INTERVIEW
Source of Inspiration
Elisabeth Bohnet: Olivier, you publish AI-generated furniture and objects on your Instagram account ai.furnitures. What is your profession and what interests you about designing with AI?
Olivier: I trained more in the field of international relations. I've since changed direction and founded a publishing house and a music label with my colleague. At the same time, I have developed my work as a photographer, without any academic training. So I'm not working in the field of design.
Regarding your second question: I myself wasn't necessarily receptive to what you see of midjourney and artificial intelligence in terms of aesthetics. I still think it reflects a lot of the mainstream trends that exist in our society. As far as I understand AI, it's really more about what AI can capture and process and then pass on. For me, it was interesting to see how you can play with Midjourney to get results that are not mainstream and reflect a different set of facts. I allowed myself to experiment a little with design and furniture, which has always interested me, and was relatively surprised that there weren't many people doing this before. I don't have great technical skills and I'm not a designer, so of course Midjourney was a bit of a cheat. I was able to achieve some pretty amazing, pretty interesting results quite easily. I also find it a great source of inspiration.
How did you come to design furniture using AI? Was it pure interest?
Olivier: I think there was a kind of frustration. My girlfriend and I talked a lot about it, especially about the fact that we could design furniture or decorative pieces that we actually liked and that we could cross all the influences that we were very interested in. Of course, you usually can't get over the material and technical hurdles. In the end, it made sense to familiarise myself with Midjourney and the tools of AI. There is something very fascinating about exploring new things. It is a relatively new world that is full of possibilities. However, you also quickly realise that it is very parameterised. So it was interesting to discover the possibilities of freeing yourself from the many limitations. As I said, I really see it as a source of inspiration and not an end in itself.
We are still in the early stages of AI, but it will be interesting to see if, in the long term, designs generated by AI can be translated into models for 3D printing. I don't know what applications there might be, but I'm very curious to see where progress in AI at this level will take us. I think there are still a lot of surprises in store.
What is your process? Do you have a specific object in mind? Do you add images that you like into Midjourney to generate a style that you want to see?
Olivier: The process can look very diverse. Sometimes I have concrete ideas. Sometimes I have ideas for prompts that are new, or allow me to lose myself, to ask something different from the AI, with results that may be disastrous. That's the game, that's what the designer is experimenting with. It may actually be that I feed the AI with images. Personally, I really like the idea of experimenting on a semantic level to achieve certain styles. There is an appropriation of certain terms and words that you can't use because they have too strong a connotation. I don't have a specific example right now, but there are words that lead the AI to results that I don't like aesthetically, for example. Today I know that I'm not allowed to use them. Sometimes when writing prompts, it's necessary on a linguistic level to go round the thing a bit or make several attempts to see which terms are associated with certain aesthetic realities and which are not. It's a kind of search for clues. Similarly, there are already some terms where I know I can potentially get results that are close to what I have in my head. It can be very exhilarating, because in the beginning, when I was messing around with Midjourney, I had results that were really very different, and sometimes there was nothing to see at all. It's not like that anymore. When I formulate something in the header, the result is pretty close to what I had in mind. I think we've become a bit of friends, the AI and I, which is a fun thing to see.
For me your designs on ai.furnitures are very different from many other designs. As you said, many other designs on Instagram are stylistically very futuristic, very technical. Your designs are dominated by natural materials such as wood, stone, glass and even leather. The designs look a bit retro, which I like. I recognise a bit of the style of the 50s and 60s or sometimes references to Art Deco.
Olivier: Of course, there are some decades that I like more than others. The 70s, the 80s, the whole Art Nouveau style, the Wiener Werkstätten, these are always influences and inspirations. I'm a big fan of Vienna and Art Nouveau in general, these are things that really appeal to me.
So "Jugendstil", is that an instruction you give to the AI?
Olivier: Honestly, I don't know if that ever happened. For example, I think the 70s influence me even more. It's constantly changing. The idea is really to let yourself drift a little bit, not to be too restrictive because the tool allows it. It's vagabonding on that level, I like that.
What you said about futurism in relation to AI is interesting. I think there are a lot of people who use AI as a way to create entirely new worlds, which is of course more than possible. I personally like the idea of playing with AI to design things that are plausible and possible, even if none of my AI creations have materialised in real life at this point. It would be great if that happened one day, but we're clearly not at that point yet. In fact, there are AI creations that are not necessarily recognised as such. I had the opportunity to work with a friend who was really surprised and said, "Wait, is that AI?" I stood there and thought, "Yeah, it says AI and ai. furnitures on it," but he had only seen the picture and then was a little disappointed.
It's interesting, there's actually a bit of that feeling of disappointment when you realise that an object doesn't actually exist. I think: "How much longer? Maybe it will change soon". For me, it's actually this idea of trying to play a little bit with what's real and what's not real. But more than anything, it's aggregated inspiration.
That would be my next question: What are the goals you are pursuing with the designs? Do you want to realise these projects?
Olivier: At the moment it's more of a game and as I mentioned, it's about creating things that then give me new ideas, which in turn create other ideas. There is a kind of domino effect that encourages aesthetic reflection. The goal, of course, would be to maybe realise one of the pieces one day. That would be great. But at the moment, that's not what's driving the project – if you can call it a project at all.
What do you think about copyrights? For me, the question arises on two levels. Firstly, at the level of input: What images do I give the AI? And secondly, at the output level: Who owns the rights to the images that the AI spits out. Do you have any thoughts on this?
