I’m at the Scottish AI Summit and connecting with it better than last year. It’s in part a shift in practise, what seemed aspirational and abstract for policymakers in the past is coming alive in their heads for them now.
The Scottish AI Alliance has a new Chair, Catriona Campbell with a background in product design. Her keynote featured a “future-back” exercise, positing different AI futures - Do Nothing, Living in a Simulation, Escape to Mars, Neo-Luddism etc - decide where we want to be, then wind back and take next steps to design useful products.
I sat there thinking “you’ve missed at least one out - Hunger Games”. In this scenario, in the wake of environmental catastrophe, technology and knowledge advance focuses on a small, well-insulated elite, while the people in the outlying districts are repressed, denied the possible, left to starve and freeze when plenty is available, and periodically harvested.
AI cynicism infects all my perspective. While watching Toju Duke, who’s taken on the Responsible AI remit at Google, as she outlined six principles of a responsible AI framework - Data Ethics, Fairness, Explainability, Human Rights, Safety, Privacy & Security - I thought, how are each of these framed in a way that advances the tenets of late stage capitalism? How is the language being used to service corporate profit outcomes? What’s in the gaps, that these focuses leave behind?
A few years ago it was possible to read most discourse about “AI ethics” internally substitute “ethics of capitalism”, and yield at least as much, if not more sense from the same statement.
Now you can do the same with “product”, in the expectation that any product will be in some way delivered, designed, or packaged in an AI-assisted way. A member of the panel on the role of standards in trustworthiness said “we’re only concerned about trustworthiness when there’s harm”. But almost everything is weaponisable; almost everything is productisable. When something has a price, then you are buying the expectation of safety. In our system, if you’re not paying, everything is an unsafe product.
Try it! Try internally substituting “product” for “AI” in every talk title and opinion piece. See if it makes you float or sink, internally. Does it now make a crushing sense, or does it jar in a way that tells you our trajectory is off?
In some cases it makes no difference, just shows the focus - “Products in the Scottish Fintech Sector”. In some cases it makes you look harder - “The potential for products to disrupt education”. In a few cases it looks like the alignment of incentives just isn’t there - Products and racial justice? Climate products? And in a few it makes me smile at what it shows - “Encountering Products through Art”.
I would like to encounter fewer products. I would like to experience less design. I think “replace AI for product” is a neat trick to catch the values mismatch - between what government funders claim their desired outcomes are, how their networks of fundees believe they need to act in order to reach those outcomes, and the everyday needs of all of us unwilling, or unwitting participants in the Hunger Games.
While suffering chronic AI angst, I can’t indulge in despair. If the arc of the moral universe bends towards justice, then there must be a future-back I can believe in. It’s called “Jerusalem”, following Blake’s “dark, satanic mills” - the untapped possible smothered by excess of tech acceleration, “Jerusalem bound in chains, in the Dens of Babylon”.
What does it look like? It looks like building material carefully prepared from local substances - each built structure uniquely reflecting the geology and fauna in the 50 metres surrounding it. It looks like the skills and knowledge needed to find and build that structure, then calmly live in it. It looks like planning for a constant regrowth and an elegant decay. It feels like body-consciousness extended to a scale that I suppose is Gaia, but with readily-available snacks. Like co-creation but without design; visible, mobile, and welcoming. An easier place.
Comments powered by Disqus.