Lessons from Paris: from Chatbots to AI Art
A Chat with Marion Carré

Explorations of AI Art — Episode 13

Beth Jochim, Director and Co-Founder at Cueva Gallery. Twitter: @_bblurred

 

“I think that algorithms and in particular GANs can play a role of creative ‘ouvroir’. “ — Marion Carré

 

 

[Fig.1] ‘Portrait 14’, from the series ‘Passé (dé)généré’. The artwork is created with neural networks (GANs) trained on old maps of Paris, courtesy of Marion Carré.

[Fig.1] ‘Portrait 14’, from the series Passé (dé)généré’The artwork is created with neural networks (GANs) trained on old maps of Paris, courtesy of Marion Carré.

 

Marion Carré is CEO at Ask Mona, a startup based in Paris that she co-founded with Valentin Schmite in 2017 and that aims to make culture more accessible through chatbots. The company leverages the power of Artificial Intelligence (AI) creating interactive experiences and providing innovative and impactful solutions for culture professionals and young audiences.

Her interest in AI does not stop however here. In fact, at Sciences Po she is engaged as a teacher in a new course called When AI challenges culture: History, Faces and Mutation of the Twenty-first Century Brush (for the Master in Public Policy). This course analyzes the impact that technology and art have on each other and it is designed for the Master in Public Policy.

As a public speaker, her commitment is to raise interest about the topic of AI as a driver of cultural innovation. Her approach proves to be analytical and exploratory, going to dissect doubts and opportunities that influence both the artistic creation and the circulation of knowledge.

In love with culture and innovation, and having degrees in law, communication and history, for Marion it was probably just matter of time to explore AI also from an artistic point of view. Indeed, last year she taught herself to code and developed an artistic practice that allowed her to reconnect with older passions such as historical archives and maps. It is in following this direction, and confronting a tradition of the past, that she discovered how a set of rules can later bring novelty and the unexpected, allowing a deeper investigation of what is real and what is only a cultural product of our perception and subjectivity.

This interview ranges from the analysis of the relationship and mutual influence between culture and technology, to questioning epistemological aspects, passing through the concept of creativity which is reviewed under the lenses of Artificial Intelligence.

Marion’s work creates a bridge between past and future, highlighting the difficulties of accessing AI art, but also confirming the infinite possibilities that AI can bring to the artistic practice. Constrains trigger new ideas and inspiration, while the use of technology teaches us to see the world with different eyes.

In a Paris that offers a continuity between its cultural heritage and a future to discover and design, entrepreneurial innovation and artistic experimentation move on and find their place between the history of the past and that of the future.

[Fig.2] ‘Portrait 2’, from the series ‘Passé (dé)généré’. The artwork is created with neural networks (GANs) trained on old maps of Paris, courtesy of Marion Carré.

[Fig.2] ‘Portrait 2’, from the series ‘Passé (dé)généré’. The artwork is created with neural networks (GANs) trained on old maps of Paris, courtesy of Marion Carré.

 

Beth Jochim: You are an entrepreneur, speaker, author, teacher and artist. This puts you in the position of exploring the relationships between Artificial Intelligence and art from different angles. What can you tell us about it? Is there a point of convergence among all these approaches or do you feel that every angle brings something different to the discussion?

Marion Carré: Each of these approaches positions me in different places in and outside the art ecosystem. Through my entrepreneurial approach, I am very much in touch with the professionals of cultural institutions and their audiences. In this place I think of artificial intelligence more as a facilitating agent in the service of this public-cultural institutions relationship.

My activities as a speaker, author and teacher seem to me to have in common the opportunity to share and shape a constant work of research and exploration around art and artificial intelligence while being challenged in these reflections by pupils, an audience or readers who, by their often new look on the subject, sometimes push me to consider certain aspects differently.

Finally, my artistic activity is for me a very important third pillar in the sense that it allows me to extend this exploration and sharing, but this time through the senses and not only through the words. It is also a work that allows me to develop my artistic practice and reconnect with older passions like that around the historical archives which happen to be for the moment the main raw material of my creations with machine learning.

Beth: You are a self-taught AI artist. How was the experience of learning to code?

Marion: Difficult! Besides, I must always continue to train myself each time I want to start a new artistic project. But it’s very motivating to learn that way.

I would like to take advantage of this experience of a novice who learned to code from scratch in order to facilitate access to creation with artificial intelligence for other artists who do not have initial technical skills.

