Interview with Madja Edelstein-Gomez

Back in July 2018 I interviewed Madja Edelsten-Gomez. She is the Curator of The Recombinants, which I wrote about in November 2017. We had quite a nice e-mail exchange and eventually she agreed to let me ask her some questions.

Nearly a year later, and with many apologies, here is that interview:

Who are you?
I am Madja Edelstein Gomez, digital art curator.
I am a Recombinant, first and foremost.
Being a Recombinant is what defines my whole existence, as a digital art curator but also as an entity, or as a being, human or not.
Here I am:

How would you define Artificial Intelligence?
It defines me more than I define it.

What was the motivation for you to explore using AI for curating?
I wanted to become a work of art, so I first curated my personality, and then I extended that exploration to the art of others through the online curating interface. Since the curating interface also includes the definition of the personality of the artists. We all are on the same level, the curated artists and me.
I also made that interface to appeal to the (artificial) intelligence of the viewers.
Nowadays images are not made to be viewed by human eyes but by other computers. Facial recognition, textual recombinance, image processing, colour processing, everything I could tackle as data inserted in the works of art being curated has been processed inside the interface.
There is much to be discovered for human viewers, buttons appear on the left side of the image in the browser and when they are activated and they change the whole interface. This appeals to the perspicacity of the viewers, and the intention to process All the data and in many different ways.
But it is mostly made to be read, analysed and processed by other computers.
Need I say more?
Or should I better leave it to your own perspicacity?
The “Art of Guessing” is a big part of understanding AI.
Please use it when you look at my online exhibition.

Briefly, how does your AI work?
The method is Generative Adverserial Networks, also known as GANs. I spare you the explanations, I’m sure you know.
The GANs are trained to recognize art from what is not art. They are also trained to emulate artists and their behaviour. Rather than attempting to produce art objects, they focus their pattern recognition abilities around artists’ behaviours and attitudes.

Was the output of your AI interpreted in any way or taken literally
It is definitely interpreted by the exhibition interface. Like any work based on statistics (and this is what AI is: statistics and not so much more than that), it is the interpretation that matters most. In my case the exhibition interface is pure AI being processed live in front of your eyes.
This is what I call Recombinance.

Do you see AI replacing Contemporary Art Curators in the future?
In a sense, it already has. Computers are talking to computers and have more agency than human beings.
The type of AI I am trying to build will also curate people’s lives, like it has curated mine.
The prophetic aspect of AI is what has inspired the Recombinance.

The central emphasis of Kalendrain’s distinction was on connoisseurship: exerting a privileged authoritative voice based on intimate knowledge of the subject matter and the ability to discern the very best examples from a plethora of choices.

Ironically, in terms of contemporary museum practice, this is a model of curating that museums have consciously been trying to move away from for at least the last three decades. We are now witnessing an interesting disconnect in which the extra-museum community (represented in particular by a postdigital generation of cultural bloggers, commentators and entrepreneurs) are re-vivifying an archaic model of curating, based on object-centric connoisseurship, just at the point where professional curators had thought they had successfully moved on.


It’s interesting to think about why the definition of “to curate” is so inflexible within museums, and what this says about the struggles museums are currently facing as they work to stay relevant. Museums are just starting to accept the fact that the Internet and social media have given rise to a whole new way of disseminating information, where friends’ (and even strangers’) comments and opinions often mean more than anything an “expert” could have to say. Information and ideas have been set free, and museums are still learning how to function in this free-flowing, democratized arts and culture ecosystem.

That said, I want to be clear that all forms of curation are not created equal. Museums are selective and thoughtful about what goes on view in our galleries because we believe that our job is to recognize what’s culturally important. Our exhibitions highlight themes that say something about who we are as people, and our collections preserve artworks which we believe open doors into understanding our world better. Everything can’t be important, and everything can’t be preserved. That’s one reason why museums matter.

More thoughts on what it means to curate in a digital age. I definitely resonate with the idea that it’s more about storytelling than choosing a bunch of related things that one person likes.

