sc-sc-scatterbrained

nora-khan-notes.txt

Nora Khan, 11/2/2017

trevor paglen & ian cheng & co. - revealing inherent bias in image recognition algorithms
    viz. face sentiment & gender scores - what is a woman, what is a positive and negative face?
    viz. cheng's de-god-ifying the gaze of drones

white-color crime map :)

the idea that the body and face predict who you are and what you will do
    the new phrenology?
    the idea that a massive amount of input data somehow legitimizes this with a false sense of science (and even removes the ethical quandary - it's the computer that's making the judgment calls, right?)
    how does this feed back into how we view ourselves?
    and note how this extends to other data: your credit score, your shopping patterns, ...

consider the dwindling utopian vision around the early days of tech/networks/etc.
    whole earth catalog - stewart brand
    fuller/utopias/communes
    merry pranksters
    psychedelic drugs - what the dormouse said
    declaration of the independence of cyberspace

>> commodification/normalization of the image of the globe. interesting... what did we lose?

what would it mean to build a geodesic dome in cyberspace? lol

on technologically imposed limits to expression & flattened identity
    demands user as static object/identity/brand (vomit)
    interface design favors utility & affinity & spectacle
        what would the alternative be? interesting to consider a new language of interface
    relentlessly anthropomorphized AI
        but it can embody emotions without responding
    attempt to achieve unearned trust
        convincing veneer that hides biases and whatnot

artists intervening:
    sondra perry "graft and ash for a three monitor workstation" - holy fuck
        questions "right" body and erasure – "neutral" is never neutral
    ian cheng
        simulation-as-art, agent-behavior-as-medium
        no end state
        what does it mean that we want to narrate it? this is something like trying to understand the output of a neural network. and sometimes things happen that are *just not human*. but it seems we're pretty good at storytelling in a situation we can't understand.
        viz. david o'reilly's mountain
        >> what's the difference between a simulation and a game?
        interesting on a personal level that he seems pretty uninterested in the fidelity & aesthetics of his projects - a hell of a lot of clipping and Z-fighting. kind of inspiring in a way that he can be lauded and shown at PS1 despite that

TCP over carrier pigeon lol

how we talk about AI governs how it's conceived. we use a lot of metaphors:
    we should rework & investigate these closely given the variety of AI that exists or could exist
    mind-as-computer, sea of data
    black boxes
        nvidia intending to "open the box" behind its self-drving cars
        this narrative makes us pretty irrelevant – we set things up but then we're out of the picture, and we can't even easily correct mistakes unless we go out of our way to engineer that in
        >> ask anurag more about DARPA's Explainable AI project
        >> interesting comparison here to our own brains... WHAT DOES IT ALL MEAN
        what is behind the desire for explainability? is it entirely to avoid a PR disaster? to have something/someone to blame? or so we feel somehow related to the AI? somehow superior to it? or at least not inferior? are we scared of it being too alien? what similar demands do we put on humans?
            and like... what form would an AI's explanation even take? this feels like the "explanatory gap" in philosophy - what good is an explanation if it's not satisfying?


khan suggests that we should put a hold on further AI development until we can decide upon & engineer a suitable standard for transparency & understanding
    something something "an AI that is worthy of us"
    funny to think about the idea of a society putting something on hold. hahaha we don't do that