Return to Video

34C3 - Humans as software extensions

  • 0:00 - 0:15
    34C3 preroll music
  • 0:15 - 0:19
    Herald: And yeah, please give
    a warm welcome to Sebastian!
  • 0:19 - 0:25
    applause
  • 0:25 - 0:28
    Sebastian: Hi, and thank you for the
    introduction. Thank you all so much for
  • 0:28 - 0:32
    having me. So, what am I gonna
    talk about tonight? I will
  • 0:32 - 0:36
    of course, being an artist, talk a bit
    about some art projects but mostly
  • 0:36 - 0:42
    I will speak about the research I do
    as part of my practice, specifically
  • 0:42 - 0:47
    about humans as software extensions.
    And towards the end I will
  • 0:47 - 0:52
    share some thoughts on why being a human
    software extension could actually be
  • 0:52 - 0:57
    something maybe not positive but brings
    some new possibilities. So I will try to
  • 0:57 - 1:06
    end the talk on a positive note. In 2008,
    the science fiction movie "Sleep Dealer"
  • 1:06 - 1:11
    speculated about a future which couldn't
    be more timely. The border between Mexico
  • 1:11 - 1:16
    and the US has been closed. Therefore,
    immigrant workers in the US have been
  • 1:16 - 1:22
    replaced by robots. However, the robots
    are remotely controlled by people in
  • 1:22 - 1:29
    Mexico who have their bodies plugged
    directly into the network. Then, 2 years
  • 1:29 - 1:35
    later, in 2010, the CEO of CrowdFlower,
    which is a crowdsourcing and cloud work
  • 1:35 - 1:41
    platform, Lukas Biewald speaks of a
    similar situation: "Before the Internet,
  • 1:41 - 1:46
    it would be really difficult to find
    someone, sit them down for 10 minutes and
  • 1:46 - 1:51
    get them to work for you, and then fire
    them after those 10 minutes. But with
  • 1:51 - 1:56
    technology, you can actually find them,
    pay them the tiny amount of money, and
  • 1:56 - 2:01
    then get rid of them when you don't need
    them anymore." Biewald's remarks, however,
  • 2:01 - 2:08
    are not Science Fiction. Instead, they
    describe a contemporary condition. So,
  • 2:08 - 2:14
    humans as software extensions: What is
    this condition? I would summarize it as
  • 2:14 - 2:19
    people extending computational systems by
    offering their bodies, their senses and
  • 2:19 - 2:25
    their cognition. And specifically bodies
    and minds that can be easily plugged in
  • 2:25 - 2:30
    and later easily be discarded, so bodies
    and minds algorithmically managed and
  • 2:30 - 2:36
    under the permanent pressure of constant
    availability efficiency and perpetual self
  • 2:36 - 2:43
    optimization. So, as such, humans as
    software extensions are both, the
  • 2:43 - 2:49
    foundation and the result, of a mega
    structure which Benjamin Breton calls "The
  • 2:49 - 2:55
    Stack". It's a computational totality of
    planetary scale so somehow we can imagine
  • 2:55 - 3:01
    it as a planetary scale computer
    consisting of a stack of layers from rare
  • 3:01 - 3:08
    earth minerals to data centers to bots and
    people which in this model are exactly the
  • 3:08 - 3:14
    same. So, in this computational totality,
    even the smallest nodes can be addressed
  • 3:14 - 3:21
    and can be programmed. What The Stack also
    describes is a new geography so Google
  • 3:21 - 3:28
    Maps defining borders or, as just seen in
    Sleep Dealer, the US building a wall while
  • 3:28 - 3:35
    still being able to plug right into the
    bodies of Mexican workers. But to be
  • 3:35 - 3:42
    clear, from my point of view, the model of
    planetary scale computation as a totality
  • 3:42 - 3:49
    is as much a reality as it is also a
    gigantic fantasy and ideology of power,
  • 3:49 - 3:53
    optimization and efficiency. So this is
    the state of self-driving cars right now,
  • 3:53 - 4:01
    at least from the perspective of a Ford.
