by Louisa Clement
Reinforcing her ongoing artistic engagement with transformation and optimization of the body and with communication at large, Louisa Clement's latest project pioneers the creation of a clone of herself – here denominated Representative -. Through the advanced development of a mechanical, digital as well as sculptural self-portrait, this new work brings together concerns and theories which have thus far informed the conceptual core of the artist’s practice. In doing so, it voices existential questionings which eclipse abstract principles and firm beliefs; What lies beyond the binary of self and other? Is the recognition of a non-self crucial in the definition of concepts such as gender, desire, identity and human consciousness?
As the genre of portrait has been enduring substantial challenges since the invention of digital photography, and the very idea of 'self' has come under pressure to legitimize its nucleus while technology and social media enact valid extensions of our persona, Louisa Clement raises the bar of artistic and technical experimentation seeking to unravel the enigma enveloping the essence of human nature. At a point in time when the narrative of modernism is evaporating and the 'self' is under threat of gradually yet inexorably dissolving into data, Clement also lays bare the perilous scenario of AI-induced emotional recognition and addresses the ever-narrower gap between human experience and machine learning.
Louisa Clement with her Representatives | Courtesy Der Spiegel
THE PERFECT SELF-PORTRAIT
The Representative is the perfect self-portrait in the present, it is the artist herself immortalized in silicone. An artificial clone about as close to human nature as technological advancement can get today. Biological data have been generated using microphotography, filmic motion studies, an innovative personalized Artificial Intelligence and the latest 3D body scanning technology that captures microscopic details. In cooperation with a specialized company, Clement has converted the data into an ultra realistic life-size doll supported by a wired aluminum skeleton which enables movements. This skeleton is enclosed by a TPE body, which is a replica of the artist body. Skin tone, shape, features and details are reproduced to the highest degree of precision. The body has high internal temperature and it is sexually functional.
A personified Artificial Intelligence based on the personality of the artist herself further shifts the boundaries and maps out almost untapped territories of human/robot twofoldness. The artificial intelligence encodes an algorithm-based profile of the artist character which is then fed into the robot head creating a backlog of experiential cognition. Furthermore, the artificial intelligence socializes and further develops through conversations with the audience - in other words, it absorbs impressions from the outside. At the same time the Representative has constant access to the artist’s social media profiles and digital accounts in a seamless stream of online data exchange and comparison.
Louisa Clement, Double Bind | Exhibition view at Kunsthalle Gießen (2021)
Implemented in cooperation with Professor Dr. Vera Demberg of the University of Saarbrücken, the AI has been programmed developing a chatbot which processes various fields of performances. Starting from computing a question-answering system that uses keywords to find logical answers to any question and in turn generates corresponding counter-queries, different teams composed by students of Computational Linguistics under Dr. Vera Demberg have achieved remarkable results by inputting emotional reactions which are coupled with facial expressions assigned in correspondence to any statement proposed and by calibrating the choice of language based on Louisa Clement's vocabulary, modulation and syntax. Lastly, an advanced voice generator program has been added in order to replace the doll's voice with the artist's inflection and timbre.
"Where do you get to when you no longer know whether you are an avatar or not?".
- Louisa Clement
GENESIS OF A PROJECT
Through the realization of a sculptural AI-based self-portrait, Louisa Clement draws the sum of the foundations of her work to date. Since the Heads series (2014–2015), Louisa Clement has been concerned with the embodiment of the human that surrounds us. Working on the idea of a serial, motionless motif, Clement photographed mannequins of various fashion shop windows and backed them with monochrome background. Featureless heads look out at us, seeming to speak to us yet silently challenging us. Interlocutors hard to identify as their faces are indistinguishable. Disquieting and perturbing when grouped up in a seemingly anonymous enveloping crowd forcing the viewer to confront with the oppressive feeling of homologation.
In the thematic complex Avatar (2016), Clement explores the implications of re-presenting oneself under a digital sign. Is the ‘self’ fed by LinkedIn profile, Instagram images and stories and Tinder photos to name but a few, or does it increasingly nourish itself through these parallel and existing self-portraits? How deep is the influence of technology and social platforms on how we really feel? Are we conscious players in the virtual performative
arena or rather victims of relentless dynamics? .
The shortcomings of human relationships as communication fluidly reshapes upon patterns of eroding attention span and over-sharing were further examined with the VR work Aporias (2019) which was developed in collaboration with Acute Art London. Three black puppets sit at a table, only visible through VR glasses. They speak out and a table talk may develop, but the three puppets are rather self-focused and unaffected by visitors. Equipped with peculiar artificial intelligences and deliberately programmed to be moody, their algorithm-based world view is not compatible with human, sensory-based perception. They talk to each other incessantly, communication is underway, yet real understanding does not occur.
