fields

What Would It Take to Read the Label?

February’s Futures Collider at RFI put three provocations in a room and asked people to act out the consequences. Seven minutes each. Go…

Vanessa Johnston opened from law. AI is dismantling frameworks we’ve spent centuries building: copyright, IP, privacy, because all of them depend on one thing we can no longer reliably do: attribute. Identify an author. Locate a thought. Trace an idea back to a person. AI doesn’t destroy these protections deliberately. It just makes them structurally inoperable.

Tim Fawns followed with a metaphor I haven’t stopped thinking about. AI is a contaminant, already in the reservoir. It’s in the treatment systems. You can try to find fresh water, but it’s homogenising the supply, filtering out flavour, taking out nutrients we can’t fully name. His worry isn’t that we reject it. It’s that we get used to it. That Melbourne water starts tasting fine.

Then Ascelin Gordon made the metaphor uncomfortably material. Those data centres keeping AI running use actual water, enormous, often unsustainably sourced quantities of it for cooling. Plus energy. Plus e-waste cycling every five to seven years. And the impact is almost impossible to calculate: we don’t know which data centre we’re hitting, where it sits, how it’s powered, or what the local aquifer can bear.

Three speakers. Three registers. One thread: the cost is real, but it’s been made illegible. We can’t attribute the thought. We can’t taste what’s been filtered out. We can’t read the energy label on our query.

The challenge I put to the room: imagine RFI in 2035, looking back. Did we default to convenience? What did we resist on principle? Groups of RMIT staff, students and members of the public had 45 minutes, then performed their answers, on their feet, two minutes, no posters – only props. It was genuinely brilliant. We had four people walk out – “I’m not doing a skit”!

But here’s the question I’m sitting with: if the cost of AI – legal, cognitive, environmental – is structurally designed to stay invisible, what would it actually take to make it legible? And who has an interest in that never happening?

Thanks to Kirsten Black Hannah Bornsztejn (she/her) Tiobstia Alemu Dale Leszczynski Toni Jones Alan Hill Diane Brown Nicholas Mau

hashtag#RegenerativeFutures hashtag#FuturesCollider hashtag#AIethics hashtag#DesignEducation hashtag#RMIT