There are days where I’m on dad duties and everything is a colossal juggle: home schooling my eldest, toilet training a toddler and the never-ending pile of laundry. Instinctively my inner engineer retreats, dragging me into a daydream. Often, I think about the times where I have long uninterrupted sessions to tap away at a computer. Then there are the other times where I dream about the things I wish I had more time to build. Eventually, I start thinking about artificial general intelligence (AGI): the idea that one day we will have machines that can understand and learn at the same intellectual level as a human.
Usually these thoughts spring up somewhere between singing, ‘Poops in the toilet go splosh splosh splosh’ (to the tune of ‘The Wheels on the Bus’), and helping my eldest with her spelling practice. I often can’t help but giggle. Maybe I read too much Asimov as a kid or maybe it’s this 1965 quote from NASA:
Man is the lowest-cost, 150-pound, nonlinear, all-purpose computer system which can be mass-produced by unskilled labour.
I guess I start to wonder how much time the people pursuing AGI have invested in raising and teaching children. Because the raising kids thing is hard, certainly the most difficult undertaking I have ever embarked upon. Fostering constructive responses from my children given a universe of possibilities and situations, while simultaneously discouraging unhelpful behaviours? DiFficult. Yes. With a single capital F.
Maybe I’m being naïve, but often it feels like there is this myth, that with the right swish of mathematics all this child-rearing complexity will disappear. That by taking a bunch of numbers and combining them together in just the right way, we can somehow sidestep all the complexities inherent in raising our organic or artificial children. But to me, that sounds a lot like wishful thinking, similar to Mark Zuckerberg’s wishes that more algorithms will somehow reverse the algorithmic erosion of our social fabric.
I think that’s one of the reasons why I love the work of American sculptor, Tom Sachs. His practice is unapologetically analogue and draws much inspiration from the American era that ended in the 70’s – that period just before the arrival of cheap, ubiquitous computing power.
While hosting a recent office-hours, Sachs threw away this one-liner:
Look at my Keynote skills. I mean, man, this boomer is a hacker…
It was a hilarious self-deprecating joke. In a lot of ways Sachs’ work is the antithesis of our current digital era and he isn’t shy at admitting that computers aren’t his thing. But to say Sachs knows nothing of algorithms or programming would be a mistake.
Tom has built his own fully functioning McDonalds restaurant, a Swiss Passport Office and spent decades crafting his own space program. The audience is often invited to participate in these large-scale works.
With all the precision of a NASA pre-flight checklist, or the countless hours of refinement in a McDonalds operating manual, Tom uses a mixture of video and instructional drawings to indoctrinate the audience so they can perform their role within the work.
In many ways this indoctrination, like any other form of education, is how Sachs programs the audience to be competent ‘bots’ during a demonstration of his sculpture. Tom uses similar techniques when training the studio employees who engineer and manufacture all his work. I know it sounds a bit cultish, and that’s certainly an idea that Tom leans into:
This place is a cult, and I mean that in the scariest, most Manson-family kind of way, in that we’re totally committed to this way of life.
NASA checklists and McDonalds operating manuals? Stitched together with a strict, cult style dogma? Without any visual reference, it’s an approach that sounds like the assembly line of a factory, rather than a team of artisans wielding their skills to explore the frontiers of human culture.
Therein lies the strength of Tom’s programming approach: His studio is a team of artisans building unique ‘one-offs’; a full-sized free-standing Apollo Lunar Excursion Module or a 1:25 scale model of Le Corbusier’s Unite d’ Habitation.
The best glimpse of how Tom resolves this factory vs artisan paradox is a recent design workshop that he ran in collaboration with MIT Architecture. Instruction was a couple of minutes of video, some rules and a visual recipe:
By Friday, Aug 14th, in remote collaboration with my studio team, you will build your own 3v geodesic dome from a material, and at a scale appropriate to your current environment – wherever and however you find yourself. Try to avoid buying things, make do with what you have. In addition to building a Dome, you will create two communication materials.
A 1-Pager:
- Communicate what your dome is about
- Must have Name, Date and a Title for Your Dome
- 8.5 x 11 sheet horizontal – folded in half – be mindful of the gutter.
A 1-min Film:
- In that minute you need your name, the title of your dome, and to convey what’s important about it.
Note – This is not a class about Dome geometry, it’s about bricolage and construction. Don’t get lost in the math. If you have a reason you want to use another Dome geometry we must talk about it at the first session. Don’t fuck around.
People from all around the world got involved and executed the programming Tom had set before them. But the resulting domes looked nothing like a mass-produced product that had rolled off an assembly line. 1-Pagers from of all these domes were printed and bound into a zine called ‘How to Build a Geodesic Dome’. Here are a few pages:
*
At some point, long before we reach artificial general intelligence, the way we raise technology will need to change. If one of our artificial kids becomes at least as capable as one of our organic kids, then the same amount of labour and effort will need to be expended in creating and raising them. And long before that time, all of our artificial children will have special needs, requiring specialised guidance and nurturing within our society. Who or what will have the responsibility for meeting the special needs of our artificial children?
It’s becoming pretty clear that the artificial infants who decide what we see on social media already need specialised care. Who or what protects these artificial infants from the malicious actors who set out to harm them and distort what we see? Another infant? Or something completely different?
If you look at places like the AI Now Institute and 3AI, the work on improving how we teach and integrate these technologies into our society has already begun. My hunch is that the way we raise our children (both artificial and organic) will slowly converge. As a software developer, and a home-schooling stay-at-home dad, that’s a massive head spin. Will we end up training AI with less of sudoku-style grids of numbers and more like something that resembles formal schooling? Or will school end up more like sudoku-style grids of numbers? I don’t have the answers, but I do know that moving forward we will need to resolve the cognitive dissonance inherent in this evolution: like the way Tom Sachs juggles the factory production vs artisan duality in programming participants within his community.
Comments:
You can join the conversation on Twitter or Instagram
Become a Patreon to get early and behind-the-scenes access along with email notifications for each new post.
Hi! Subconsciously you already know this, but let's make it obvious. Hopefully this article was helpful. You might also find yourself following a link to Amazon to learn more about parts or equipment. If you end up placing an order, I make a couple of dollarydoos. We aren't talking a rapper lifestyle of supercars and yachts, but it does help pay for the stuff you see here. So to everyone that supports this place - thank you.