The conceptual framework of Louisa Clement’s Representative was first investigated by the series Moulds (2019) and Body (2020). Defining an empty space, a void that reiterates the existential struggle of our condition, here the ancient theme of the incarnation of human desire is knitted together with the commercial. The Moulds – dark bronze cast covers of sex dolls – touch on an anthropological constant, a longing from which the effect produced by art itself is also derived, but in doing so they echo a persistent claim of availability paired with an omnipresent logic of exploitation.
During the production process, Clement routinely photographed the shells, but discovered that her cell phone camera reproduced the concave hollow forms as convex volumes. The series Body delves into this perpetual twisted interplay which fundamentally transforms emptiness into an abundance to be craved. Screens like dystopian veils of Maya seamlessly project us in and out of the real, taking up a space where perception
gets distorted and desires are commodified.
Louisa Clement, Head 54 (2015)
Louisa Clement, Avatar 8 (2016)
Louisa Clement, Mold 1 Female (2019)
Louisa Clement, Body 5 (2019)
Louisa 0 and her three algorithmic identities
- essay by Ulrich Trottenberg
"We have become accustomed to anonymous voice assistants like Siri and Alexa, to their vast knowledge and their ever friendly willingness to share their knowledge and skills with us or to entertain us with our choice of music. With no preparation and faster than any human, Alexa can answer questions such as, “Alexa, how much is 16 to the power of 64?” And when asked if she has consciousness, she answers, “Yes, I think about a lot of things,” or even, “Yes, I think, therefore I am.”
Louisa Clement's three algorithmic identities are not anonymous. They don’t just have abstract general knowledge and don’t just provide evasive answers – to ‘unsuitable’ questions – like Alexa; they represent Louisa. They look a lot like Louisa, and answer questions in much the same way as Louisa would perhaps answer them. They even answer very personal questions that one might not even pose to the real Louisa.
Algorithmic identities? Or better yet: robotic clones? Talking Louisa dolls? Active Repräsentantinnen (Representatives) of Louisa's personality? Louisa's living voice assistants? What should we actually call the three animated art objects, these results of artificial intelligence (AI), machine learning (ML) and computational linguistics? Louisa calls them Louisa 1, Louisa 2 and Louisa 3. In this list, she – Louisa, the real person – would be Louisa 0. (Mathematically, 0 has always been a very special number).
Louisa 1, 2 and 3 are art objects, but they are also technical achievements from the world of AI. If one has a little knowledge about the beginnings of artificial intelligence, then one might think about what Alan Turing would have asked them if he had met them.
In 1950, before the term ‘artificial intelligence’ was even in use, Alan Turing, one of the greatest mathematical minds of the last century, conceived of the 'Turing Test': this thought experiment, which is still frequently quoted and discussed today, aims to establish a benchmark for how close AI has come to human intelligence. According to the test, an AI system (robot, computer, algorithm) is considered to have fully passed the test if the system can be given any task or asked any question and it is impossible to conclude from the solutions and answers whether one is dealing with an artificial or a human intelligence. (In the case of the task ‘16 to the power of 64’, it would, of course, be immediately clear from the speed of the answer that you are not dealing with a human being. For such tasks, a pocket calculator is already far superior to a human.)
New advances in the field of AI are reported every day in the media. AI was the official theme of Science Year 2019, and since at least this time, research and economic policy has been paying great attention to the topic, and gushing – somewhat naively – about its virtually un limited possibilities.
In fact: AI's fields of application are extremely diverse and already range from image, face and handwriting recognition to language processing, including automatic translation, conceptualisation and 'creative' writing; from automated medical diagnostic procedures, robotic football teams, (not quite yet) autonomous cars and driving assistants to art – all the way to Louisa 1, 2, and 3. The vast increase in computer performance over the past 30 years has been crucial to AI's success. A smartphone today can already calculate as fast as the world's most powerful computer could 30 years ago – and the volume of available data worldwide is growing at an almost explosive rate.
Our questionandanswer games with commercial voice assis tants have become more challenging. I have not yet had a chance to speak with Louisa 1 to 3, but I assume they are as smart as Alexa. It is difficult to predict how far I will be able to go with my knowledgebased questions. Actually, I think Louisa 0 is still the more interesting conver sationalist. But maybe that’s not generally true at all. Perhaps I would learn things in conversations with Louisa 1,2,3 that Louisa 0 wouldn’t say to me or that she would find embarrassing. For example, I could ask Louisa 1 without any inhibition, “Louisa, are you in love?”, or even more intimate questions. Maybe Louisa 2 would give a completely different answer to the same question. After all, the three Louisas do not (only) have static knowledge. They are constantly learning. They gather new information with every conversation they have, and as a result, the three Louisas constantly evolve as learning machines. They are not only AI Louisas in a general sense, but also ML Louisas in the strictest sense. The three individual Louisas can learn very different things, and possibly develop in different ways; comparable, perhaps, to identical twins who grow up in different environments.
