#WrightNews 29/08–04/09

Making gods, making robots

Photo by Edilson Borges on Unsplash

Genesis 1:27: “So God created man in his own image, in the image of God he created him; male and female he created them.”

A normate-satisfaction machine

In Genesis, God creates man and woman in his own image, but really God is constructed in man’s image (and I mean man deliberately here), but really man has created God in his own image, and has continued to do so.  “If man is to have contentment in God”, wrote Ludwig Feuerbach

¹

, “he must find himself in God.”  God is a construction that best fits a human need, but paradigm shifts in human need must map to paradigm shifts in what gods we want to create.

In The Age of Early Divinity, Venkatesh Rao riffs on the adage “First we make our tools; thereafter our tools make us” by positing that we then make our gods; thereafter, we become our gods.  We are, he posits, in the age of “early divinity”; that is, we are becoming gods, but we are not very good at it.  Indeed, the technology we’re building now is being built with the same mindset with which we built our gods; that is, to serve ourselves and what we think we desire.  But this act of building has to build upon something, and that something is going to be pre-conceived notions of the human: what a human being is, how a human being thinks, how intelligent a human being is, and how acceptably one fits into acceptable models of humanity.  Michele Elam, in “Signs Taken for Wonders: AI, Art & the Matter of Race”, a fascinating paper that explores the consequences of AI from a racio-cultural perspective, warns that technology is being built upon very narrow, Enlightenment-influenced models of the human.  Put more specifically, these models are built on the belief that there is a universal human subject, and therefore universally-human attributes, but this is from the perspective of what Rosemary Garland-Thomson calls “the normate”

²

: “the imagined everyman whose self-determination, independence, rational thinking ability, and physical sturdiness makes American democracy philosophically possible.”  It’s not far from Toni Morrison’s assertion that “in [the USA] American means white. Everybody else has to hyphenate.”  Elam quotes Lorraine Hansberry, whose character Tshembe Matoseh in Les Blancs

³

says that:

Race — racism — is a device. No More. No less. It explains nothing at all. It is simply a means. An invention to justify the rule of some men over others. But it also has consequences; once invented it takes on a life, a reality of its own… And it is pointless to pretend that it doesn't exist merely because it is a lie!

Race as device — an epistemology, not an axiom — fits with Paul Preciado’s similar insistence about gender relations, in that binary ideas of sex and gender are epistemologies that have not always existed, but have instead grown from the prerogatives of power structures whose reinforcement of such categories and taxonomies shored up their own power and dominance.

Everything is made in man’s image — and I am here deliberately using man as pseudo-universal — tools, gods, structures, ideas.  But this narrow image excludes as it includes, and those who are excluded — people of colour, differently-abled people, transgender people, the neurodivergent — dissolve and become invisible at the point of their exclusion.  As Spivak writes of women outside the mode of production narrative, so do these excluded “efface as they disclose.”

⁴

AI will extend this effacement, is the crux of Elam’s paper, but it doesn’t have to be so.  Though our future is far from certain, there are spaces being made.  Among other examples, she presents Rashaad Newsome’s AI Being 2.0 — a digital ‘griot’ designed to be free from bias and instead modelled on compassion, the very antithesis of the normate on which current technological models are built.  It’s a vital counterpoint to something like GPT-3, an impressive language model that uses deep learning to produce putatively and convincingly human responses.  It’s problematic, though: “despite its new tricks, GPT-3 is still prone to spewing hateful sexist and racist language” according to MIT.  Why it does this is not clear, but it seems to be building on something already there.  It’s modelled on human beings.  We make our tools; thereafter our tools make us.

AI is fascinating, then, because it’s really about human beings, an idea not lost on James Manyika in “Getting AI Right”, his introduction to the issue of Daedalus which also features Elam’s paper.  There is something performatively robotic about expectations of ‘normate’ behaviour.  It’s a space into which neurodivergent people have to cross into, but they must camouflage if they do.  My own experience of such is extensive and current.  And although such camouflaging strategies have helped me make space for myself in a normate, neurotypical world, it has come at great cost.  I burn out, melt down, fade away.  It takes a physical and a cognitive toll.  All so I can be, if I may riff on Deleuze and Guattari’s concept of the human as a “desiring-machine”

⁵

, I become a normate-satisfaction-machine.

I’m really lucky, because I am privileged in ways that means I don’t “efface as I disclose.”  I am aware that my perspective and my suffering pales in comparison to many others who are defined so completely by lack, by the cumulative and constant harm of spoken and unspoken deficit narratives.  There is a double-empathy problem.  The double-empathy problem was coined by scholar of autism Damien Milton to express the real problem that autism is defined by neurotypicals.  The truth about autism can only come from autistic people.  And so it goes for everyone.  There must be space for all voices.  It’s why we still need the arts: how better to express oneself than via art?  I like Elam’s assertion that there needs to be a symbiotic relationship between programmers and artists if the AI of the future is going to be reflective of compassionate human beings.  I write these essays in the hope that I might be some part of it, in my own very atomic way.

Why AI?

Why am I reading and writing about AI?  It’s originally because I have been writing a novel about advanced AI for a while now, but I’ve been stuck, so I thought research might help.  As is usually the case with me, I have wound up rabbitholing away from my original intent, so I figured I do something with what was buzzing around in my head and post it in this newsletter.  I am hoping to incorporate some of these ideas into the text.  However …

What’s happening?

I’m back to teaching tomorrow, which will inevitably slow everything down (and I’m pretty slow to begin with).  I struggle a bit with transitions, especially between discrete and very different routines (home vs school), so I know I’ll need to take care over the next few days. I have to work out how to mask, though this always throws me off, because masking too completely is a) exhausting and therefore impossible and b) makes me feel that my default, ‘real’ self is unacceptable.  So it’s a fine balance, and a specially-constructed mask that’s required.  And this makes me think about autistic and otherwise neurodivergent teachers going back to school who might be weaving their masks right now.

Alex W. 🦅🐍 @curtaindsleep

Spare a thought for your autistic teacher colleagues, #edutwitter, who might have taken ages to get used to being off and now have to get used to going back to work. If they’re teaching they’re likely masking their autism. It’s the trade-off.

So, finally, because it’s back-to-school tomorrow, I shall end this newsletter here.  As always, get in touch on Twitter @curtaindsleep, and do look after yourselves.

Until next week,

Alex

1. The Essence of Christianity, 1841

2. Extraordinary Bodies, 1997

3. 1970

4. "Can the Subaltern Speak?", 1988

5. Anti-Oedipus, 1972