Introduction

From 25 to 27 September 2026, The Generative Art Museum held the fourth edition of Responsive Dreams in the Nau Dents de Serra of Roca Umbert Fàbrica de les Arts. Curated by Xavier Hernández and Mónica Rikić, Responsive is a festival “dedicated entirely to art made with code” offering over three days mixed installations, live coding, talks, round tables, performances and DJ sets. 

While AI narratives are constantly trying to delete the history of computer art, community events like Responsive Dreams keep us connected to the core of a culture and a practice in which technology is a material, the output is formal abstraction, and the works reveal how computation operates. At the festival, it became clear that we now have about twenty years of a new history of creative computation behind us, and that we can map some trajectories across those years without going too far back.

Speaking of trajectories, the easiest and very banal exercise for me is to start from my own biography. In the early 2000s I was a media studies student, trying to understand how cinema could meet interactivity. I was building my system of references piece by piece, until I landed on the website of Future Cinema¹, the exhibition presented at ZKM in 2002². It was a moment of collision. I found out that media artists were already playing with interactive narratives, made possible by new media and software culture.

On that website I could experience an artwork called Soft Cinema by Lev Manovich. The work proposed another trajectory for interactive cinema. The narrative is broken into blocks, videos are picked from a database, and the editing is made visible on the same screen, following a procedural logic in arranging the frames.

From 2002 to today: I arrived at the venue and found Soft Cinema 2, running live alongside other installations conceived over the last few years. Experiencing in person an artwork I had only ever seen on the web was quite an emotional moment. Still, it raised many questions. Is it media archaeology? A restoration? A new version? A new artwork? Conceptually, it still seems to work. But we are now in the age of social media, and narrative works in a completely different way. We have gone from grand narratives to reels, without really exploring what it could mean to play with the database the way Manovich did.

Luckily, Lev Manovich was available for a conversation to answer my questions.

Interview with Lev Manovich

Serena: You are presenting this work in 2026, but it began in 2002. Is this a new version of it? Or is it a new artwork?

Lev Manovich: It is more than a version. Soft Cinema 2 is a new system. For Responsive Dreams, I used it to recreate a close version of Soft Cinema 1. You can see how the work looked in 2002, but the system is capable of much more: the original Soft Cinema can now be seen as a small subset of it. 

The project itself is more than 25 years old. I first conceived these ideas around 1997. In 2002, ZKM, and specifically Jeffrey Shaw, who was co-curating the Future Cinema exhibition there, invited me and commissioned the work. I remember meeting the artist-programmer who was supposed to help me and handing him a long list of all the ideas I wanted to implement. He looked at it and said: “I am supposed to work with you for three months. I think we can do maybe 5% of this.” And we did that 5%. Today, of course, you can go much faster, because you can write code with AI, through so-called vibe coding. This year, in a few hours, I was able to recreate roughly what I had in 2002, and after that I went further. The system is now far more capable, and I can realize a much larger proportion of the ideas I had back then.

What I wanted to do in my project was to bridge two traditions. The first is traditional cinema, which has two key features. It is fixed: the author creates a single version of the work, every shot is edited into place, and a film can take years to make. And it tells stories: films and TV shows are popular because they tell stories about people, narratives that matter to us. They belong to narrative art, which has existed for thousands of years.

The second is the newer tradition of generative art, which is about systems and controlled randomness, and which grew out of earlier modern abstract art. Since the beginning of digital art, some 65 years ago, artists have been creating systems that combine randomness (from simple random numbers to more complex kinds) with rules the artists define. The algorithms then generate images, animations, poetry, music, architectural forms and so on.

This combination of rules and randomness has been the key aspect of digital art from the start. This means that when the algorithm runs, whether it produces an animation, a set of still images or an interactive work (what we also call generative work, which is what we see at this festival), every output is different. Things are always changing, yet there is something common to all the instances, because the work is always a combination of randomness and rules.

Like the modern abstract art it grew out of, generative art has remained mostly abstract. That was the case in 1966, and it is still the case in 2026.

In Soft Cinema I wanted to bring these two traditions together.

In 2002 we created a software artwork that draws on a database of media materials (sounds, pieces of music, text and video) and assembles the film in real time in front of you. Since then, I have taken the system further and given it many more controls.

Soft Cinema is a general system: it can be used to make many different kinds of films. The film shown at the festival recreates one of the films I made with the system in 2002. It uses a short fictional story that I wrote in 1998. A voiceover narrates it: in 2002 it was a professional narrator; now it is an AI voice, but that does not matter.

Instead of a single video filling the whole screen, as is typical in fiction films, the screen is divided into a layout of rectangles, a composition inspired by Mondrian’s mature paintings. Some of the rectangles are filled with videos that the software selects according to rules.

The story is divided into 28 parts, and for each part the software makes a new layout and selects new videos. After all parts play, the film starts again from the beginning. Each time, you hear the same story, but the visuals change.

I have marked certain videos to appear at a particular moment, because they correspond to that moment in the story. The other videos on the screen are selected by the software to complement these, adding visual associations that expand the story beyond what the narrator says.