    So I would now like to give some examples
  • 4:01 - 4:06
    about what I mean by humans as software
    extensions and what effects these, like,
  • 4:06 - 4:13
    this way of managing digital labor has. 3
    years ago, I worked on a piece in which I
  • 4:13 - 4:18
    explored digital colonialism and, among
    other things, I explored Google's and
  • 4:18 - 4:24
    Facebook's attempts to integrate into
    their services those 2/3 of the world
  • 4:24 - 4:31
    population who are not online, yet. As you
    all know, Google wants to use balloons,
  • 4:31 - 4:37
    Facebook wants to use drones and they want
    them to circle above these areas that do
  • 4:37 - 4:43
    not have any internet connection basically
    sucking that which is below up into the
  • 4:43 - 4:48
    network. So the piece of which you see a
    little bit here is called "How to Appear
  • 4:48 - 4:54
    Offline Forever" and it consists of a mix
    of found material like videos, images, and
  • 4:54 - 4:59
    questions and there's also a layer of
    stories written by people from Silicon
  • 4:59 - 5:08
    Valley, Sri Lanka and Zambia which are all
    locations of importance to this piece. And
  • 5:08 - 5:13
    so in order to get in contact with people
    in Sri Lanka, I ended up using the
  • 5:13 - 5:18
    outsourcing platform Upwork which offers a
    highly efficient interface for hiring
  • 5:18 - 5:24
    freelancers from all around the world. On
    that platform, you can sort freelancers by
  • 5:24 - 5:31
    price, skills, rating and then you can
    pick whoever you think fits the job best.
  • 5:31 - 5:36
    And their user experience of hire and fire
    is well crafted. So it's software
  • 5:36 - 5:40
    extensions that can be plugged in and
    removed again easily. It doesn't matter
  • 5:40 - 5:54
    where they are, who they are as long as
    they get it done. So once the freelancers
  • 5:54 - 6:00
    in Sri Lanka got to work I noticed that it
    was not only able but also encouraged to
  • 6:00 - 6:06
    spy on them. Upwork records every
    keystroke and regularly takes screenshots
  • 6:06 - 6:12
    while freelancers work, building a growing
    diary of their activity. So I found myself
  • 6:12 - 6:17
    in a situation in which I wasn't only
    being surveilled by corporations or
  • 6:17 - 6:24
    states, I was also doing the same myself,
    managing my extensions spying on them in
  • 6:24 - 6:29
    order to monitor their performance. So
    this is not them, like corporations or
  • 6:29 - 6:36
    states, spying on us, us protecting
    ourselves against them through encryption.
  • 6:36 - 6:41
    This is all of us fighting for our place
    in the network trying to be valuable
  • 6:41 - 6:46
    nodes. So this is me and one of the
    freelancers which is a very lovely lady I
  • 6:46 - 6:52
    got to meet in Sri Lanka later. So if you
    look at this historically and if you go
  • 6:52 - 6:58
    back just 15, maybe 20 years, outsourcing
    via the internet was a practice that could
  • 6:58 - 7:04
    only be employed by big IT companies,
    mostly from the US, India. Today it is
  • 7:04 - 7:11
    cheap and easy and it can be done by
    anybody. A new service by Amazon called
  • 7:11 - 7:18
    Amazon Key illustrates this rather new
    situation perfectly. With Amazon Key, you
  • 7:18 - 7:24
    can remotely grant access to an apartment
    while you're not at home using the Amazon
  • 7:24 - 7:31
    Key camera, lock and app, you can spy on
    the otherwise completely invisible workers
  • 7:31 - 7:41
    from your smartphone. So this is the lock
    remotely unlocked. So here outsourcing
  • 7:41 - 7:49
    doesn't generate free time. Instead, it is
    born out of necessity. It is marketed as
  • 7:49 - 7:55
    making possible the transformation and a
    liberation from being managed to also
  • 7:55 - 8:01
    being able to manage others and, in this
    case, you're able to deal with them like
  • 8:01 - 8:08
    without ever having to meet them in
    person. So now everybody, not only can but
  • 8:08 - 8:13
    has to, and actually, of course, wants to,
    we all want to use people as software
  • 8:13 - 8:19
    extensions and with that comes also that
    we have to remotely track and rate their
  • 8:19 - 8:28
    performance. So let's recap: Factory
    workers are extending machines with their
  • 8:28 - 8:33
    bodies. Freelancers have escaped the
    factory but have to offer themselves as
  • 8:33 - 8:38
    flexible extensions to the modern media
    assembly line, for example like which I
  • 8:38 - 8:45
    saw on Upwork. So now micro-entrepreneurs
    have to invent their own shops, offering
  • 8:45 - 8:52
    their creativity in the form of little
    packages that are called gigs. You can buy
  • 8:52 - 9:00
    such gigs at a fixed price on platforms
    like Fiverr. For those who do not know
  • 9:00 - 9:07
    Fiverr: Initially each each gig on Fiverr
    was priced at exactly $5 of which the
  • 9:07 - 9:14
    platform kept $1. So some months ago, I
    found a way to directly access all videos
  • 9:14 - 9:19
    uploaded to the platform in real time,
    including every single video that people
  • 9:19 - 9:25
    on Fiverr are producing for their clients.
