March 18, 2023 Externalities Kodak Film Timnit Gebru AI Katy Kelleher Swimming Pollution The Memory Palace
All the way back in 2007, I was visiting friends in Rochester, NY during the summer. It was a hot day, and we went to the Genesee River that flows through the city to take a swim—unaware that the water was contaminated.
Rochester is the home of the Eastman Kodak Company, one of the world’s largest manufacturers of photographic films and papers. For decades, the water Kodak used during production was processed in a treatment plant and then released into the river. What the company didn’t know (or didn’t want to admit) is that the silver particles coating each film, which make it sensible to light, couldn’t be fully removed from the water—and ended up as toxic residue in the riverbed.1
Today, I still love film photography, how it slows down the photographic process, forces creativity, and creates a “more authentic” look—but I also find it hard to ignore the contamination caused by the production and development of film. In other words: My love for film photography is somewhat tinged by thoughts of that riverbed in Rochester.
I keep returning to the topic of externalities, that dark flipside of seemingly any impressive technology. Deliberately taking the human cost aside, the pure environmental impact is staggering: Starting with the rare earth metals in phones and electric cars to the electricity use of cryptocurrency or the wastefulness of online shopping, many technological breakthroughs or feats have unintended consequences.
Katy Kelleher’s book The Ugly History of Beautiful Things hasn’t come out yet, but the description already resonates with me: “(…) whenever you find something unbearably beautiful, look closer, and you’ll inevitably find a shadow of decay lurking underneath.” Those toxic silver halides, it turns out, can be found in practically everything.
What troubles me about externalities is how easy it is to miss or outright ignore them: When a problem is removed from sight, it can take years or decades to discover the consequences—when it might already be too late.2 Electric cars are great for removing emissions in a city, but the electricity powering them still needs to be produced elsewhere—making all promises about “zero tailpipe emissions” ring hollow.3
Currently, there’s lots of attention on AI and machine learning, with all the largest technology companies racing to create applications for this new technology. Unfortunately, the externalities loom large once more: This 2020 article from MIT Technology Review analyzes a controversial paper from Google engineer Timnit Gebru, who points out the resources required for training a model—apparently around the same amount of CO₂ “as a round-trip flight between New York City and San Francisco”.
Gebru’s draft paper points out that the sheer resources required to build and sustain such large AI models means they tend to benefit wealthy organizations, while climate change hits marginalized communities hardest. “It is past time for researchers to prioritize energy efficiency and cost to reduce negative environmental impact and inequitable access to resources,” they write.
They would go on to reduce the amount of silver halides used. “Before the 1960s,” Rebecca Solnit writes in A Field Guide To Getting Lost, “light and air themselves seem to have had an almost undersea depth and luminosity, in which skin glowed opalescently and everything seemed to have a faint aura slaughtered by the newer black-and-white films made with less silver in the emulsion.”↩︎
This is most devastatingly illustrated by an episode from the Memory Palace called Butterflies.↩︎
It’s also a question of equality: Norway is one of the world’s richest countries due to oil and gas exports and uses its wealth to subsidize electric vehicles, effectively shifting the emissions abroad.↩︎