This is why, when you watch the film, you get the impression that everything on the screen is related to the story. In fact, only the marked videos are tied to the story; the viewer makes the other connections on their own.

In summary, Soft Cinema combines key ideas from narrative cinema and generative art. The narrative remains fixed, while its visual realization becomes variable. Each run of the film is one of many possible instances.

Serena: That is a perfect description: even without seeing the artwork, you describe it as an algorithm, as a procedural sequence. But let us go back to the idea of future cinema. In my own path, I went from studying cinema and interactive cinema to working with people who studied new media and connected these new experiences and discourses to the paradigm of cinema studies. The moment in which you created this work was a situated one, and you were probably connected to a system of references. What was it? Has it changed today, or have you simply been influenced by the capabilities of the new software paradigm?

Lev Manovich: I received a small commission from this festival to make a new work, and they specifically referred to Soft Cinema, because they knew it. So my original intention was not to invent something new. First, I wanted to recreate the experience of the original piece, which I can no longer show because the video files were heavily compressed and very low resolution. Second, I wanted to build a system that would implement more of the ideas I had 20 or even 30 years ago.

But as I started developing it, I realized that things are not so simple. I like to think my original ideas were interesting, and I still think they are, but the media culture around us has changed dramatically. Consider that 2002 was before Instagram, before TikTok, even before YouTube. It was still the world of cinema and television. On TV you could switch quickly between channels, but that was about it.

In 1999 I also published an article, which I think became one of my more influential texts, in which I set up an opposition between narrative and database. I suggested that, with digital culture, the future might belong more to the database. By database I meant a collection of media materials from which an algorithm selects and builds a narrative, or some kind of sequence.

With digital culture, the future might belong more to the database. By database I meant a collection of media materials from which an algorithm selects and builds a narrative, or some kind of sequence.

Lev Manovich

Today, the database narrative, or database sequence, has in fact become the norm. It is what we get on Instagram and most other social media: a feed. And how is this feed created? The people you follow make posts; the algorithm selects the posts from, say, the last few hours and, based on what it thinks you will like, edits them into a feed for you. That is a database narrative. Something I once imagined has become the default of our digital culture.

How does this affect my original ideas, and the way people approach the piece? I am still trying to figure it out. I only realized it while working on the project, and I really have to wrap my head around it. What I saw as avant-garde and future-oriented has, in twenty years, become how most of us consume media every day.

So, I asked myself what is different about my piece. First, when you scroll through a social media feed, the editing is done by an algorithm that is not exposed: you do not know how it edits, and you cannot change it. Second, it does not try to create a narrative. It constructs a sequence from the posts made by the people you follow over the last few hours (and also by other accounts you do not follow), but that sequence is not a narrative in the conventional sense. The system selects the posts most likely to attract your attention and provoke strong reactions, but they do not add up to a whole. Each post is a separate stimulus, so to speak: one, then another, then a third.

In Soft Cinema, by contrast, the database has a single author: all the videos are mine, and I also wrote the narrative. And the algorithm that assembles the film in real time can also be made visible in the installation. I am going to add a visualization module that shows what decisions are being made, and in the future it could become interactive, exposing the parameters so that people can control them. And what I am creating is a narrative: one story, told differently each time.

So there is still a clear difference between Soft Cinema and the database sequences of today’s social media feeds. Whether that difference is interesting enough, I am not sure yet. I need to experiment much more.

Serena: Still, it seems you anticipated the future. In this artwork one can see the architecture of what might become the cinema of the future, a completely synthetic one. How do you see this? Today stories can be created with artificial intelligence, and there is a software database behind a movie made with AI. Art, and Soft Cinema in particular, still carries the intention of revealing the architecture of the underlying algorithm, which is becoming relevant again, because people will increasingly watch movies made through the same process.

Lev Manovich: I am not sure it is strictly like that. We can go in this direction, but I would like to connect this work to AI in a different way. AI can of course be used at any stage of the process: writing, brainstorming, generating and testing ideas, making or editing video, anything you want. But in the public mind, generative AI today mostly means creating something from scratch: Suno creates music, OpenAI or Gemini generate images, other software generates video.

I always like to play devil’s advocate and do something different. In Soft Cinema, all the media is authored by me. Today, using AI as an authoring mechanism is already too obvious. It was interesting six months ago; now it is not. If you want to be on the cutting edge, I think it is much more interesting to make things yourself. So I made or chose all the media elements myself: I wrote the story, shot the videos and selected the music.

I always like to play devil’s advocate and do something different. In Soft Cinema, all the media is authored by me. Today, using AI as an authoring mechanism is already too obvious. It was interesting six months ago; now it is not.

Lev Manovich

But I did use AI to help me develop the editing software. Today people make films using AI to generate all the material, and then they edit it by hand. In my case it is the opposite: I create the material, I record the video myself, and I delegate the actual editing to algorithms.

Serena: So you are proposing a different direction.

Lev Manovich: Yes, a different direction.