Setting the conceptual stage for her artificial, AI- equipped, sexually functional doppelganger, Louisa Clement takes a quantum leap as she devotes herself to the exploitative altar of digital capitalism. Clement’s advanced self-portrait is also to be seen as a subjectified article offered to the desire of others. Sheer byproducts of present- day ubiquitous self-marketing, both the artist body and her ‘self’ are reduced to available item, commodified and perversely straddling the real and the virtual. In an existential quest to fathom human destiny, the Representative enacts a radical attempt to give recognition to the self, vulnerable entity doomed to dissolve into data.
A few words about machine learning: socalled learning algo rithms form the core of ML. ‘Algorithm’ is a digital buzzword that is in constant use, often misunderstood and suspected of being dangerous. In fact, algorithms are the core of everything digital, they control all digital processes and devices, and form the basis of every computer program. In their traditional form, they comprise a sequence of precise instructions and consist of a finite number of precisely defined, individ ual steps. Unlike conventional, rulebased algorithms, learning algorithms do not calculate a result simply by processing a sequence of commands. Instead, they first go through a learning phase. In this process, internal numerical values (parameters) are modified by processing a large, often vast amount of sample data so that the algorithm independently recog nises and rehearses patterns in the data, finds new features and gradu ally improves its functionality, and consequently its results. It is said that the system is ‘trained’ or that the algorithm ‘learns’. ML systems have learned to beat the world’s best human opponents at challenging games like chess or – most impressively – Go, as well as to weed out spam mail, distinguish diseased from healthy tissue, recognise facial expressions and identify hate mail and fake news on the internet, etc. To do so, the patterns they are to recognise do not need to be predetermined or ex plained to them. In many cases, they find them themselves.
Nevertheless, compared to human learning, machine learning is an extensive process: based only on a few examples, a child learns to distinguish a dog from a cat and an apple from a pear. Conversely, an algorithm generally requires many thousands of training examples be fore it can recognise decisive features (patterns) and more or less re liably master the distinction. Louisa 0 posed and answered thousands of questions. On the ba sis of this initial information, Louisa 1, 2 and 3 have learned to be Louisa. But if, in an algorithmic sense, Louisa 1, 2 and 3 are no different from Louisa at all – what about the ‘uninhibited questions’? Maybe I would hesitate to ask overly personal questions because I don’t want to offend Louisa’s representatives, and thereby Louisa. At this point, I realise that such a representative, a machine, could perhaps evoke feelings in me. This immediately brings up the major issue of ‘emotional AI’, and it quick ly becomes controversial. Emotional AI is not the only philosophical and ethical controversy that this exhibition will and aims to trigger. All the historical questions on the subject of humanlike machines and artificial humans, etc., which have been addressed in many different projects and art contexts, re surface: Homunculus, Frankenstein, World on a Wire, 2001 – A Space Odyssey, Matrix, Her, Klara and the Sun, and so on. The fascinating visions of science fiction that the topic of AI inspires are back, but not as figments of the imagination or theoretical constructs, instead they are tangible and physical, and can be experienced by everyone.
Although one’s experience of the three Louisas today may be rather playful and not an immediate threat – discussions regarding onto logical questions of identity and the legal and ethical aspects of AI, as well as questions concerning the control of AI development, are still in evitable in the face of an encounter with the Louisas. As an artist, Louisa 0 wants to ask these questions. She wants to provoke.
Most AI/ML experts and developers are still in agreement that AI’s major successes are limited to ‘weak AI’, i. e. solving specific individual problems, and that they will remain so for the foreseeable future. How ever, the question arises as to why these many specialised areas are not able to grow together or to combine on a longterm basis in order to gradually approach a ‘strong AI’ (a comprehensive AI that is no longer limited to specialised tasks). Yet, the imagination reaches its limits when one considers the possibility of combining, in the distant future (?), the tremendous achievements of neurology and molecular biology (gene scissors) with the developments in AI that will take place over the next 50 years. We are more than happy to leave such fantasies to the trans humanists.
Back to Louisa 1, 2 and 3. What central questions do the three arti ficial humans ask us? Are they ‘only’ questions about the identity of such systems? Clearly, the three Louisas do not have emotions and empathy, but they do trigger emotions in their human counterparts, e. g. when they are offensive. And beyond that: don’t the three art objects also rep resent much more? Don’t they also contribute to the big questions we ask ourselves in the context of global AI developments? For example, questions concerning the ability to explain and control AIbased deci sions. The three Louisas are not making important decisions for us, but we have to address the question of how we will remain in control when AI algorithms are making vital decisions, for example in judicial matters, in medical diagnostics or even ‘just’ in economic processes.
In addition to confronting the ethics of machines and algorithms, the three Louisas – in the land of technology sceptics – also confront us with the question: where do we actually stand (in Germany) with AI, and how do we deal with it? Where do we go from here? Do we even have a role to play in international development? Are we helping to shape it, or are we just lagging behind? This visionary and equally provocative Louisa project shows us that Germany – aside from its combination of classical knowledge of engi neering, foundation in theory and competence in advanced AI research – can make an important artistic contribution to illuminating the question of human identity and enriching people’s quality of life with applied AI.