Olivier: That's a very difficult question to which there is still no answer. Even at a legal level, it's complicated. I don't believe that the AI creations belong to me, they are not my creations. I'm just operating a machine. Whether that's enough to make it something that can be appropriated, I don't know. In the same way, it is problematic to see AIs in other areas using content created by humans and then making it publicly accessible. There is a kind of copyright exemption here that is problematic. Among the creations that exist on ai.furnitures, there are very few that are derivatives of things that already exist. That is, they are prompts that are written rather than the input of a photo.
Of course there is a form of authorship because it is my idea, but it is the execution of the machine. Is it still my idea? I don't know. At the moment, I'm not bothered by the fact that people can potentially continue to use these very creations. For me, there is actually no copyright as long as the creation is not real. That almost ties in a bit with the debates that were going on at one point about NFTs with digital property. I personally don't believe in it. So it's going to be difficult to say that it's purely my property. But I think this question of copyright with AI is going to go on for quite a while. I don't know how it will be decided and I can't make a prediction.
You may have heard of Boris Eldagsen, he's a photographer from Berlin. He won the 2023 Sony World Photography Award, and he submitted a photo that he processed with AI. He had mentioned in the submission that there was something wrong with the photo. After winning, he declined the award and said that he thought it was important that we start talking about the topic: "We need an open discussion in the photography world about what we want to consider photography and what we don't."
Also, we recently spoke to architectural designer Tim Fu, who worked for Zaha Hadid Architects and specializes in working with artificial intelligence. He perhaps formulates the opposite: for him, photography is a technique for producing an image. In this respect, it is the same as AI: it too is nothing more than a technique for producing an image. This is perhaps a rather philosophical question, but what do you think of these two opposing ideas?
Olivier: It's interesting that he has a very AI-friendly perspective. To be honest, however, there is a big difference for me: the original photography as we used to know it was supposed to reflect reality. In my opinion, we are in a different realm when AI generates images. You can get extremely realistic photos, but they will never represent reality. For as long as there have been photos, there have always been photomontages. This has meant that reality has sometimes been distorted, but here we are really inventing something new.
I think there was a photo of the Pope in a Balenciaga down jacket. It was quite funny because it went viral and people didn't understand that it was an AI-generated photo. I think we're going to move in that direction anyway because the advances that have been made, particularly by Midjourney, are enormous. Even if you wanted to ban it, I don't think you could. The question of whether it's photography or not: to me, the real difference is that there's no human at the trigger.
When I do a prompt on Midjourney, I don't see myself as an author or artist. In the same way, someone who produces an image with Midjourney is not really an author or artist for me. He is the author, but he will not produce, the machine will produce. The AI is not just the tool, for me it is also the author. That's where it really differs from a camera, which is the tool. At first glance, the camera will not compare information, it will not think, it has no form of consciousness, not even an artificial one, it will simply execute. It's completely different with AI.
But didn't you say that the AI only carries out what you tell it to do?
Olivier: Yes, but AI also interprets it. AI doesn't just perform it, it depends on the context. If you ask ChatGPT for a very clear answer, of course it will give it to you, but it will also be able to interpret it. For me, that's the real difference with the camera. I think it's still early to answer that, because we will be much more dependent on future developments. On the other hand, I think the Berlin photographer is right when he says that we need to start talking about it very seriously. Because it raises a number of issues in this question that are also becoming increasingly important socially. We will be forced to talk about it anyway, that's for sure.
With regard to AI, I have often heard that it is no coincidence that many artists or photographers achieve good results with AI. According to the motto "shit-in-shit-out". You are also an example of how you can achieve better results if you have already familiarised yourself with visual possibilities. Which creation in your work do you like best?
Olivier: To be honest, I'm not very good at naming my favourite things.
If you see an evolution in your furniture, your objects, since you started last year, how would you describe it?
Olivier: I find it easier to produce a certain style now than when I started. I don't know to what extent Midjourney adapts when it is fed by a user. To be honest, this is beyond my technical knowledge. In any case, it's interesting because my designs seem to become more consistent. Maybe it's also because I'm a bit more proficient with the tool. It goes less in all directions than at the beginning. There was a moment of curiosity at the beginning. You do a thousand prompts in an hour because you want to try things out. Now it's more like, "Okay, I have an idea, I want to push it, I'll see how this thing reacts, and then I'll keep gaining inspiration." As a result, I don't know if I have a favourite object.
As far as the "shit-in-shit-out" is concerned, it's like everything else. I would say that you shouldn't expect miracles. Just like with a camera. But I also believe that it can be learnt in the same way, albeit with less effort. Because with AI, there is still a side where I think the machine interprets itself. It's true that there are a lot of shit-ins about AI, it's still quite scary. At the same time, it's interesting because it ultimately democratises the whole thing even more. After all, it's affordable for a lot of people. Even if there are now increasing restrictions that are really problematic, like subscription prices and especially data collection. For me, this is a major problem with AI, that it is ultimately only controlled by a few players. That's quite scary for the future, because AI will control more and more parts of our society. I'm not sure what solutions there are and again, it's a case of wait and see.
Do you have a message for the design industry and how do you see the future of AI-generated design?
Olivier: I think you have to persevere; you have to continue to exist and continue to produce physical objects. With a label and a publishing house, I strongly insist on materiality, on concrete objects. Therefore I know how difficult it is, how time-consuming, how resource-intensive and how unrewarding in the world we live in today. But if anyone listens to me, I would tell them to get on with it.
It's really about AI not replacing creative work, but perhaps becoming a part of it and providing a few ideas and inspiration from time to time. The real problem would be if it replaced human interaction, which in turn creates inspiration, stimulation and so on. I myself sometimes feel like I'm betraying my own moral concepts when I use AI. But I still think it's possible to do both, and to have one eye on the left and one eye on the right, to try to nurture one and the other mutually.
