A Is For Another

Here is what I made with my 2020 Hivos Digital Futures Fellowship: A Dictionary of AI called A Is For Another. “A Is For Another answers the question: what is AI? Or, how do we understand what it means to be human and non-human through artificial intelligence? This dictionary presents how intelligence, humans, machines, data and mind exist across a variety of cosmologies, sciences and art practices as perceptions change over time.”

After I finish my dissertation, I will be expanding and developing this further: workshops, teaching, inviting new entries. Contact email on the website if you would like to stay in touch about this project.

 

Cloud Cosmogram. The last of 2019.

Over many months this past year, Johannes Bruder and I worked on writing an essay called Cloud Cosmogram as part of the Data Farms project. And with the designer Selena Savic, we developed a companion visualisation.

“Taking a step back from seeing the cloud as a medium that virtualizes social relations, we take the technicalities, standards and certificates that guide data centre operations to compose an alternative cosmogram of the cloud. In so doing, we move away from the metaphorical elusiveness implied by the figure of the cloud to ask how data centre operations enable, challenge, and to some degree replace, traditional sociality through the organizational intricacies of ‘post-human institutions’: a ‘new form of architecture for data and machines … almost liberated from human intervention and entirely shaped by technological rationales’. Our cosmogram does not reveal a disguised materiality – it is neither another mapping of its very material infrastructure nor a visualization of its equally material carbon footprint. Instead, we concentrate on the co-location of objects and technical practices that determines the place of the human in the data centre…”

New Projects and Publications

It’s a busy week; it is all busy weeks till the end of the Northern year. Three things I have been working on have come out into the world: announcement of a new seminar series I will host at the School of Disobedience in 2020; a journal issue on the Spectres of AI that I guest-edited with a collective; and an essay-report about the work with Transmediale’s Affective Infrastructures Study Circle. More details below.

Over the past year I have been developing a talk slash drafts of something called a “speculative inventory about almost-cyborgs”. Over time this has morphed into a seminar at the second edition of the Berlin-based School of Disobedience. It is called A is for AnOther: Beyond Artificial Intelligences. I will also be working with a small bit of (new) funding to bring together the syllabus and materials from the course to a website near you (stay tuned!).

A mammoth project has now gone out into the world. I worked with my home institution, Leuphana University’s Centre for Digital Culture’s Spheres Journal Editorial Collective to bring out a new issue. The Spectre of AI, Issue #5 is now out. I am especially pleased with two pieces I shepherded through the process: Manan Ahmed Asif’s essay about the colonial pre-history of data science embedded in the US tradition of Area Studies; and Noopur Raval’s brilliant and thoughtful response-companion piece to it. I think this is the #decolonisingAI research we need in this world.

(I wrote something too, a comment on an essay about AI and its erasure of difference)

Third and not least, a report of the project I did with the Transmediale 2019 team. A year ago Femke Snelting and I worked with curator Daphne Dragona to moderate a Study Circle on the theme of Affective Infrastructures. We brought together a group of scholars, artists, academics and technologists to work together to develop an event as part of the Transmediale Festival. I wrote about our process, and about ‘affective infrastructures’ here. Daphne wrote a fine piece about it too, as did my co-moderator Femke Snelting, here.

Things in the mix

Things are changing. I’m on the road. There are some new ideas and projects in development, yet I’m trying to keep my eye on the big one (the dissertation).

2019 has been a year of saying no to big conferences and events, and saying yes to smaller and more focused events and workshops. As I write I am sitting in the surreal-y beautiful library of the Villa Serbelloni at the Rockefeller Foundation’s Bellagio Center. I have been here for three weeks as a Resident and part of their thematic month on AI; I have been working on another dissertation chapter and a shorter paper based on it. In this time I have had 13 other amazing scholars, artists and technologists to learn from and share space and time with. It’s been amazing.

