Discover more from Links
Maybe every issue will be an AI issue from now on…
Over the years, Marsha Farmer had learned what to look for. As she drove the back roads of rural Alabama, she kept an eye out for dilapidated homes and trailers with wheelchair ramps. Some days, she’d ride the one-car ferry across the river to Lower Peach Tree and other secluded hamlets where a few houses lacked running water and bare soil was visible beneath the floorboards. Other times, she’d scan church prayer lists for the names of families with ailing members.
Farmer was selling hospice, which, strictly speaking, is for the dying. To qualify, patients must agree to forgo curative care and be certified by doctors as having less than six months to live. But at AseraCare, a national chain where Farmer worked, she solicited recruits regardless of whether they were near death. She canvassed birthday parties at housing projects and went door to door promoting the program to loggers and textile workers. She sent colleagues to cadge rides on the Meals on Wheels van or to chat up veterans at the American Legion bar. “We’d find run-down places where people were more on the poverty line,” she told me. “You’re looking for uneducated people, if you will, because you’re able to provide something to them and meet a need.”
Farmer, who has doe eyes and a nonchalant smile, often wore scrubs on her sales routes, despite not having a medical background. That way, she said, “I would automatically be seen as a help.” She tried not to mention death in her opening pitch, or even hospice if she could avoid it. Instead, she described an amazing government benefit that offered medications, nursing visits, nutritional supplements, and light housekeeping—all for free. “Why not try us just for a few days?” she’d ask families, glancing down at her watch as she’d been trained to do, to pressure them into a quick decision.
Once a prospective patient expressed interest, a nurse would assess whether any of the person’s conditions fit—or could be made to fit—a fatal prognosis. The Black Belt, a swath of the Deep South that includes parts of Alabama, has some of the highest rates of heart disease, diabetes, and emphysema in the country. On paper, Farmer knew, it was possible to finesse chronic symptoms, like shortness of breath, into proof of terminal decline.
The recruiter was a chipper woman with a master’s degree in English. Previously she had worked as an independent bookseller. “Your experience as an English grad student is ideal for this role,” she told me. The position was at a company that made artificial intelligence for real estate. They had developed a product called Brenda, a conversational AI that could answer questions about apartment listings. Brenda had been acquired by a larger company that made software for property managers, and now thousands of properties across the country had put her to work.
Brenda, the recruiter told me, was a sophisticated conversationalist, so fluent that most people who encountered her took her to be human.
But like all conversational AIs, she had some shortcomings. She struggled with idioms and didn’t fare well with questions beyond the scope of real estate. To compensate for these flaws, the company was recruiting a team of employees they called the operators. The operators kept vigil over Brenda 24 hours a day. When Brenda went off-script, an operator took over and emulated Brenda’s voice. Ideally, the customer on the other end would not realise the conversation had changed hands, or that they had even been chatting with a bot in the first place. Because Brenda used machine learning to improve her responses, she would pick up on the operators’ language patterns and gradually adopt them as her own.
It was the spring of 2019. My time as a creative writing student had just come to an end, as had my funding, and the rent was due; I needed a job. I sent the recruiter my CV. Several phone interviews later, I was signing up for training slots and watching a 45-minute PowerPoint presentation on fair-housing law. I did a little maths: an operator made $25 an hour, and worked between 15 and 30 hours a week, depending on how lucky they were in the weekly shift lottery. It wouldn’t be enough to cover my rent, but I had no other leads. I packed my things and moved back home to live with my parents in New Jersey.
The first video game was a 1952 research product called OXO — tic-tac-toe played on a computer the size of a large room:
Fifteen years later Ralph Baer produced “The Brown Box”; Magnavox licensed Baer’s device and released it as the Odyssey five years later — it was the first home video game console:
The Odyssey made Magnavox a lot of money, but not through direct sales: the company sued Atari for ripping off one of the Odyssey’s games to make “Pong”, the company’s first arcade game and, in 1975, first home video game, eventually reaping over $100 million in royalties and damages. In other words, arguments about IP and control have been part of the industry from the beginning. […]
Forty years of context may seem like overkill when it comes to examining the FTC’s attempt to block Microsoft’s acquisition of Activision, but I think it is essential for multiple reasons.
