WR 15: Deepfake drama, scam text sorties, and a reader survey!
Weekly Reckoning for the week of 22/1/24
Welcome back to the Ethical Reckoner. In this Weekly Reckoning, we cover an attempt at voter suppression using deepfake robocalls, voter engagement using big data, and how those wrong-number texts you’re getting might provoke international conflict. Then, I muse on how technological revolutions actually take time, and it’s vital not to squander those preparatory transition periods. (But which transition are we in? You can probably guess…)
Also, I’ll ask again at the end, but I’m doing an Ethical Reckoner Reader Survey! If you have a few seconds to help make the newsletter better by filling out this short survey, I’d really appreciate it.
This edition of the WR is brought to you by… an actual sponsor! SciSpace is an awesome AI-powered research tool that uses generative AI not to make up papers (unlike some genAI I could mention) but to summarize and organize them so you can research more efficiently. I’ve found it super helpful for discovering and sifting through papers, and you can try it for free here. If you love it and want unlimited access, get 20% off a monthly subscription with code EMMIE20 and 40% off a year with code EMMIE40 here.
The Reckonnaisance
Deepfake robocalls crash the US election
More: The call, which was likely created using AI deepfake technology, urged people to skip the primary and “save your vote for the November election” in an apparent effort to hurt Biden, who was the subject of a write-in campaign after internal party machinations led to his name being left off the ballot. The callback number was the personal cell phone number of the former chair of the state Democratic Party who’s been helping organize the write-in campaign, and she got at least a dozen calls asking her about the message. (Biden still won in the state.)
Why you should care: Obviously, it’s not great that we’ve got more generative AI in our elections, especially clear attempts at voter suppression. My theory of AI deception in elections is that it doesn’t have to be widespread to be effective. Large-scale disinformation campaigns may not convince anyone, but contribute to the “liar’s dividend” where politicians can claim that any news they don’t like is fake—which Trump is already taking advantage of. Small campaigns with targeted messaging can actually change people’s minds. In the US, elections are often decided by a relatively small number of votes (whether because it’s a local election or because it’s a presidential election, where the electoral college system means your vote effectively only matters if you’re in one of a handful of swing states). So, a disinformation campaign can be extremely small-scale and targeted, but still make a huge difference, and I worry this is just a test of the waters—especially considering that some social media companies are stepping back from monitoring election content.
Scam texts might provoke international conflict in Southeast Asia
Nutshell: “Scam dens” in Myanmar have been capturing Chinese workers to forcibly work their scams, and China is starting to crack down from across the border.
More: Ok, this is a crazy story to begin with, one that I’ve been meaning to write up for a while but it never felt news-y enough. Anyway, you know those seeming wrong number texts you get saying it was great to meet you, or asking if Anna is there, or something like that? If you respond, eventually they’ll try to get you to “invest” in a crypto scam before ghosting with all your money (in total, multiple billions of dollars in 2022). Wild enough, but it turns out that a lot of the people sending these scam texts are victims of human trafficking who are forced to work in brutal “scam dens” in Southeast Asia. Many of the victims of both the trafficking and the scams are Chinese, and the Chinese government obviously isn’t happy about that, but Myanmar—which is embroiled in a civil war—isn’t doing much to stop it. So now China is getting increasingly enmeshed in that conflict, arresting kingpins in Myanmar and pressuring local leadership to close scam dens. China is also maybe supporting a rebel group that’s shutting down many scam centers and is also launching PR and diplomatic pressure campaigns against the scam dens, which are also implicitly against Myanmar.
Why you should care: That region is already troubled enough and doesn’t need more conflict. There’s not much you or I can do about it, but next time someone asks if they can talk to Juan, spare a thought for the poor person who sent it and the really terrible physical and geopolitical system they’re in.
Apple Vision Pro sells out on launch, but headwinds remain
Nutshell: The $3500 Vision Pro, Apple’s new extended reality (XR) headset, sold out on launch, but demand may be tapering off.
More: This is good initial news for Apple, but there are indicators that not everything is roses. Apple didn’t make that many headsets for the launch (estimates are 60,000-80,000), so it wasn’t that hard to sell out. The headset also is launching without native apps for popular platforms like YouTube and Netflix—who are either peeved at Apple for the recent app store disputes, don’t think a Vision Pro app would be a big revenue driver, or both—leaving it heavily dependent on its version of Safari and a potentially clunky typing interface.
Why you should care: Well, unless you have $3500 to drop on a first-generation device, you probably don’t. But Apple is betting—and I’m hoping, if only for the sake of my thesis—that this ushers in the era of “spatial computing,” a paradigm shift equivalent to the launch of the Mac or the iPhone. We could be witnessing history—or one of the biggest product flops of all time.