I’m excited about some upcoming talks.First I will be at Cosmopolis #2: Rethinking the Human in Paris on October 26th with Tabita Rezaire, Nishant Shah, Emo de Medeiros, Francois Knoetze, Tricky Walsh, Yuk Hui, Clapperton Chakanetsa Mavhunga and Mi You; and then keynoting at the India HCI conference in Hyderabad on November 2. These events seem quite different, and they are, but the work I will present at the two is about [changes in, the reshaping of] the figure of the human in the context of AI/ML.

And there is a lot more in development.. Stay tuned…

Things before, and things ahead.

I did not take much of a summer holiday this year, though I did have a short (always too short!) trip to Italy with friends; attended a dear friend’s fantastic festival-glamping wedding in the Devon countryside; and spent a lot of time stay-cationing, watching TV shows and beating the Berlin heat with ice-cream and homemade yoghurt. Oh and around all that ice-cream eating, I:

  • ran a session at the Citizen Lab Summer Institute at the University of Toronto that asked how and if computational tweaks and tactics could be designed in to mitigate specific dimensions of online abuse. This was a bit of a follow-on from my work with them on gender and digital security.
  • spent five days at Akademie Schloss Solitude outside Stuttgart working with my co-juror, the artist, writer and attorney, Rasheedah Philips, selecting five recipients of a ‘Digital’ Solitude Residency. Check the fantastic lineup of ‘Web Residents’ selected by Jonas Lund and in response to the theme Rigged Systems.
  • did a talk at Re-Imagining AI a symposium organised at IXDM, Basel (with Johannes Bruder). Johannes and I have also been writing a new piece together called Cloud Cosmogram for the Data Farms project by Tanya Notley, Ned Rossiter & Brett Neilson at the University of Western Sydney. We’re working on a visualisation. It’s all coming together.. !
  • did a guest lecture titled Ceci n’est pas Ethics: Computational Ethics and the Making of Machine Autonomy at my new, temporary home, the Hochschule für Gestaltung, Karlsruhe. The talk was based on a chapter I’m writing for the dissertation, so it was fresh, but is not ready for wider circulation. There has been a lot happening on our AI+Media Philosophy Project (KIM, in German) project. We also did a short kick-off workshop with our partners. Stay tuned for a lot more coming out of the KIM Group.


And now, we’re heading to another season. In addition to managing reviews for a brand new Social Sciences and Humanities Track at the ACMFAT* Conference, here is what else is coming up. I will

  • moderate two sessions at Berlin’s International Literature Festival in September. There’s a special focus this year on ‘Automatic Writing 2.0’ (link goes to PDF);
  • be on a panel at an Akademie der Künste event in Berlin, AI + Art? ;
  • speak to new Fellows of the Mercator Foundation about AI;
  • spend most of October as a fellow at the Rockefeller Foundation’s Bellagio Centre in Italy working on a paper related to some of my PhD research, and my work at the KIM project.
  • keynote at the India -HCI conference in Hyderabad in early November.

Please come and say hello if you’re at these places!

Two new essays

I spent most of the past winter and spring writing. I’m happy to report that two pieces of work from that time are now out in the world. The first, for Deep Dives, a digital imprint by Point of View (an organisation I worked with way back in the day), is titled You Auto-complete Me: Romancing the Bot. It emerges from my work over the past year on AI and affective computing (for Transmediale, and an academic and cultural symposium at the HKW in April, called Life Forms). This non-academic, personal essay was a lot of fun to work on; not only did it allow me to write ‘outside’ of academic norms, but it also challenged me to write about computational topics for a diverse and general audience. This has always been important to me as someone whose practice has been embedded across different communities. And, how do you tell compelling stories about technology? I think there might be space to write in strong and sophisticated ways inside academia too (at least I am gambling on there being a place for this! We will know soon! )

The second essay, Tipping the Scale: Notes on the Topologies of Big Data Platforms (opens as .pdf directly), was commissioned by IT For Change, a Bangalore-based technology policy and research organisation, as part of a series called Platform Politick. Platform Politick accompanies the organisation’s fascinating new body of work about platformization from the South. In this essay I address scale as a feature, and aspiration of, big data platforms. How does the demand for big data platforms to be more contextually sensitive and aware square off against their ambitions to scale? How does accountability work when most mechanisms for accountability are scaled to fit particular situations and do not necessarily address the vastness of platforms? So, I tried to think through how scale forces us to rethink what platforms are, what they mean, and how to regulate them.