First, the video game market has proven to be extremely dynamic, particularly in terms of 3rd-party developers:
Atari was vertically integrated
Nintendo grew the market with strict control of 3rd-party developers
Sony took over the market by catering to 3rd-party developers and differentiating on hardware
Xbox’s best generation leaned into increased commodification and ease-of-development
Sony retook the lead by leaning back into vertical integration
That is quite the round trip, and it’s worth pointing out that attempting to freeze the market in its current iteration at any point over the last forty years would have foreclosed future changes.
Most people won't miss 2022.
It was a year defined by a lingering pandemic, rampant inflation, unstable economies, multiple humanitarian crises, a summer of climate hell in many parts of Asia, Europe and North America, and more than anything else, the outbreak of a devastating war in Ukraine. Wherever you turned, it was hard to escape the sense that things were falling apart.
That wasn't all that happened though. Our two media outlets, Future Crunch and The Progress Network, spent the year reporting a very different set of stories, the ones you didn't see on the evening news or in your social feeds. Unlikely as it may seem, 2022 was also a year of uplifting human rights victories, extraordinary conservation wins, big milestones in global health and development, and an unprecedented acceleration in the clean energy transition.
We teamed up to bring you a comprehensive list of all that good news. Our goal isn't to try convince you to take one side over the other in a debate about optimism and pessimism - the world is far too muddled for that. Instead, it's to remind you that away from the headlines, millions of people from every corner of the planet did their best to solve the problems that could be solved, and stayed open-eyed and open-hearted even in the most difficult of circumstances.
I’m no true believer. But I’m less than infinitely hostile to crypto. This is becoming a pretty rare position, so let me explain why:
Crypto Is Full Of Extremely Clear Use Cases, Which It Already Succeeds At Very Well
Look at the graph of countries that use crypto the most:
Do Vietnamese people love trading monkey gifs? Are Ukrainians especially susceptible to Ponzi schemes? Is Venezuela laden with techbros?
Vietnam uses crypto because it’s terrible at banks. 69% of Vietnamese have no bank access, the second highest in the world. I’m not sure why; articles play up rural poverty, but many nations have more rural poor than Vietnam. There’s a history of the government forcing banks to make terrible loans, and then those banks collapsing; maybe this destroyed public trust? In any case, between banklessness and remittances (eg from Vietnamese-Americans), Vietnam leads the world in crypto use.
Ukraine has always been among the top crypto countries: in 2021, NYT called it “the crypto capital of the world”. Again, this owes a lot to its terrible banking system.
In previous pieces, I argued that there's a real and large risk of AI systems' developing dangerous goals of their own and defeating all of humanity - at least in the absence of specific efforts to prevent this from happening.
A young, growing field of AI safety research tries to reduce this risk, by finding ways to ensure that AI systems behave as intended (rather than forming ambitious aims of their own and deceiving and manipulating humans as needed to accomplish them).
Maybe we'll succeed in reducing the risk, and maybe we won't. Unfortunately, I think it could be hard to know either way. This piece is about four fairly distinct-seeming reasons that this could be the case - and that AI safety could be an unusually difficult sort of science.
This piece is aimed at a broad audience, because I think it's important for the challenges here to be broadly understood. I expect powerful, dangerous AI systems to have a lot of benefits (commercial, military, etc.), and to potentially appear safer than they are - so I think it will be hard to be as cautious about AI as we should be. I think our odds look better if many people understand, at a high level, some of the challenges in knowing whether AI systems are as safe as they appear.
This post is about China, and language, and how the press responds to criticism and handles its mistakes. It centers on a big story from ProPublica and Vanity Fair six weeks ago, which I thought was fishy from the start and only looks worse on further exposure.
