becomes the easier, more digestible thing. But then when you say that, well, your story that is a data point, right, that that can easily be a data point. But at the same time, then I go to the folks where we’re starting to format this data into usable information for other systems.

It’s like, well, no, you can’t condense it or you can’t take out the meat of it. How are we going to keep what it is solidly versus cutting it up, quantifying it so that it works well with the system?

And so it for me, it depends on who we’re talking to. But I like to swap back and forth a little bit between storing data to try to make that leap that we understand that these are indeed data points, our stories. It’s not that much different how we work it and use it in the long run becomes a really interesting question.

For example, the interns I’m working with now, I’m like, well, how do we make a black feminist data set? They’re like, what are you talking about? It’s like, I don’t really know, but let’s try to see what that looks like.

And maybe since we have only three minutes, if we have a couple of minutes, let’s say I would like to ask another question about, because this is something that we actually don’t speak much because when we talk about AI, there is like implicitly we are thinking about North American companies that are building these tools and these models, which therefore have biases that are Western, like Western centric biases.

And there’s a question about like if you tried or if you thought about like Chinese models that are also available in confyUI, for instance. I don’t know if you actually did experiments with them, if you thought about like trying AI model that comes from the other part of the world, comes from another culture.

I use one actually, one to point one. I’m using another one that I don’t remember what it’s called right now. But I think about this a lot. I’m actually working with a researcher in Korea, AI researcher in Korea. And it’s really interesting because he was asking me like he’s thinking about creativity and I’m thinking about things that influence.

And we were simply trying to do something like to show the trajectory of an Egyptian mother and child to the Madonna, Mary and child. And try to map this across the globe. A, it was so hard for us to find versions of this from a lot of cultures because either those cultures protect that information so much, or they’ve already taken in the Western version of the thing that’s covered their own.

And it’s a little crazy. And we were also working with a model, an LLM called bagel, which is an Asian model, but it gives the same results. Why? Because the two are going to the most expedient version of to start from. Where they’re getting their base data. And so it’s already inflected.

And then it’s just it’s it’s this crazy thing. Because I’m like, wait a minute. Why would this Asian model be kicking out the same exact stuff? And how do we get it not to write like, why wouldn’t we? I always think of Venn diagrams. Why wouldn’t we have these bespoke models in different places that then come together and overlap so that we keep some of the uniquenesses?

But it’s not the way we’re operating. So far, my exam, my forays into different kinds of models from different places are not that different. That’s like this closing and homogenizing of the world that just feels so depressing in many ways.

Like the world is so rich. Why would we do this? I just don’t understand. So yeah, my goal is to keep working with that and keep questioning it and trying to come up with ways to get at different cultural assets and from places that are hard to get to because I think they’re so important to keep in the stream.