What’s next: one more essay waiting to be published; and another one in the works….

Talks and events

October 2019. A speculative inventory of almost-cyborgs. In conversation with Nishant Shah at the Brighton Digital Festival.

October 2018. Making the Black Box Speak. Season 2 Episode 3 of The Future of Demonstration. A performance event and salon series by Gerald Nestler and Slyvia Eckermann. Vienna, October 24-26, 2018. Video here.

May 2018. Solar punk and going post-post apocalyptic. re:publica 2018. Berlin.

March 2018. Dangerous Conjectures: Race, Nation, Class. Haus der Kulturen der Welt, Berlin.
_ Data discrimination, Dystopia and the Future of Citizenship; in conversation with Nishant Shah; Video here.
_ Thinking Through Racisms. Video here.

September 2018. Soundbytes: “Thoughts of the week” as part of a Deutsche Bank series. Videos ‘What’s the idea of intelligence?’ here and ‘Should AI have ethics’ Video here.

Joining KIM

Excited to announce that I have a new, short-term and part-time position at the Künstliche Intelligenz und Medienphilosophie (KIM) Research Group at the Staatliche Hochschule für Gestaltung (HfG), Karlsruhe (the Karlsruhe University for Art and Design). I will be working on the project AI and the Society of the Future supported by the VW Stiftung.

OWNW Talk on Affective Infrastructure and #MeToo: Script

Here is the full text of the talk I gave at the first salon organised by the new Otherwise Network in Berlin on April 6, 2019. You can watch the talk here:

(Thank yous – Otherwise Network. Audience)

Humbling to have the space to articulate ideas and think with people.

And especially on such a beautiful Spring day when you could be outdoors.

This talk will have three parts and I hope to lead you all through it gently and smoothly. There is theory and if theory is not your thing, then just hold tight and breathe, just like when there is turbulence and you’re sitting at the back of the plane.

One. Giving Data to Life

in the beginning was data, but it was small
the personal story, the case study, the story told through sobs after the third drink. or not at all.

Quote. You taught me my first workplace lesson. I was 23, you were 43. I grew up reading your smart opinions and dreamt of being as erudite as you. You were one of my professional heroes. Turns out you were as talented a predator as you were a writer. It was more date, less interview. … I escaped that night, you hired me, I worked for you for many months even though I swore I would never be in a room alone with you again. Endquote. (By Priya Ramani calling out abuse by MJ Akbar)

Women who work at microsoft on the Xbox team say that they have been called ‘bitch’ at work at least once. Another refused to sleep with a colleague and was threatened with death. Yet another was asked to sit on a colleague’s lap at a meeting with HR and executives present and no one said anything.

About 18 months ago, these small data became a hashtag.
the data became a flood
the details didn’t matter
you just had to say two words.
the data had volume velocity variety
so then we could call it big data.

But like big data, it was not raw, but it was made of things that felt raw.
but, data is always cooked. it always comes from someone making choices about what to put in and what to keep out.

it is no surprise that the most popular and dominant forms for organising information – the spreadsheet, the dataset, the hashtag – have converted what was always my story and your story, into a data story. the body had to become data so that it could be seen.

so we had the shitty men in media list
the shitty men in the arts list
the list of sexual harassers in academia
the hollywood list
the bollywood list

let me talk about just one list, because all data are different and have their own histories.

in 2017, two Dalit feminist students of Indian origin, Raya sarkar and inji pennu, both in California, put up a spreadsheet called LoSha List of Sexual Harassers in Academia. The include 50-odd names of Indian men academics living in India, Europe, the UK and the US and included some illustrious names. Before our eyes the spreadsheet expanded to 70+.

(Disclosure: I have a friend whose name is on the list. Also an acquaintance who lives here in Berlin)

The list had just a few columns
Who
What
Where
When

data, which offers all kinds of analytical possibilities instead was referred to as empty, desperate, chaotic, and anarchic. it made no sense to just make a list online.

and the thing is, people have been making accusations for a long time. Like in the whisper network. but like really great infrastructure, it is mostly invisible. that’s its point. but it was not enough to just talk among ourselves. something is significantly broken.

