Actually, there is such a thing as a free lunch
Why squint, when you can blink? Why get old, when your online persona doesn’t have to? Why grow crops, when you can make flour out of thin air? This week we found pieces that challenge some of our most basic assumptions about the world, and as usual, are both excited and skeptical about what might come next.
1: The better to see you with, my dear
It’s pretty standard sci-fi fare to have glasses or contacts that can zoom in, identify objects, and give a heads up display. Scientists got a little closer to making that a reality using contact lenses that can recognize patterns of blinks and use those as commands. One command these lenses understand is “zoom.” Blink twice and the lenses sense small variations in electrical signals in your eye, which activate polymers that expand within the lens, making it more convex.
Leaving aside how these things know what’s in focus or not (I get a headache just thinking about it), using blinks as interface signals is fascinating. A related example of invisible inputs we saw this week was Arnav Kapur’s rig that recognizes pre-verbalization signals. Essentially it responds to phrases you’re thinking about saying, but don’t say. Creating invisible gesture and command input could lead to a whole new set of covert interactions with computers, and a whole host of changes to social norms and interactions as well. Let’s just hope your allergies don’t act up.
Scientists create contact lenses that zoom when you blink twice
Me: Accidentally blinks twice while driving.
2: Neural nets not so smart?
It’s a single data point and bears skepticism, but researchers in Milan have just released a paper describing their efforts to replicate the results of various neural net-based recommendation systems. They analyzed 18 different systems and found that only seven could be “reproduced with reasonable effort”; of those, six could be more easily replicated using more traditional computing techniques.
It’s an old joke that most machine learning algorithms could be replaced by simple if/then statements, but this paper implies there’s a lot of truth behind that punch line. It could mean that many of the models we rely on today aren’t necessary, and neither is all the computing power (and electrical power!) they require to be maintained.
You say: "We added AI to our product"
I hear: "We added a bunch more IF statements to our codebase"
Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Recommendation Approaches
We considered 18 algorithms that were presented at top-level research conferences in the last years. Only 7 of them could be reproduced with reasonable effort.
3: Faces in the crowd
We often see stories of individuals caught up in facial recognition schemes, whether identified at major sporting events or fined for trying to block a camera’s view. This New York Times report from Hong Kong adds a new wrinkle: both protestors and police worry about being recognized, and both utilize tools to try and break the other’s anonymity.
Police began removing their identification numbers from their uniforms, seemingly to avoid being reported for brutality or other misconduct. Protestors combated this using a Telegram channel that acts as a crowd-sourced Shazam for officers’ faces; pictures of suspected police are posted, and participants post the names and information of those it can find. Pro-government channels also exist for crowdsourcing identification of protestors.
While these examples aren’t using facial recognition technology, they do make us question what it means to have an expectation of privacy or anonymity. Is it reasonable any longer to hope to move around, especially in a major city, and not be identified?
In Hong Kong Protests, Faces Become Weapons
A quest to identify protesters and police officers has people in both groups desperate to protect their anonymity. Some fear a turn toward China-style surveillance.
4: The avatar of Dorian Gray
Paul Ford wrote about the punctuated equilibrium of experiencing others’ aging on social platforms. Every few years, digital acquaintances will update their avatars and you experience “the tiny moment of shock as you see a proto-jowl where before there was smooth skin…The dyed orange hair is brown, or the bushy natural do is replaced by a shiny bald head.”
What’s striking about this experience is the way that social media and other digital products are designed for the present moment — the new and the now — even as many of us are well over a decade into our relationships with these products. The internet is 30 years old, social media about half that. But very few digital products have any sense of time, growth, or aging embedded in their experience design.
This dissonance will only continue to increase over time. We’ve already seen it come up in significant ways, like Facebook figuring how to handle deaths amongst its users. But we also see it in mundane ways, like the fact that Google search results have no explicit sense of history (try learning JavaScript from a 10-year old blog post!). What does a 30 or 40-year relationship with Twitter, Facebook, LinkedIn, Instagram, or Google look like? How does our digital detritus follow us or fade away? How might the way those products serve us change or accumulate value over time?
Every week or so someone I know on social media, but not in my daily life, updates their avatar, and suddenly they’re three or four years older…
5: Tea, Earl Grey, hot
Solar Foods, a Finnish company, has quite literally figured out how to make food out of thin air. Its product is a protein powder called Solein that is made out of CO2, water, and electricity. It looks and tastes like wheat flour and could potentially be a revolutionary new food source. Solein is produced indoors through carbon-capture technology, which means that it’s not dependent on arable land, irrigation, or other constraints common to most food production. Innovation like this is particularly necessary and compelling as we begin to see how dramatically climate change is threatening the world’s food supply.
NASA has created food out of thin air and it could be the solution to global hunger
The Finnish company, Solar Foods, has designed a high-protein powder made from CO2, water and electricity.
6: The problem isn’t your data, it’s our data
Data privacy is often framed as a personal responsibility issue: each of us should be responsible for good data hygiene, for reading the terms of service, and for giving informed consent. This has not only led to a victim-blaming stance that shifts the responsibility away from governments and corporations (like we saw in the reactions to the Russian FaceApp), but it also frames the value of personal data too simplistically. As Martin Tismé points out, data about other people can increasingly be used to make accurate statistical inferences about you, even when you opt out of your own data being used. It’s like herd immunity but in reverse; I may be very careful about my own data, but if the people around me aren’t, I’ll be just as vulnerable. Tismé posits that “data as oil” is an outdated metaphor, and that we should instead be thinking of data as more akin to CO2.
“Our own consent (or lack thereof) is becoming increasingly irrelevant. We won’t solve the societal problems pervasive data surveillance is causing by rushing through online consent forms. If you see data as CO2, it becomes clearer that its impacts are societal not solely individual. My neighbour’s car emissions, the emissions from a factory on a different continent, impact me more than my own emissions or lack thereof.”
Data isn't the new oil, it's the new CO2
If you see data as CO2, it becomes clearer that its impacts are societal, not solely individual. This isn’t to abdicate individual responsibility or harm; it’s adding a new lens that we often miss entirely.
One playable thing
This week Alexis learned that the app she has been wanting for ages now exists! RunwayML is a visual creation UI for machine learning, giving you simple interfaces to play with ML models without having to dig deep into code or (god forbid) linear algebra. Of course, our friend Allison McCartney immediately started using it to create some true nightmare fodder. Click through to the thread for some bizarro-world presidential candidates.
I decided to play with the AttnGAN model on Runway ML by describing the top Democratic candidates and seeing what kind of images it could synthesize, a thread of nightmares:
Six Signals: Emerging futures, 6 links at a time.
If you don't want these updates anymore, please unsubscribe here.
If you were forwarded this newsletter and you like it, you can subscribe here.
Powered by Revue