    So through this crack in the surface, or
  • 9:25 - 9:32
    you could also call it a security problem
    or a privacy issue they have, I could look
  • 9:32 - 9:37
    at the leaked stream of videos and I did
    this for days and weeks and months and I
  • 9:37 - 9:43
    downloaded videos worth of like more than
    100 GB and I was looking for some patterns
  • 9:43 - 9:51
    to understand this marketplace. So to give
    you some ideas about what I saw there: On
  • 9:51 - 9:58
    the platform it is dog-eat-dog. Be the
    best the cheapest, the most creative, the
  • 9:58 - 10:03
    most efficient, be just like a proper
    software extension. Never sleep, work all
  • 10:03 - 10:09
    the time. At the same time, everybody is
    also fighting against the platform's
  • 10:09 - 10:17
    algorithms and clean interfaces that hide
    most gig workers on page 2, 3, 4 and so
  • 10:17 - 10:28
    on. As many gigs offer unrealistically
    short delivery times for creative work, it
  • 10:28 - 10:34
    becomes clear that they themselves use
    bots, generators and templates, simulating
  • 10:34 - 10:41
    creative work and creating yet another
    layer of man-machine complexity. So using
  • 10:41 - 10:47
    automation in order to not be replaced by
    automation. Their biggest selling point
  • 10:47 - 10:55
    seems to be their low cost coupled with a
    truly natural interface: a human being. In
  • 10:55 - 11:01
    contrast, there is another group of people
    consciously offering their bodies often as
  • 11:01 - 11:07
    carriers of messages like screens. So
    here, the fantasy of the universal
  • 11:07 - 11:12
    addressability and availability of all
    nodes manifests itself in the distant and
  • 11:12 - 11:23
    often exotic body that acts as a screen.
    This goes hand in hand with gigs offering
  • 11:23 - 11:38
    personal porn or erotic videos and fetish
    videos. So here, I would summarize: Being
  • 11:38 - 11:44
    a software extension on a hyper-
    competitive platform fosters and demands
  • 11:44 - 11:49
    something that I would call survival
    creativity, that means coming up with
  • 11:49 - 11:56
    whatever it takes to survive in a
    competitive environment. And as I
  • 11:56 - 12:01
    reminder, Fiverr might be an extreme
    example but it exemplifies a development
  • 12:01 - 12:08
    that has become a reality for many already
    and it's not like it's them and us, we are
  • 12:08 - 12:18
    all human software extensions. So far I've
    managed to talk about software without
  • 12:18 - 12:23
    mentioning artificial intelligence even
    once, that's nice I think. Instead I've
  • 12:23 - 12:29
    drawn this a bleak picture of a quasi
    totality of work and exploitation.
  • 12:29 - 12:35
    However, automation and artificial
    intelligence supposedly imply a future
  • 12:35 - 12:43
    without work, right? So in the previous
    examples, platforms, software, artificial
  • 12:43 - 12:48
    intelligence acted as scientific
    management. The Taylorist boss
  • 12:48 - 12:53
    algorithmically distributing and
    modulating human workers as software
  • 12:53 - 13:01
    extensions. Now, one great post work ideas
    is to not only automate the management of
  • 13:01 - 13:09
    working bodies and minds, but instead to
    completely replace all human nodes with AI
  • 13:09 - 13:15
    as well. While I think this is an
    excellent foundation for discussing our
  • 13:15 - 13:22
    society's obsession with work, I would
    also argue this hypothesis is as appealing
  • 13:22 - 13:30
    as it is flawed, unfortunately. So my
    observation is this: AI, artificial
  • 13:30 - 13:35
    intelligence, is an appropriation and a
    possible extrapolation of existing
  • 13:35 - 13:42
    knowledge and skills, yes. And as such it
    might as well do our jobs but it is first
  • 13:42 - 13:49
    and foremost used as a scheme to fragment
    work into tasks that can be done anywhere
  • 13:49 - 13:56
    24/7 and to make this labor invisible.
