Future #53: Synthetic people in real spaces
Our ability to create synthetic media is only getting more uncanny, with simulated voices and images that are increasingly indistinguishable from the real thing. But while the media we create may be less and less constrained by physical reality, many other experiences are deeply tied to the spaces in which they happen. From the pandemic transforming our urban spaces to Cuba’s revolutionaries grappling with the country’s tenuous relationship with the internet, our physical reality still shapes us in powerful ways. This week, we look at these issues, as well as the meta-question of what it means to report on the complexity of our current reality and its hyperobjects.
— Alexis & Matt
1: “Do not simulate” & post-mortem identity
Twitter blew up over a difficult, nuanced issue last week, as it was revealed that the new documentary “Roadrunner” used a reconstruction of Anthony Bourdain’s voice to say words he’d written but never said. The director, Morgan Neville, defended the choice, saying he believed that he had the blessing of Bourdain’s family and ex-wife. (Ottavia Busia challenged this in a tweet.)
Helen Rosner did some legwork and wrote a thoughtful summary of the controversy. In it she points out that this all feels “creepy” for a whole host of reasons. Beyond the issue of consent, which is still rather murky here, the filmmakers did not indicate that the clip was synthetic speech. This could raise doubt in a viewer’s mind that any part of the documentary could have been manipulated. And while other films (and even commercials) have simulated performances from dead performers, there may be more complicated reactions when a simulation is representing the person themself, rather than a character (á la Carrie Fisher in Star Wars).
The issues this raises are clear. First, consent to be posthumously simulated shouldn’t be assumed. This isn’t just relevant to celebrities; Microsoft has patented a chatbot that can mimic a dead person’s speech patterns after processing their online posts. It may soon become necessary to express your consent, or lack thereof, in wills and other legal orders. Second, disclosure of some kind seems necessary to maintain trust between creator and viewer. Designs could be subtle and specific to the medium — Rosner discusses various forms of “creative signaling” that can make synthetic media clear without the interruption of explicit disclosure. Either way, more transparency is required to honor both the deceased and the work’s audience.
→ The ethics of a deepfake Anthony Bourdain voice | The New Yorker
2: The call is coming from inside the neural net
While the Bourdain documentary is a complicated ethical issue, this one is not. A father gets a call from an unknown number; the voice of his son, clearly agitated, comes through. He’s in Mexico, he’s in jail, he’s only got a few seconds to talk, and he’s being accused of something he didn’t do. The son gives a number of a local lawyer who is helping, and before the father can ask any follow-ups, the line goes dead.
If you’re reading this you’ve already guessed the twist: the son was fine. Someone had simulated what his voice would sound like under duress and was attempting a classic phishing scam. What’s less well-understood is how; most voice simulators require many samples to sound authentic, and those samples typically would not include gasps, heavy breathing, or other signals of stress. This podcast from MIT Tech Review (with a full transcript, if you prefer to read) goes deeper into the how and the why of it all.
The moral of the story? If your mother calls to say she loves you, check it out.
→ AI finds its voice | MIT Technology Review
3: Reporting on hyperobjects
“Our 21st century existence is characterized by the repeated confrontation with sprawling, complex, even existential problems without straightforward or easily achievable solutions. Theorist Timothy Morton calls the larger issues undergirding these problems “hyperobjects,” a concept so all-encompassing that it resists specific description.”
Charlie Warzel makes the newsletter two issues in a row, with this must-read about journalism in our current age of [waves hands vaguely about]. He describes the inherent tension of trying to responsibly report on events that are incredibly complex and highly uncertain. Do you write about the voting rights bill being voted down, or the larger thread of our unraveling democracy? Do you report just-the-facts about recent wildfires, or do you point out how this is one more signal of our worsening climate emergency? What is a proportional response to unprecedented events?
This inherent tension in creating journalism for the current moment reflects the broader societal crisis we’re having. We need to communicate the narrative of what’s happening in a way that engages with complexity but helps synthesize it, that faces potentially bleak futures yet gives us ways to act, and that helps us truly understand the many hyperobjects that make up our current reality.
→ We are not ready | Charlie Warzel
4: Temporal zoning in urban design
Pandemics in history have often brought notable changes in urban design, from tenement reform to waste management to modernist architecture. Even the overly enthusiastic steam radiators common in New York City were developed as a response to the 1918 flu pandemic, in order to allow windows to remain open during cold winters. In response to Covid-19, urban dwellers have already seen a host of changes, though we have yet to see which ones will become permanent. One of the more common responses to the danger of indoor gatherings has been some form of “open streets”, where areas that are usually used by cars are closed to traffic for certain hours or days of the week in order to provide more space for outdoor recreation. In this New York Times opinion piece, Sara Hendren dives deep into this idea of “temporal zoning” as a means for making cities more livable. She argues that by using “time as the sculpting tool”, we can make cities that are much more flexible, livable, and welcoming to diverse needs without the costs associated with redesigning the built environment in more permanent ways.
→ The simplest tool for improving cities is also free | The New York Times
5: Protest without internet
We often comment that what we perceive as “the internet” is actually multiple internets, fragmented by platforms, infrastructure, and nation-states. Nowhere is this more apparent than in Cuba, a country whose internet is unrecognizable to much of the rest of the world. Cell phones were illegal until 2008, and what internet access exists today is extremely limited and exorbitantly expensive. The more common way that Cubans get access is through “el paquete” — a week’s worth of internet content downloaded onto a USB drive and viewed offline on outdated laptops. For a number of years, Cubans had also built one of the world’s largest local mesh networks, called SNET, but it was shut down by the government in 2019.
Given all of this, it’s especially remarkable that videos of anti-government protests have recently made their way out of Cuba on YouTube and Twitter, giving us glimpses into Cuban life and politics that have been invisible to the outside world for decades. This piece unpacks what that may mean for Cuba’s future, US-Cuba relations, and more.
→ The contrarevolución will be livestreamed | The Pull Request
6: Gallium melts in your hand, powers your cell phone
The tech press generally spends most of its time on new inventions: devices, services, and software that are based on tried-and-true science and technology underneath. To continue shrinking devices and expanding devices’ capabilities, those underlying assumptions need to be refreshed from time to time.
One such refresh comes in the form of gallium nitride (GaN), a crystal that can handle a lot of energy more efficiently than its silicon-based predecessors. Applications for this tech are primarily focused on power distribution and management, including faster phone chargers, voltage regulators in electric vehicles, and even missile defense systems. The tech is relatively simple, so GaN chips can be made in older factories, bypassing the global semiconductor shortage. And gallium itself is a byproduct of aluminum manufacture, making the material very cheap to acquire.
Gallium nitride may not, in itself, be a particularly fascinating breakthrough, but we have no doubt that it — and other advances in materials tech — will be what gets us to smaller, more powerful, and more environmentally-friendly devices.
→ The novel material that’s shrinking phone chargers, powering up electric cars, and making 5G possible | Wall Street Journal
One poetic thing
Digital artist Martin O’Leary created this scrolling video, composed of images created by a GAN (generative adversarial network) prompted by lines from the poem “Kubla Khan” by Samuel Coleridge. It’s the closest thing we’ve seen to what it might look like if computers dreamt.