India’s ruling party raising concerns with voter data collection
Nutshell: The BJP’s Saral app, which started as a volunteer organizational tool, is being used to track voter engagement before India’s 2024 election.
More: Door-to-door app-download campaigns are getting scores of people to install the app and input their personal information. Their engagement with the app is used to track local party strength. This is done in “booths” of 700-800 people, meaning the BJP can know on an extremely granular level which voters they need to target.
Why you should care: This is reflective of an increasing trend towards techno-authoritarianism in India, with things like the national digital biometric ID system aadhaar and government exceptions to the new data protection law serving to increase state power. Campaigns everywhere thrive on data collection, and any party can make their own app, but the BJP has a clear advantage over them when they can promote it on state-owned airwaves. And while this shows that non-AI technology can be hugely concerning in elections, this is also a possible blueprint for the kind of targeted generative AI campaigns that I’m concerned about.
Extra Reckoning
I’ve been thinking about the pace of technological change and how we remember and historicize periods of transition. When we think back on history, we think in clean cuts: pre-printing press, post-printing press. Pre-electricity, post-electricity. For younger readers (myself included), pre-Internet, post-Internet. Maybe even pre-smartphone, post-smartphone. But I’m at least old enough to remember the smartphone transition period, when the early generations of iPhone were going through big design changes and slowly spreading through my peer group. Thankfully there was no iMessage until 2011 and thus no green-vs-blue bubble enmity, or else I might’ve been mocked until the end of my freshman year of high school in 2013, when I was gifted my mom’s old iPhone 4 to replace my slider phone. Looking back on my photos from that time reveal a lot of random screenshots, bad selfies, and this not-entirely-terrible picture of a tulip(?) that I’m shocked came from an iPhone 4.
But most days, I barely think about what it was like to not have a smartphone—and specifically, a highly capable smartphone that has dozens of features that didn’t exist on early iPhones. All technology has a transition or adoption period, some longer than others (it took decades to electrify most US households). But we tend to think in these snapshots when we look back—or forward. The idea of sudden technological change informs our most utopic and dystopic thinking. Take nuclear fusion, for example. The dream is clear: one scientific breakthrough, and the next day we’re all the Jetsons with cheap, limitless power and no need to fear climate change. But in reality, even if we had a major breakthrough tomorrow, it will likely take until the 2030s or 2040s to even get a prototype off the ground, and it could take another decade after that for plants to be powering infrastructure. We can’t rely on fusion like it’s a knight on a white horse galloping in to save the day. The knight has to trudge through a few swamps and slay a few dragons first.
On the other hand, this can also be a good thing. Take quantum decryption, which is the use of quantum computing to decrypt information encrypted with classical cryptography (so your emails,1 WhatsApps, national security information, etc.). Encryption is really important, and if it were to break overnight, that would be correspondingly really bad. Thankfully, that probably isn’t going to happen—even though we have theorized algorithms (as far back as 1994) for quantum decryption, quantum computers are too small and error-prone to effectively decrypt much of anything right now. And yes, some things evolve exponentially, but quantum is so far advancing slowly. We’ll have plenty of warning time before we have to deploy the quantum-resistant encryption methods that organizations like NIST are already developing.
While we dream of—or fear—sudden breakthroughs, the reality is generally one of incremental progress. And it can feel like a long transition time. I wanted an iPhone for ages before I was finally allowed to have one. But then one day, I had one, and the transition period was done. I didn’t need to do much to prepare for having a smartphone—although smartphones are so tied up with other issues like social media and screentime that there’s a lot to be said for preparing kids for having them. On a bigger scale, there’s much that can be done in these interregnum periods when we’re anticipating technological change, like how NIST is developing quantum-resistant encryption. But if we don’t, one day we wake up and the change has happened; the electricity has turned on, the computer has been booted up, we’re in a new era, and there’s no going back.
We’re in one of these weird transition periods now with AI. But some are using this time more wisely than others. The EU finally agreed on a text for its AI Act, China has passed a boatload of AI regulations, but the US has… some voluntary commitments and executive orders. If we can’t get our act together, we’ll wake up and we’ll officially be behind in regulating what’s looking like the next major era of technology. It’s vital that we appreciate the transitory nature of transition periods, on an individual and societal level, and not squander them.
I Reckon…
that it would really help me out if you filled out this short reader survey!
Sometimes.
Thumbnail generated by DALL-E 3 via ChatGPT when I uploaded the tulip photo and asked it to “Make an abstract painting version of this flower”.