    What we see here is a piece called
  • 13:56 - 14:02
    "Segmentation Network" which I made last
    year it plays back over 600,000
  • 14:02 - 14:08
    segmentations manually created by
    mechanical crowd workers, Mechanical Turk
  • 14:08 - 14:14
    crowd workers, for Microsoft's cocoa image
    recognition data set. These so-called
  • 14:14 - 14:19
    segmentations are based on photos we have
    uploaded to Flickr and they are used in
  • 14:19 - 14:26
    machine learning, training AI what it can
    see and what not. So you can automate as
  • 14:26 - 14:33
    much as you want but at some point you
    will have to train and especially maintain
  • 14:33 - 14:40
    the machines and the software. So I would
    say AI creates yet another layer of badly
  • 14:40 - 14:49
    or unpaid care and maintenance work which
    is often invisible on purpose. So I would
  • 14:49 - 14:54
    say this needs to change. And I think
    Feminist theory and practices have to lot
  • 14:54 - 14:58
    have a lot to say about this issue but
    that's another talk and I would like to
  • 14:58 - 15:08
    hear it. So here is a point in case and
    maybe a solution: At the end of 2011,
  • 15:08 - 15:13
    while still being students and sharing a
    studio Silvia Larusso, a friend and
  • 15:13 - 15:18
    artist, and me we started to take a
    screenshot of every single captcha that we
  • 15:18 - 15:25
    had to solve while navigating the web. So
    over the years, proving that we are human
  • 15:25 - 15:34
    time and time again, we captured hundreds
    of captchas. This year, we thought about
  • 15:34 - 15:39
    how to make this little thing or these
    like little things, the collection of
  • 15:39 - 15:43
    these things, into something which is as
    valuable and as expensive as possible and
  • 15:43 - 15:49
    so we published the complete collection as
    a series of five handmade Leporello books.
  • 15:49 - 15:56
    So each of these books is one year. And if
    you if you expand all these Leporello,
  • 15:56 - 16:02
    they have a total length of 90 meters
    chronicling five years of micro labor as
  • 16:02 - 16:08
    well as the history of captures. So if you
    look at it, captures started as a
  • 16:08 - 16:15
    technique to merely prevent spam and then
    they kind of morphed into a method for
  • 16:15 - 16:23
    deciphering house numbers and transcribing
    books. And then lately it's become a means
  • 16:23 - 16:31
    of teaching image recognition to AI
    software. While we were collecting these
  • 16:31 - 16:38
    captures, Gabriela Rojas Lozano in 2015
    filed a class action lawsuit against
  • 16:38 - 16:43
    Google and she claimed that Google
    operates a highly profitable transcription
  • 16:43 - 16:49
    business built upon free labor which it
    deceptively and unfairly obtains from
  • 16:49 - 16:57
    unwitting website users. Unfortunately,
    her claims were rejected. So here the
  • 16:57 - 17:01
    judge states that it's like you spent only
    a few seconds on this it cannot be
  • 17:01 - 17:08
    expected to pay for such a small job. So,
    however, her attempt to sue Google was
  • 17:08 - 17:14
    still a success I would say because it led
    to the proof that Google has perfected a
  • 17:14 - 17:20
    magical process in which work is
    transformed into literally nothing. So
  • 17:20 - 17:27
    welcome post-work society! How does this
    magical trick work? It's rather simple.
