User:Cristinac/Conference

http://constantvzw.org/site/Discrimination-Big-Data.html Brussels C

Pattern writing coach/missing, suspicious, broken, impossible? data take the last line of the text and be judged

Antoinette Rouvroy

lawyer, to study and to see the ways that new techniques have shifted the way we relate to the world/governing/producing/subjectivizing

algorithmic governing symptom of a cry for representation=> big data. there is nothing left to represent patterns are new in the sense that they don’t function on visual epistemics not ethnic, not classist ..can it be.. datamining techs rely on different statistics=>p.o.l.i.t.i.c.s. categories are already there,

Geoffrey Bowker era of the database big data tends to decategorize no national idea of what it means to belong zizek.the fragile absolute: or why the christian legacy is worth fighting for. loss pf categories. loss of class. replaced by subsets. instead of taking data, to think in terms of what we pay attention to. dominant visibility. if you’re not captured by big data you become invisible. oppressive/discrimination eugenics "closing the temporal loop” to perform a normative society

Solon Barocas discrimination-socially salient groups in society that are targeted for unjustified reason escaping the process worry&enthusiasm type 1 or type 2 error?? premise on the belief that we can make strong statements based on these models accuracy. labelling as objective

A: words denote a transformation in the field of statistics The Man Without Qualities-interesting description of statistics one excludes the points of data that are at the extremities; big numbers exhaustivity. everything that can be described singular can be taken into account what is the most far removed from reality normalcy is avoided in big data conventions of equivalents business of statistics/could be targets for public contestation/contestability data mining escapes these burdens The theory of the normal man theory-Adolphe Quetelet immanence 60s immanence? over transcendence systems of preemption. algorithmic governing doesn’t target the present/individuals, but the virtual, the things that are not already there not emancipatory

G: is a historic who looks for continuity. deviant behaviour at airports. v strict. the moral categories that are being used; ex: deviant. data analytics instead of big data. name is questionable drawing away from panopticon, heading into oligopticon (Bruno Latour)

A: breaking the normal pattern of your profile is impossible, it’s just an occasion to prove the machine right. the more you disobey the more it works: machine training Walter Benjamin: bourgeois houses full of kitsch. concern for public anonymity, the public needs to be sure they still exist……………………….. the public house has been replaced by Facebook, twitter etc. it’s not exhibitionism it’s the need for proof of existence

G: The D/Fiction by Design, algorithms to evacuate interiority (silicon valley)

S: big data-marketing speak. the tool to make sense of the data machine learning is the response to what was the failed attempt to develop AI when we mention the duality good/bad worker, the data sets have already been defined category is indebted to the humans at the outset the attraction of this technique is that they will reveal unintuitive information ex: data mining for terrorists normative force of abnormal if you are part of the minority, you will be much more difficult to assess=different assessment system? data as justification for dismantling the responsibility of the state. ex: the uk. sending a letter letting people know that they are using more energy. shifting responsibility to the user from the government

G: behavioural paternalism. still the paternalist state exists but http://www.amazon.com/Save-Everything-Click-Here-Technological/dp/1610393708

q from public G: ex: Pandora. radio stations being created without the need for genres Big data will not go away even if it won’t prove useful

A: performative. even if it has false predictions, it will still expand on the notions that we have in the future traces your future trajectory Ryan Calo http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2309703 the goal is not to make you buy things but collect information. when all you crave cake? at that time of the day you are more susceptible it works in such a way that we are not authors of our own desires.

q from the public: distinction for supervised/unsupervised learning. are you looking at a relational database of a configured one? example: a professor correcting test papers. objectivity drops etc. moving from considering we are using them/they are using us=>interacting with the machine

q from the public: the public has grown accustomed to relying on algorithms to simplify their life take away the choices. they think they will be always in the majority. learning generically to discriminate.

G: computer scientists-priers of our time. democratising the process

S: deep learning. unsupervised learning. requires very little intervention on the part of the machine. breakthrough in speech recognition and object recognition.

A: accountability for using an automatic system of recommendation. automatic systems tend to let us believe that they dispense us from making decisions and errors (there’s no error and no fault, everything has been forgiven in advance). ex: automated decision release criminals. we don’t want to know about the causes anymore.

Femke: relationship with data mining. conversation with an algorithm different from training data the annotator. not just talk back to the software but have the possibility to access the moment the judgement was taken.

S: for these techniques to have some sort of validity. if the background conditions that were taken will remain the same. don’t become seduced by discussion of technique and not take it as a given

A: government by facts. The fact has not by itself any vale and even as soon as it exists as fact it means it carries with itself its conditions revival of positivism. presents the background as smth inevitable that we cannot change. the natural landscape and environment.tyranny of majority https://en.wikipedia.org/wiki/Tyranny_of_the_majority what is lost here is the idea that pretended objectivity is not justice and to contest the categorisation

S: what about solidarity? traditional legal systems still can apply here. distributional issue across all groups. data and other techniques will be instrumental in doing so.

q from the public: you usually think in categories. all that will be gone. the moment of the decision still carries authority. derrida. decision has to be derived from knowledge. regime of data behaviourist. Max Weber Theory of Bureaucracy. http://kalyan-city.blogspot.com/2011/04/bureaucratic-theory-of-management-by.html

q from public: who to sue. F: who to shout at. (sexist slide in presentation)

G: patterns of using categories in human behaviour. “i am now inhabiting a system. i have been created by a highly distributed system donna harroway, a cyborg manifesto https://en.wikipedia.org/wiki/A_Cyborg_Manifesto

S: ex ssbkyh computers seeing faces where they shouldn’t. to see the images as the way human parse the world and the the other as a computer error symptoms we don’t have direct access to. accountability: google is learning the prejudice exhibited by its users.

A: refusal of representation, mediation. uncertainty gives value. https://en.wikipedia.org/wiki/Writing_Degree_Zero http://plato.stanford.edu/entries/qt-uncertainty/ https://en.wikipedia.org/wiki/Uncertainty_principle

pattern code “don’t rely on unreliable sources"