CG models, VC values, and AI disguises

CG models, VC values, and AI disguises
By Ethical Futures Lab • Issue #31 • View online
This week, we investigate systems that are in transformation, from fashion to supply chains, and we also take a look at the tactics that people use to respond to the systems around them (even as they change). There’s also a mind-blowingly cool analysis of LEGO and UX design.
Alexis & Matt

1: The 3 As: how we engage with imperfect tech
As technology has become integrated into every aspect of our lives, we often find ourselves confronted by systems that don’t do what we want them to. In some cases that is because they are designed to do something we want to avoid (e.g., interactions that come with troublesome privacy implications). In other cases they are simply misinterpreting what you’re trying to do. The latter issue has been a thorn in the side of geneticists who work with Excel. The names of multiple genes, like MARCH1, are automatically converted to dates by the spreadsheet software, leading to erroneous gene name conversions in about 20% of papers with Excel gene lists. There are workarounds, like adding an apostrophe before the value or converting column types. But instead, a few weeks ago the HUGO Gene Nomenclature Committee (HGNC) decided to actually rename all the problematic genes to avoid the issue altogether.
As we’ve been tracking a variety of human interactions with technology, we’ve identified three distinct tactics that people use when their systems aren’t working for them, which we call the Three A’s. The first is response is Adversarial, where people develop hacks and workarounds to achieve their goals despite the system (we write about these a lot and find them fascinating). The second is Adaptive, where we adapt our own behavior to the system’s rules and expectations, contorting ourselves to match machine perception. These include examples like using an apostrophe so Excel doesn’t turn your gene into a date, or employing hyper-articulation so that voice assistants can better understand you. The example above, where geneticists just gave up and renamed entire genes to please a piece of software, is the third tactic, which we call Acquiescence — where people just stop trying to do the thing they wanted altogether because it’s too frustrating.
Excel Kept Messing Up the Names of Genes, So Scientists Renamed Them
2: Strike a pose
Our friend Chris Kent pointed us to this Vogue piece written by a model about the potential impact of AI-powered virtual models on the fashion industry. We’ve written a lot in this newsletter about virtual celebrities recently, but most media takes tend to range from “Hey, did you know this is a thing now?” to “Whoa, this is creepy and scary!” So we were pleasantly surprised by the nuance and thoughtfulness in this article. Sinead Bovell explores a wide range of ethical implications, both positive and negative. She discusses how virtual models have personas (including gender, ethnicity, and political stances) that are often different from those of their creators, and explores the responsibility those creators have to authentically represent identities other than their own. She talks about “robot cultural appropriation” that could happen otherwise, or if brands use CGIs to champion activist causes without having to actually invest in those causes themselves. But on the flip side, Bovell looks at the positive environmental impact virtual models could have, and she also explores how AI-powered models might allow for new kinds of shopping experiences where accurate models of our own bodies could try on clothes for us.
I Am a Model and I Know That Artificial Intelligence Will Eventually Take My Job
3: Motivations and values in venture capital
We came across this writeup of Sahil Lavingia’s experience building Gumroad from early 2019 and found it very relevant today. We often discuss the underlying assumptions and motivations that subvert our best intentions, and Sahil’s experience with venture capital illustrate some of these contradictions very clearly. In particular we were impressed by his analysis of building a growing business that isn’t growing fast enough, the derisive moniker “lifestyle business” for something that is sustaining and thriving, and how the usual “growth hacks” VC-backed companies are taught may have little to no effect on growth. Through his story, we see how venture funding can often incentivize certain choices that prioritize exponential growth at all costs. While there is a place for venture capital in entrepreneurship, Sahil’s experience shows there may be more sustainable and more rewarding ways forward to create value in the world. 
Reflecting on My Failure to Build a Billion-Dollar Company
4: Digital disguises for good and evil
In our last issue, we wrote about adversarial image cloaking software from the University of Chicago — a program that allows anyone to digitally obscure faces in photographs in order to protect their images from unwanted facial recognition. The software does so by making pixel-level changes that are undetectable to the human eye, but make the photograph “read” very differently to facial recognition software. This week, we saw a similar technique used by a team at McAfee, but with a very different goal. The team managed to manipulate a photo of one person so that it appeared the same to the human eye, but tricked facial recognition software into recognizing it as a photo of somebody else entirely. 
This kind of targeted misclassification could lead to all sorts of nefarious behavior, like allowing someone who is on a no-fly list to board an airplane. The researchers at McAfee conducted the research explicitly to set off these kinds of alarms: their goal is “to demonstrate the inherent vulnerabilities in these AI systems and make clear that human beings must stay in the loop.” It also reveals how the same techniques can be used for radically different outcomes: in one case, allowing individuals to better protect their privacy and in another, allowing people to falsify their identities.
The hack that could make face recognition think someone else is you
5: Shopping: the same, but different
As you’ve no doubt seen, Covid-19 has accelerated the shift toward a delivery-based economy, with Amazon as one of the biggest beneficiaries. After Amazon recently announced that its year-over-year profits had doubled, The Verge reported that the company was in talks with Simon (the largest owner of malls in the US) to lease its “anchor store” spaces for use as fulfillment centers. 
We’re interested in this phenomenon as an example of how our digital behaviors reshape our physical world. Think of it like a retail “ghost kitchen”, where instead of a restaurant without tables, Amazon is building stores without checkout counters. In both cases the basic value chain hasn’t changed: raw materials are turned into products which are delivered to consumers. In both cases, however, the experience a consumer has around that transaction has radically shifted, leaving what once was a highly integrated set of tasks to be split apart and performed separately. Look for more speculation around how digital behaviors remake physical space in an upcoming post.
Amazon reportedly considering mall space for fulfillment centers
6: Reporting from the future
Speculative practices are a big part of our day jobs, imagining how signals we see in technology and media might play out and how we can make small interventions to head toward a more desirable future. This summary from Nieman Lab discusses some examples of journalistic speculation, including two of our favorite podcasts: Science Vs. and Fast Forward. 
Speculation, especially when informed by research and interviews, is a compelling storytelling and educational tool. That said, this review correctly points out that our current climate of misinformation, distrust of the media, and a decline in critical thinking makes it challenging to publish speculative journalism and have the desired outcome. Some may take it as a prediction and act accordingly, where others could use its fictional aspects to cry “fake news” and discount the endeavor altogether. In our view, it’s critical to imagine possible futures in order to act with purpose, and to avoid or accelerate the arrival of these possibilities.  
Speculative Journalism Can Help Us Prepare for What’s to Come. Could It Also Promote Misinformation?
One extremely nerdy thing: LEGO UX
We love this so much. George Cave, a designer and engineer from Austria, did a deep (and we mean deep) dive into the design of LEGO interface panels. You know, the pieces that make up the control panels for your submarine or Batmobile? Turns out you can learn a LOT from these “2x2 decorated slopes” that translate to real-world UX design. 
The six different codings in use in the LEGO interfaces: size, shape, colour, texture, position, operation
The six different codings in use in the LEGO interfaces: size, shape, colour, texture, position, operation
The UX of LEGO Interface Panels
Did you enjoy this issue?
Ethical Futures Lab
By Ethical Futures Lab

Six Signals: Emerging futures, 6 links at a time.

In order to unsubscribe, click here.
Powered by Revue