  • 17:27 - 17:31
    You take a job, let's say transcribing
    books, and you fragment it and you
  • 17:31 - 17:36
    fragment it more and more and more until
    suddenly the job is magically done without
  • 17:36 - 17:42
    anyone having ever worked on it. Because,
    if nobody has to get paid, nobody had to
  • 17:42 - 17:47
    work either, right? Hence, the judge's
    statement is proof that this magic
  • 17:47 - 17:52
    actually works. It gets better. In the end
    Google still ends up being paid even
  • 17:52 - 17:58
    though they have just made the job
    disappear magically. So what I want to
  • 17:58 - 18:05
    suggest now is to seize the means of
    magic. What about this: Fragmenting those
  • 18:05 - 18:11
    platforms that algorithmically manage us
    to such a great degree they simply do not
  • 18:11 - 18:18
    exist anymore? Magically, their job will
    still be done and in the end we get the
  • 18:18 - 18:24
    money. I guess in like a less magical
    version we could call this platform co-
  • 18:24 - 18:30
    operativism. If you do not believe in
    magic then I have two more suggestions
  • 18:30 - 18:38
    about what we, as a software extensions,
    can do. So I will show you two clips from
  • 18:38 - 18:43
    my latest video piece which is called "I
    Will Say Whatever You Want in Front of a
  • 18:43 - 18:49
    Pizza". You can see the full version
    online on this website, all that's gonna
  • 18:49 - 18:53
    be, I think there's going to be a
    screening somewhere here in the in the
  • 18:53 - 18:57
    video lounge. Anyway, I will show you two
    bits even though it's a 12-minute video
  • 18:57 - 19:03
    and just looking at two bits doesn't make
    too much sense still I think they show a
  • 19:03 - 19:08
    point I wanted to make pretty nicely. So
    the video is narrated from the perspective
  • 19:08 - 19:15
    of a cloud worker and this is the
    protagonist:
  • 19:15 - 19:23
    Protagonist: [unintelligeable speech from
    video] Instead, I was working for a pizza
  • 19:23 - 19:30
    delivery company. Brian, your next
    automated pizza delivery is scheduled for
  • 19:30 - 19:36
    Saturday February the 25th at 12:00 p.m..
    To confirm, text yes. To decline, text no.
  • 19:36 - 19:39
    Text help for help.
    Other voice: Thank you, Papa John, you
  • 19:39 - 19:43
    handsomeman. I shall call
    you the Carbs Vixen.
  • 19:43 - 19:47
    Protagonist: We're sorry, we didn't
    understand. Please confirm or decline.
  • 19:47 - 19:51
    Other voice: When I make love I imagine
    you tossing some dough shirtless
  • 19:51 - 19:55
    Protagonist: Dude, our automated system
    isn't set up yet. This is a real person
  • 19:55 - 20:00
    texting you. I make minimum wage, please
    just tell me if you want the pizza
  • 20:06 - 20:13
    S: So, that's a protagonist. To give you
    some background: In 2016, Donald Trump's
  • 20:13 - 20:18
    team hired a Singaporean teenager through
    Fiverr which is the platform I talked
  • 20:18 - 20:24
    about before, and they hired her to
    convert a PowerPoint into a Prezi which is
  • 20:24 - 20:29
    the software I'm using for the video,
    basically outsourcing the Make America
  • 20:29 - 20:34
    Great Again campaign, which is it's a true
    story. laughter in background So at some
  • 20:34 - 20:39
    point in the video, the protagonist who is
    now working and only as a pizza delivery
  • 20:39 - 20:45
    bot but also as a cloud work on Amazon's
    Mechanical Turk platform, preparing data
  • 20:45 - 20:50
    sets for AI, at some point he gets to know
    the Singaporean teenager as a fellow
  • 20:50 - 20:58
    worker. And she has got an idea.
    Protagonist: One day, I found a thread
  • 20:58 - 21:05
    started by her. She talked about the
    political implications of what we did and
  • 21:05 - 21:18
    what kind of things we could do. Of
    course, we on Mechanical Turk had created
  • 21:18 - 21:24
    the data set. It was sad to see how
    depressed many of us actually were.
  • 21:24 - 21:28
    Other voice: Researchers asked 500 workers
    to complete a survey which contained a
  • 21:28 - 21:32
    standard clinical depression survey. 170
    workers agreed to share their Instagram
  • 21:32 - 21:40
    post for the study. Out of those 170
    workers, 70 were clinically depressed. By
  • 21:40 - 21:45
    the way, the depression filter is Inkwell.
    It turns photos to black and white and
  • 21:45 - 21:52
    adds high contrast. Better not to use it.