The first thing that happened in response to the LoSHA list was that a group of senior Indian feminist academics wrote an open letter in defence of one person on the list, one of their friends, The old feminists were suspicious of social media, of accusations that could be made on google spreadsheets, without “due process”. It is important to note that most of these older academics have been closely involved in social justice movements and activism on rape, sexual harassment, dowry deaths, honor killings and so on. Some of them have been dear mentors and inspirations.

The younger women responded: you oldies are obsolete. your stupid blog is obsolete. your networks are obsolete. the women responded that they had taken to social media and the internet because nothing else had made good on its promise of due process.

So the immediate outcome of the list of sexual harassers in academia was actually a conflict about whose feminism was purer.

This is not unlike the moment three years ago when a group of people set up a website detailing Jacob Appelbaum’s sexually abusive behaviour and sexual bullying. The first thing that happened was that a group of six women wrote a response quite similar to the one the Indian feminist academics did. The website, Ourresponse.org is now not available and has anime characters dancing across it instead.

But what I really want to get to here is what happens the day after the revolution, when the maidan, the square, has to be cleared of torn posters and coffee cups? It is not actually as simple as making allegations online. People who make allegations of sexual abuse know this already.

Women are having to quit jobs, and find new ones, to organise evidence in cases where things do go to court, or having to go through internal complaints committees, or having to withstand online and offline attacks for making allegations, find emotional resilience to go through the aftermath of putting your story out there. They are sharing information, preparing with and supporting each other, encouraging each other in WhatsApp groups, email lists, offline. This is the work of care and support that comes after.

The other big shift taking place is the flowering of uncertainty and unhappiness, a discomfort about how broken sociality is. Consider this.

The actor Henry Cavill says in an interview to Australian GQ.

quote It’s very difficult to do that if there are certain rules in place. Because then it’s like: ‘Well, I don’t want to go up and talk to her, because I’m going to be called a rapist or something’. So you’re like, ‘Forget it, I’m going to call an ex-girlfriend instead, and then just go back to a relationship, which never really worked’. But it’s way safer than casting myself into the fires of hell, because I’m someone in the public eye, and if I go and flirt with someone, then who knows what’s going to happen? End quote

So are having to deal with hesitation, awkwardness, self consciousness, nervousness, uncertainty, fear. This emotional management calls for a coordination of mind and feeling, watchfulness, an awareness of body, a re-negotiation of social scripts, a struggle to find the right words.
How might the difficult, uncertain, awkward aspects of intimacy might offer new territories for thinking about the emotional management we have to be doing to create new ways of relating.

I do not want to valorise some old fashioned gendered notion of romance that existed in the past and say that was a kinder gentler time and that things have become complicated now. It was not better back then, it never was, it was just a time when women were either silent or muttered to each other under their breaths.

In moments of uncertainty, we rush to establish things that reassure us. So consider the rise of sexual consent apps like Good 2 Go and SaSie. Or my favourite, Legalfling that puts consent on the blockchain.

And I quote from their website:

“Communication and understanding is key. Talking about rules and boundaries before sex can be awkward and uncomfortable for some. Having an app that clearly shows the rules of engagement as well as personal preferences, can remove misunderstandings and prevent unintentional bad situations.”

or Netflix passes a five second rule saying their employees cannot make eye contact for longer than 5 minutes or stay in a lingering hug.

Two.

slide :the world is in bad shape and I am overwhelmed: – agnes varda

I want to consider MeToo and its aftermath as a kind of affective infrastructure. And this is the part where I draw on Lauren Berlant’s concept of the same name.

But first we have to work really hard to dismantle our idea that infrastructure is material. That infrastructure is roads, water systems, nuclear power plants, airlines, museums. Brian Larkin, Susan Leigh Star and others reminds us that infrastructure is as much things as the relationships between things. It is values and symbols, beliefs that sustain organisations, it is a multiplicity of cultural factors.