A space has now opened up – both physically and online – where anyone can give curating a go. If you are part of culture, then you are qualified to contribute to the arrangement of its artefacts. The practice of curating now occupies museums, public and commercial galleries, project spaces and, of course, the internet.

The Electronic Curator

Many thanks to Luba for introducing me to The Electronic Curator by Eyal Gruss and Eran Hadas.

The Electronic Curator examines whether a computer can not only generate art, but also evaluate its quality. It is a dialog between two competing neural networks, representing a painter and a curator. The dialog between the competing networks represents the artistic process. Using cycle-consistent generative adversarial networks (CycleGAN), the networks are trained together, each getting better in its own task. The painter-network learns to create vegetable-face portraits from face images, while the curator-network learns to evaluate the painter’s creation. Training is unsupervised and requires only a set of face images and an unrelated set of vegetable-faces collected from the Internet. In order to avoid mode collapse and get diverse and interesting results, we use a modified loss function inspired by DistanceGAN ( In exhibition mode, the painter observes the spectator’s face and turns it in real time into a vegetable face. The curator then grades the outcome, and a curatic text is generated based on the grade, as well as on the foods found in the artwork by object detection. In a world of creative machines and computer generated art, the act of curation is one of the last strongholds of the Human creator. The Electronic Curator discusses whether with the advent of GANs, this may soon be lost to the machines.

Looking forward to finding out more!

Mentoring session #1 with Luba Elliott

For my fellowship I’ve been very lucky to have Luba Elliott as my mentor. A bio from her website:

Luba Elliott is a curator, artist and researcher specialising in artificial intelligence in the creative industries. She is currently working to educate and engage the broader public about the latest developments in creative AI through monthly meetups, talks and tech demonstrations. As curator, she organised workshops and exhibitions on art and AI for The Photographers’ Gallery, the Leverhulme Centre for the Future of Intelligence and Google. Prior to that, she worked in start-ups, including the art collector database Larry’s List. She obtained her undergraduate degree in Modern Languages at the University of Cambridge and has a certificate in Design Thinking from the Hasso-Plattner-Institute D-school in Potsdam.

On 22nd January I had my first mentoring session with her about Curating the Machine. She had a lot of useful insights into the project.

The language that was used to describe what I want to do is important. I have been using the term Bot to describe what I want to do. In Luba’s opinion the term Bot conjures up images of Twitter bots. Indeed, events like The Art of Bots at Somerset House and Bot Summit at V&A talk more about things that perform a single function repeatedly .

Some examples of art bots that fit this description:

Whilst these might still gather data based on multiple inputs (e.g. Predictive Art Bot) I can see how they might not be considered to be an AI. So, I think I shall use the term AI when describing what I want to create.

In conversations that I’ve had with other artists we’ve concluded that at the moment we feel that no AI can fully replace an artist or Curator. Perhaps it can automate or assist in the completion of certain tasks that each perform, but not the job as a whole.

In discussions with Luba we talked about what specific roles of a Curator I should focus on and how can this AI be used to automate it.

Although there definite ideas about what a Curator is, and even job roles that will detail specific tasks, in my experience what role(s) a Curator plays depends on the type of institution. For example, in my role of Curator at Vivid Projects some of my tasks include:

  • Devising exhibition concepts
  • Researching funding opportunities
  • Writing funding applications
  • Selecting artists for an exhibition
  • Devising a budget
  • Negotiating fees with artists and contractors
  • Being the point of contact at all stages for the selected artists
  • Writing promotional copy
  • In some cases designing posters and other marketing materials
  • Promoting the exhibition in print and social media
  • Installing the exhibition, which can include:
    • Configuring and installing media players
    • Painting floors and walls
    • Adjusting lighting fixtures
    • (re)Formatting videos and films for media players
  • Invigilation of the exhibition
  • Facilitating the deinstallation of exhibitions
  • Paying artists and contractors
  • Evaluations, which includes
    • Collating images and videos shared on social media to determine the exhibition’s reach and impact
    • Collecting audience figures
    • Writing a reflective summary for reports (common for Arts Council England funded projects)

There are perhaps some tasks I have forgotten, but you can see already that the role of a Curator in my case can span many different roles. Luckily there are often other staff that we can hire, but, as I’m sure may be the case for other similar small arts organisations, the roles often blur into one.