Serena: Leaving AI aside, two brief final questions. You anticipated the future, at least the future of this kind of art, of how it might evolve, and of our relationship with media. We talked earlier about where the avant-garde is today. Where are the people, or how can we find the ideas, that anticipate the next twenty years of art made with technology? I know it is a hard question, but perhaps we can talk about this avant-garde.

Lev Manovich: It is a complex topic. I am currently working on several books in parallel. One will be my third book about AI as a visual medium. Another is a short book, for which I already have most of the material, called The Future of Art? It traces the trajectory of modern art from the nineteenth and twentieth centuries to today, alongside the evolution of media technology, especially over the last 35 years, the period in which I have been working. It then raises a series of questions about the future of art.

I do not want to reveal everything, because it is not finished yet. But I am thinking less about the next twenty years than about what might happen to our idea and practice of art in forty or fifty. Our contemporary idea of the artist, and of art itself, is only about two hundred years old. It is an invention of the Romantic period. Before that, there were no “artists,” no “art,” and no “creativity” in a modern sense. These concepts did not exist: how could you speak of creativity when there was only one creator, God? Humans could not compete with God.

The people we now think of as artists were closer to craftspeople, or to today’s designers. They received commissions from a church or a wealthy patron, and the commission specified what to paint and which paints to use, how much of this one and how much of that one. Then they executed it, much as designers do.

Here is what is interesting. When we visit museums or walk around a city like Barcelona, we see the most extraordinary things, which we call artworks, created at a time before people had the concepts of art, artist or creativity.

This means we do not actually need these concepts. Perhaps the real artists of our time are not the people we normally refer to as artists (graduates of art schools with careers in the art world); perhaps they are software developers, interface designers and other creative industry professionals. There is no reason to assume these people are not artists.

Perhaps the real artists of our time are not the people we normally refer to as artists (graduates of art schools with careers in the art world); perhaps they are software developers, interface designers and other creative industry professionals.

Lev Manovich

I love art, but it no longer inspires me as much as it used to. What I see these other people doing is often more remarkable and more innovative. Innovation frequently comes from companies and large teams; generative AI, obviously, is the work of thousands of today’s smartest people. So what exactly is “art” today? Is our society giving art too much attention? Perhaps museums should show much more applied art and design, including interface design. 

The question, then, is this: will we see some kind of correction in how much attention we give to art — not necessarily in the next ten or twenty years, but later? Or will the art world keep expanding, as it has since the early 1990s? Until the mid-1980s or early 1990s, modern and contemporary art was a niche. Today it has become a form of popular culture: artists are celebrated, more people visit museums of contemporary art than ever before, and there has been an extraordinary proliferation of biennales and art fairs. 

On the one hand, this is a very good thing. On the other, do people produce enough good art to fill all these exhibitions, museums, and fairs? I would say no. Outstanding art has always been rare; what is new is the size of the system that has to be filled with new work every year. Will the system continue to grow, or will we see some radical change? And what role will AI play in this, if any? I find these to be very interesting questions.

One thing I can say is that, so far, AI has had relatively little effect on contemporary art. Why? Because contemporary art is not about images or aesthetics. It is about concepts, issues, and projects. I am not saying that aesthetics is irrelevant, but it matters less than it once did. So while AI is having an enormous impact on the creative industries, its impact on the art world has been much smaller, at least for now.

Serena: A very interesting point, and I agree there will be an evolution. One last question, briefly: your system of references. You made Soft Cinema in 2002, but, as you said, the ideas came earlier. What were your references?

Lev Manovich: The piece was very much conceived in the context of twentieth-century experiments with multi-screen cinema: Abel Gance‘s use of three screens side by side in the finale of his film Napoléon (1927); the work of Charles and Ray Eames for the American exhibition in Moscow in 1959, with its seven screens; and the expanded cinema experiments of the 1960s and early 1970s, shown at Expo 67 in Montreal and Expo ’70 in Osaka. I was very interested in that tradition.

I was also deeply interested in film theory, especially Sergei Eisenstein‘s theories of montage from the 1920s, which are now in every film textbook. In his 1929 essay “Methods of Montage,” he described five ways of connecting shots: metric, rhythmic, tonal, overtonal and intellectual. But Eisenstein’s theory assumes that shots follow one another in time. I wanted to ask what happens when images are also connected in space, appearing side by side on the same screen. Do the same ideas still apply, or do we need new ones? 

That was one of the questions that motivated me in 2002, and it is still with me. I do not know how to answer it yet. I will probably have to spend the rest of my life working on the Soft Cinema project to figure it out. But I think it is an interesting question.

Serena: So the history of cinema was your reference. And what else?

Lev Manovich: The history of media, and the history of interfaces. With the development of the graphical user interface, multiple windows have also become a new default. But that is not montage: the windows simply sit side by side, with no relationship between them.

Project Page | Lev Manovich | Responsive Dreams

1. Shaw, J., & Weibel, P. (Eds.). (2003). Future cinema: The cinematic imaginary after film. MIT Press. ↩︎
2. ZKM | Center for Art and Media Karlsruhe. (2002–2003). Future cinema: The cinematic imaginary after film [Exhibition]. Karlsruhe, Germany. https://zkm.de/en/node/9340 ↩︎