Even things like roads and highways are about cultural values of freedom, autonomy, and individuality. The cultural history of US highway infrastructure and engineering during the Cold War is a particularly interesting little rabbit hole.

Infrastructure is things, it is people says Abdoumaliq Simone. If you look at how roads are built, in India, it often starts with a thin man in a thin cotton shirt, hacking at the hot tar road with a spade, under the blazing sun.

So let us be wary of materiality as infrastructure and infrastructure as materiality. And let us be wary of the metaphors that come to mind: networks and grids. We are so deep in network thinking that it is difficult to consider an ontology without networks.

“Just because a space on a grid is shared intends nothing about the affective and material substance or even the fact of membership, “ slide

Lauren Berlant invites us to try.

‘Affective infrastructure’ pushes up against the notion that ‘infrastructures’ are physical and material relationships; I like to think of affective infrastructures as more like synaptic relationships, in which neurotransmitters convey messages across the gaps between neurons by osmosis, rather than how muscles are attached to bones by tendons.

Affect includes the cognitive, proprioceptive, behavioural and psychological; Facial expressions, respiration, tone of voice, posture, all transmit affect, which means “we are not self-contained in terms of our energies. … Because affect is unformed and unstructured (unlike feelings and emotions) it can be transmitted between bodies.” So a starting point for thinking about ‘affective infrastructure’ are things that are ephemeral, unstructured, things that are more in the air.

What does it mean to be connected to others not only by emotions that are shaped and shared by language, but desires that are visceral, that may be pre-linguistic and bodily, and affect that overlaps both these?

I am not saying that our shared emotions, means that we have an affective infrastructure.

quote: suddenly we dilute what we call structural by shifting the force of the normative infrastructures from the state and commodity capitalism into the ordinary that also includes the local plural intimacies and associations that make like sticky and interesting for it. but this multiplication of forms in movement, not a denial of colonial/racial/patriarchal/class inheritance ( p 408) endquote

the other signficant aspect of affective infrastructure is the idea of movement rather than something fixed and established.

Movement is what distinguishes infrastructures from institutions, although the relation between these concepts and materialities is often a matter of perspective. Institutions enclose and congeal power and interest and represent their legitimacy in the way they represent something reliable in the social, a predictability on which the social relies. Institutions norm reciprocity. What constitutes infrastructure in contrast are the patterns, habits, norms, and scenes of assemblage and use. Collective affect gets attached to it too, to the sense of its inventiveness and promise of dynamic reciprocity

Affective infrastructure trouble the tightness of sociality, of what we think of as infrastructure and shows us how things are broken. It is infrastructure of that which is destablising and shifting, and yet holds together.

It proposes that the commons concept is a powerful vehicle for troubling troubled times. For the very scenes in which the concept attains power mark the desire for living with some loss of assurance as to one’s or one’s community’s place in the world, at least while better forms of life are invented and tried out. The better power of the commons is to point to a way to view what’s broken in sociality, the difficulty of convening a world conjointly, although it is inconvenient and hard, and to offer incitements to imagining a livable provisional life.

The list, and the challenges in different approaches in feminism itself, and how to deal with generational divides, the impact of me too
our different thoughts on what aziz ansari did.
The disagreements about what Joe Biden did, smelling people’s hair
It would be easy if all violations were about rape or non consensual masturbating in front of a date or colleague, but it is not..

and there are those cases that we maybe do not agree with, where as women we look at another generation, or another women’s discomfort and feel frustrated or we disagree.

I think that affective infrastructure and what binds us as we move are not made by our movement, they are not pre-existing, they are not material, the connections are in our uncertainty. So when we resist that which is difficult or awkward I would like to think that is possibly when we are actually moving towards each other.

3.. The Life of Emotional Data
Affective computing

So I want to shift a little now and talk about affect in another context, and the ways in which affect is more important than ever and how we are being told what affect is. Because affect is already the next big thing in AI. And very soon, I gather that the kinds of affect we have just been talking about, these uncertainties and awkwardnesses and exhaustions will appear as glitches, bugs or be classified as not affect. So there will clearly be a structuring of affect to mean something very specific.