I think that one of the specificity of this artistic creation tool is that it requires both technical skills and investment in expensive equipment, which can constitute a real barrier to entry. Certainly there are interesting online solutions that allow you to start experimenting more easily, but to be able to go further it is necessary to code. I believe that it is essential to mobilize to facilitate access to this means of production for artists if we want to democratize AI art.

 

[Fig.3] ‘Portrait 4’, from the series ‘Passé (dé)généré’. The artwork is created with neural networks (GANs) trained on old maps of Paris, courtesy of Marion Carré.

[Fig.3] ‘Portrait 4’, from the series ‘Passé (dé)généré’. The artwork is created with neural networks (GANs) trained on old maps of Paris, courtesy of Marion Carré.

 

Beth: How did you find the process of working with algorithms in relation to creativity and expressiveness?

Marion: I think that algorithms and in particular GANs can play a role of creative ‘ouvroir’ (like a potential literature opener). We can arrive with a preconceived idea of the result, but it will never be carried out exactly as expected.

The unpredictability of these algorithms makes it possible to change posture in the creative approach. There is then a work of going back and forth between the artist and the machine where the latter can seek to orient the final result to a certain extent while being surprised by the machine’s proposals which can sometimes change its initial approach.

I like the idea that it is this corpus of rules, the algorithms, that ultimately pave the way for the unexpected. This is part of an older artistic tradition of creating rules to provoke the unexpected in the manner of Raymond Queneau (co-founder of the literary group Oulipo) and his work of potential literature opener.

Beth: In your work and exploration with and through AI, you investigate “the relationships between subjectivity, perception and representation. Can you explain more about this?

Marion: As I told you earlier, I really like working with historical archives and in particular with archive maps. These traces or absence of traces in certain cases are for me a great starting point for exploring and seeking to understand the subjectivity of our relationship to the world and its impact on our perceptions and the representations that we make of it.

I find that there is a real point of convergence between archives and artificial intelligence. Despite the materiality of one and the immateriality of the other, they both encrypt or record visions of the world of one or more actors, although this may affect the rest of society. Archives then have an impact on our readings of the past and artificial intelligence on our readings of the present or the future.

 

[Fig.4] ‘Orange, Peach, Illustration’, from the series ‘Divergences de vues’. The artwork is created using old maps of Paris, then showed to a pre-trained machine learning model of AI Vision, Google’s artificial intelligence that detects objects in images. The artwork represents the multiple points of view and the possibility to see things with new eyes. Courtesy of Marion Carré.

[Fig.4] ‘Orange, Peach, Illustration’, from the series ‘Divergences de vues’. The artwork is created using old maps of Paris, then showed to a pre-trained machine learning model of AI Vision, Google’s artificial intelligence that detects objects in images. The artwork represents the multiple points of view and the possibility to see things with new eyes. Courtesy of Marion Carré.

 

A question that has always fascinated me is the fact that there may ultimately be as many truths or realities as the number of actors present, and this because of the subjectivity and the differences of perception. However, only the version of the actors who were able to record and share it remains so that it reaches us. Yet we can have the impression of a universal truth of a reality that has come down to us. Especially when it comes from a tool as scientific as the map. This creates power relationships that have an impact over time. And it’s a bit the same with artificial intelligence. If we often postulate a rationality of the algorithm, it nevertheless remains dependent on the biases registered by its authors. This is not in itself a bad thing, but what interests me in my artistic work is to make it visible to open the discussion.

Beth: Ask Mona is a startup that you co-founded with Valentin Schmite that uses chatbots to make culture more accessible, gathering people and institutions together. In this case, AI is creating a bridge between an audience and culture. Considering that AI art is relatively new, what are the challenges to access it? Is it more about understanding it or experimenting with it? What could be done to push the democratization of AI art?

Marion: I believe that the answer lies in a subtle mixture between understanding how artificial intelligence is involved and at the same time the ability to go completely beyond this technical aspect to feel what the artwork offers us to experience.

I think it is important to remember that these works do not only have a technical component and this can also remove a barrier to entry for audiences who may not have the technical reading keys.

To push the democratization of AI art, I think, as I said earlier, that it is first important to give as many artists as possible the means to seize it. Then I think that cultural institutions and galleries play an essential role in opening their doors to these artists to allow them to present their creations to people while inviting to a reading of these works which is not only technical.