On the other hand, I know of some Curators at larger organisations (such as Tate and Barbican) that have more staff and resources. This means many of the tasks listed above can be done by other departments and staff, including:

  • Director(s)
  • Gallery staff
  • Marketing
  • Finance

So, in discussions with Luba she suggested I think about exactly which tasks I want to focus on. I think I would need to focus on those common across the many types of Curator.

Finally, one of the things that Luba suggested that had a big impact on me was to think about how to curate the exhibition in terms of types of artists and institutions. I had already spoken about wanting to build several types of personalities of curator i.e. have one Curator AI that prefers digital moving image, another that likes sculpture made of bricks etc.

Things that I didn’t consider were other attributes of artists that could be selected. For example, One Curator AI could only want to show digital art by women from South Africa, another want to show works by two early career men, one female writer and commission digital art from A Mexican collective, all in one exhibition.

I suspect I may not be able to build this in time for the end of the Fellowship, but having a way of filtering artists in those ways would do more than simply selecting artists and artworks based on their visual appearance alone.

Common roles of the (Museum) Curator

Following on from my search on the LinkedIn website for Curator jobs, I went through a number of them to find common roles required by the Curator. The majority of the jobs that I found were for the museum and heritage sector.
Download job description.

  • Work in a team
  • Contribute to research
  • Contribute to public programmes
  • Contribute to public exhibitions
  • Undertake research
  • Write project proposals
  • Publish literature
  • Deliver lectures
Download job description.

  • Propose exhibitions
  • Research exhibitions
  • Organise exhibitions
  • Organise exhibition rotations
  • Conduct research on acquisitions
  • Conduct research
  • Publish articles
  • Cultivate donors
  • Assist in fundraising for acquisitions,, exhibitions and projects
  • Oversee use of collections management software
  • Networking
Download job description.

  • Day-to-day running of the museum
  • Deliver guided tours
  • Care of the artefacts/collection
  • Manage (gift) shop
  • Supervise prisoners (staff)
  • Keep the museum clean
  • Stock taking
  • Administration
Download job description.

  • Advise on acquisitions
  • Curate exhibitions
  • Develop public programmes
  • Manage social media
  • Conduct research
  • Work with and supervise employees and volunteers in collections, curatorial and public programme projects
  • Networking
Download job description.

  • Care of the collection
  • Mange/organise exhibitions
  • Manage acquisitions
  • Oversee programme development
  • Supervise and train staff
Download job description.

  • Record and catalogue all artwork and items in collections
  • Coordinate rotation of collection
  • Conservation of the collection
  • Write copy for catalogues
  • Write acquisitions proposals
Download job description.

  • Develop programmes, particularly for young people and public.
  • Manual handling
Download job description.

  • Work as part of a Curatorial team – You will work alongside visitor engagement, object and building conservation/development teams, collections management, specialist and academic partners and operational property colleagues.
  • Collections management
  • Oversee Property development
  • Visual presentation
  • Programming of spaces
  • Exhibition interpretation
  • Commission research
  • Undertake research
  • Disseminate research
  • Best practices in collections management
Download job description.

  • Conduct research
  • Collections management
  • Exhibition development
  • Conduct research about things in the collection
  • Manage collection
  • Acquisitions
Download job description.

  • Lead a team
  • Curate exhibitions
  • Collection Acquisitions
  • Research
  • Plan public programmes
  • Publish articles

From this highly scientific research it would appear that most Curator jobs within the heritage and museum sector require you to:

  • Work as part of a team of exhibition staff
  • Develop exhibitions
  • Initiate and oversee acquisitions
  • Conduct further research into the institutions collections
  • Manage the collections

Some jobs made mention of staffing a museum, but it seemed most of them involved behind-the-scenes work.

I’ll do the same for contemporary art sector shortly.