There are two aspects to affective computing. There are things like Consider the following examples. Woebot, a therapy bot in an app, helps track your mood, and talks you through the reality of living with depression. And Lovot, – “powered by love” – is a silly, adorable little robot that just wants to make you happy.

There is also Paro, a soft, cuddly, baby harp seal robot that does not move and makes soft, animal noises. It is effective in therapy with Autistic children, and older people in homes, including those with dementia. Paro does not make the first move, remembers how it was held and does not feel bad if it is rejected or ignored.

Then there is the other kind of affective computing, how our faces, voices and body language are mined to feed the growing industry of ‘Emotion AI’. Emotional artificial intelligence is “the process of giving machines the ability to recognise and react to human emotion.”

Companies like Affectiva, an early pioneer in the field, are developing datasets of human affect and emotion to train future bots and digital interfaces to “understand all things human” as their tag line goes.

At its foundation, affective computing is about breaking down human facial expressions into a map of emotional states. Here is what Affectiva, one of the most important companies developing this technology says about how it works:

Humans use a lot of non-verbal cues, such as facial expressions, gesture, body language and tone of voice, to communicate their emotions. Our vision is to develop Emotion AI that can detect emotion just the way humans do.. Our Emotion AI unobtrusively measures unfiltered and unbiased facial expressions of emotion, using any optical sensor or just a standard webcam. Our technology first identifies a human face in real time or in an image or video. Computer vision algorithms identify key landmarks on the face – for example, the corners of your eyebrows, the tip of your nose, the corners of your mouth. Deep learning algorithms then analyze pixels in those regions to classify facial expressions. Combinations of these facial expressions are then mapped to emotions.

Affectiva is using this technology in the automotive context to understand “drivers’ and passengers’ states and moods, in order to address critical safety concerns” like road rage and driver fatigue. So a future car might issue a calming and authoritative order to an angry driver to pull over; or may increase the air conditioning to make a sleepy driver more uncomfortable, and therefore alert. Our faces and voices betray a lot about us, like it we are lying or angry. In the UK, one in five people lie to insurers or attempt to defraud them, so affective computing is being used to identify possible lying and cheating through facial features and voice.

But affective computing is also for humans to have a better experience relating to humanoid robots, digital assistants and chat bots. These are perceived to be awkward to interact with because they lack affect. When Google launched Duplex, its voice-based digital assistant, they emphasized how natural-sounding it was:

The system also sounds more natural thanks to the incorporation of speech disfluencies (e.g. “hmm”s and “uh”s)….adding synthetic waits, which allows the system to signal in a natural way that it is still processing. (This is what people often do when they are gathering their thoughts.) In user studies, we found that conversations using these disfluencies sound more familiar and natural.

In this, the not-human-ness of the voice assistant is trained out by real-time supervision of the software by a human. The system then learns how to negotiate around that mistake the next time. Making a -bot sound more human in this way is to reduce ‘friction’, that is, to not alert the human on the other end that they are talking to a -bot. Or to suspend disbelief for longer that we are talking to a human

So we can very quickly see how these systems of capture will create a map – this is what happiness looks like, this is what surprise looks like and so on.

But here is the thing with affect, affect is also awkward, unique, personal, bodily, physical, may be easy to capture but is difficult to analyse and process. As humans we have problems recognising and understanding what people are saying, but with the capture of enough human emotional data, we can possibly establish some clarity about what things mean.

Part 4.

So I want to close by acknowledging two very different kinds of moves here. And we need to struggle with the work of theorising these together, or at least in light of each other.

There is what Berlant urges us to acknowledge, that sociality is difficult and broken and often unbearable, and socially necessary proximity is unbearable. We are constantly dealing with the pain and privilege of others and this is hard.

But we are also seeing simultaneously a growth in the industry of the capture of emotions and affect in an attempt to make human feeling legible, and to control people and manage our errors and deviance. And through acts of prediction as well. Capturing and analysing our emotional data to predict what we might do, or establish paths of expected behaviour based on patterns emerging from prior situations.

And I leave you with that conundrum, now.