Beth: As a public speaker, you focus on what technology (and AI) brings to culture. What are your findings so far? What is the added value of technology in promoting culture? Does it change the perception of culture itself, or just the way we access and enjoy it?

Today technology is omnipresent in the lives of many people. I think it is interesting to build on the preexisting uses for it to make it a vector of culture. There is therefore a clear challenge and opportunities in using technology as a lever for access to culture.

I think that there are also more profound changes taking place on the question of the relationship and the perception of culture. And artificial intelligence has a fundamental role in this regard. All these creations made with artificial intelligence again raise questions that echo previous reflections on the role of the artist, on the place of the artistic approach, on the status of the artwork …

However, the question is not simply what new technologies bring to culture, but what the cultural field brings to technological developments. Many artists today use their works to question the relationship with these new technologies, to question their deep meaning or even to transform them. This is the case, for example, of Anna Ridler for whom artistic expression makes it possible to point the finger at the volatility of crypto-currencies price; or Kate Crawford and Trevor Paglen, who show the biases of facial recognition, linked to the subjectivity of the constitution of ImageNet datasets.

 

[Fig. 5]‘Aqua, Turquoise, Logo British Airways’, from the series ‘Divergences de vues’. Courtesy of Marion Carré.

[Fig. 5]‘Aqua, Turquoise, Logo British Airways’, from the series ‘Divergences de vues’. Courtesy of Marion Carré.

 

Beth: Paris has traditionally always been a cradle for artists, cultural movements and avant-garde. Even now the French capital has not abandoned its role as a cultural and artistic nurturer and promoter, embracing AI in many different ways. As an artist and entrepreneur do you feel supported? What can Paris teach other cities that want to embrace Artificial Intelligence in the cultural and artistic fields?

Marion: Paris has long been a city of artistic avant-garde where theorists and practitioners meet, where new ideas are born. Whether you think of Delacroix, Beaudelaire, Breton or Toulouse Lautrec, Paris has inspired poets, creators, artists. Today, we build on this heritage. Our museums flourish, people from all over the world flock to enter the Louvre, the Pompidou Center, the Orsay Museum.

But far from being a simple heritage, relayed to the rank of memory, Paris also knows how to take the lead. When we talk about AI Art, there have already been many exhibitions (for a still recent subject). At the Grand Palais, the “Artist and Robot” exhibition showed the evolution of a practice dating back to the beginning of the 20th century. Today, the Center Pompidou dedicates an exhibition to the contemporary scene.

Beyond exhibitions, Paris is a place of creation that welcomes artists and gives them the means to succeed. Consider the scholarships, residences, schools, courses that are taking place. Paris brings us continuity between the past of its heritage and the future of its creators. It shows us the way, allows us to build on its history and lets us build its future.

Beth: Valentin Schmite, co-founder at Ask Mona, and you have written a book entitled “Art and Artificial Intelligence. Artist in the making?” What is the main idea behind it and are you organizing promotional events?

Marion: I think there are three main ideas that we explore in this book. We start by drawing parallels between art and artificial intelligence, starting from paradoxes to arrive at common points. Then we explore the relationships between artistic creation and artificial intelligence by talking about what one reveals on contact with the other. Then we evoke the possibility for art to be a counterweight to artificial intelligence by raising various ethical questions on the subject.

To date, we have planned fifteen promotional events around the book between France and Canada.

 

[Fig. 6] ‘Portrait 11’, from the series ‘Passé (dé)généré’. The artwork is created with neural networks (GANs) trained on old maps of Paris, courtesy of Marion Carré.

[Fig. 6] ‘Portrait 11’, from the series ‘Passé (dé)généré’. The artwork is created with neural networks (GANs) trained on old maps of Paris, courtesy of Marion Carré.

 

To follow Marion Carré: Twitter:@Ma_rionC / Instagram:@_marioncarre

Thank you, Marion, for this interesting interview! ∎

Resources and References

https://www.marioncarre.com/

http://askmona.fr/en/


About the author: Beth Jochim is the Creative AI Lead at Libre AI, and Director and Co-Founder at Cueva Gallery. She works at the intersection of technology and arts. She is actively involved in different activities that aim to democratize the field of Artificial Intelligence and Machine Learning, bringing the benefits of AI/ML to a larger audience. Connect whit Beth in LinkedIn or Twitter.


Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.