    Protagonist: But she pointed to something
  • 21:52 - 21:58
    else. It was in our power to manipulate or
    even to change the very core of such
  • 21:58 - 22:14
    mechanisms: our data sets, machines,
    society, you, the future. We started to
  • 22:14 - 22:21
    discuss, to organize and to experiment. Of
    course, we didn't always agree. Was this
  • 22:21 - 22:29
    just for fun, like Easter eggs? Did we
    share a political agenda? In any case,
  • 22:29 - 22:42
    things had to happen secretly. Otherwise,
    it wouldn't work. Teaching Google to
  • 22:42 - 22:48
    identify a photo of an eggplant as British
    singer Chris Meloni served as a proof of
  • 22:48 - 22:54
    concept.
    S: Okay, so the ideas is this: When we are
  • 22:54 - 22:59
    extending software with our bodies and
    minds, we are also extending our reach
  • 22:59 - 23:05
    into the software, and reaching into the
    software, being part of the software, we
  • 23:05 - 23:09
    can start to manipulate these systems that
    govern us and that we have to use to
  • 23:09 - 23:15
    govern others. And once we are plugged, in
    we can manipulate data, we can create new
  • 23:15 - 23:21
    and weird and slow and inefficient
    software from within. So it can be fun,
  • 23:21 - 23:27
    like leaving easter eggs for others to
    find realizing yes, there are actual
  • 23:27 - 23:36
    people inside these systems. Which brings
    me to my third and last thought: Why being
  • 23:36 - 23:44
    a software extension also has some chances
    or possibilities. I will end the talk by
  • 23:44 - 23:49
    talking about Mark Zuckerberg which will
    make me look like a fool anyway. Being a
  • 23:49 - 23:53
    software extension can also offer a new
    aesthetic and a new way of being and I
  • 23:53 - 23:59
    think this video which could be called the
    father of all stupid demos illustrates
  • 23:59 - 24:05
    this in a rather interesting way. Here,
    for whatever reason Zuckerberg is
  • 24:05 - 24:10
    demonstrating Facebook's virtual reality
    by visiting Puerto Rico in the wake of
  • 24:10 - 24:17
    Hurricane Maria. So why do I show this
    video? Contrary to what he had intended,
  • 24:17 - 24:23
    Zuckerberg as a crudely abstracted version
    of himself turned into a software
  • 24:23 - 24:29
    extension, detaches and dissociates
    himself from the real world. This is what
  • 24:29 - 24:37
    I like. Because I think and you will agree
    with me: Software is not perfect. It's
  • 24:37 - 24:42
    full of bugs, it often behaves in
    unexpected and weird and glitchy ways,
  • 24:42 - 24:51
    doing stupid things often and like often
    over and over again, in an infinite loop.
  • 24:51 - 24:57
    Therefore, embracing the weird and
    abstract asthetic of a of a human as a
  • 24:57 - 25:03
    software extension could actually allow us
    to detach ourselves from circumstances and
  • 25:03 - 25:10
    which we are required to be our best
    working selves all the time. So being like
  • 25:10 - 25:15
    available all the time, addressable,
    programmable, to update ourselves all the
  • 25:15 - 25:21
    time. We could like use being a software
    extension, the aesthetic of being a
  • 25:21 - 25:28
    software extension, as a mask behind which
    we can hide, pretending to be a bot.
  • 25:28 - 25:31
    That's the idea. Thank you.
  • 25:31 - 25:41
    applause
  • 25:41 - 25:46
    Herald: Thank you, Sebastian, indeed, for
    making a bleak future look a little bit
  • 25:46 - 25:50
    more bright. I think we have about five
    minutes for very short Q&A, otherwise, he
  • 25:50 - 25:56
    said you have a Twitter account and you
    can also meet Sebastian next to the stage.
  • 25:56 - 26:02
    So are there questions, maybe from the
    Signal Angel? None? Okay, then just
  • 26:02 - 26:05
    another round of applause for Sebastian!
    Sebastian: Thank you.
  • 26:05 - 26:08

    applause
  • 26:08 - 26:15
    postroll music
  • 26:15 - 26:29
    subtitles created by c3subtitles.de
    in the year 2018
Title:
34C3 - Humans as software extensions
Description:

more » « less
Video Language:
English
Duration:
26:30

English subtitles

Revisions Compare revisions