Last October 28, ProPublica and Vanity Fair co-released an investigative report about the origins of the Covid pandemic. It was a splashy, major presentation, which opened with a dramatic, full-screen, black-and-white photo of one of the key figures, his face in shadows as befits the deep mysteries that were being explored. […]
Instead I talked with Brendan O’Kane, who for many years has made his living as a translator of Chinese material of all sorts — novels, news reports, party documents, everything. We had mutual friends in the Beijing community during the years when my wife, Deb, and I were living there. He is now based back in the U.S. […]
O’Kane The ProPublica article quotes part of another sentence as evidence that the lab wasn’t capable of handling infectious materials. In the original report, it said “Even at the BSL-4 lab, they repeatedly lamented the problem of ‘the three ‘nos’: no equipment and technology standards, no design and construction teams, and no experience operating or maintaining [a lab of this caliber].’” Which sounds pretty bad!
Until you click on the link the ProPublica article supplies to the original document, and see that there's another half of that sentence. And my translation of the whole sentence would be:
“At the outset of construction, the Wuhan P4 lab faced the dilemma of the “three ‘nos’”: no equipment or technical standards, no design and construction teams, and no operations or maintenance experience — but with Party members from the Zhengdian Lab [BSL4]’s Party branch leading the charge and bravely pushing forward, [the lab] achieved the 'Three 'Yes'es': a well-developed set of standards, a seasoned operations and maintenance team, and invaluable construction experience.”
Fallows So this would be like a sentence in English saying “we used to be so terrible, but now we're great.”
That's the import of the document itself. Reid is approaching it as if it's sinister. To me, it looks like boilerplate. It's a minor Party office, celebrating its own existence. (The supposedly damning passage about test tubes and Pandora’s Box doesn’t even come from November 2019; almost the exact same text appears in a document from August 30, 2019.
i feel a little embarrassed writing this bc i think people are going to judge my behavior more after i post it. as well they should honestly. also, you should not listen to me, i have no authority. but please, have mercy on me a sinner, if i were naturally good at this i wouldn't be able to talk about it explicitly. also also there is no way to express all this without being a little more cynical-sounding than i actually feel. but anyway here it is like i said
so. imagine you’re a young woman. (apparently everyone likes to do that.) you are starting to possess some degree of sexual attraction, which impacts others around you. also you live in a social world & you want things from other people. at the same time you are ambivalent about your powers of sexual attraction, which sometimes cause people to behave threateningly or at least unpleasantly towards you, and definitely put you in a lot of awkward situations you didnt ask for. & also bc you're being judged & it’s hard to get a break. what to do? […]
so what are your other options besides disavowing your sexual attractiveness one way or the other, or being overtly sexy? well, you could act like a lady
Perhaps It Is A Bad Thing That The World's Leading AI Companies Cannot Control Their AIs | Astral Codex Ten
By now everyone has their own opinion about whether the quest to prevent chatbots from saying “I love racism” is vitally important or incredibly cringe. Put that aside for now: at the very least, it’s important to OpenAI. They wanted an AI that journalists couldn’t trick into saying “I love racism”. They put a lot of effort into it! Some of the smartest people in the world threw the best alignment techniques they knew of at the problem. Here’s what it got them:
As part of DeepMind’s mission to solve intelligence, we created a system called AlphaCode that writes computer programs at a competitive level. AlphaCode achieved an estimated rank within the top 54% of participants in programming competitions by solving new problems that require a combination of critical thinking, logic, algorithms, coding, and natural language understanding.
Charcoal and burned bones found in a South African cave offer intriguing — if controversial — clues about Homo naledi
You may be familiar with the truffle oil scam, but everything else you think you know about truffles is probably a lie too.
As the end of the year approaches, it is worth considering which of our earlier views we have reevaluated. I have a nomination: I am these days less enamored of the British parliamentary or “Westminster” system of government, which no longer seems well-functioning. The British version of the idea in particular.
Some traits of the British Westminster system are the fusion of the executive and legislative branches of government, first past the post democratic elections, the relative weakness of judicial vetos, and the relative absence of federalist structures. All of those features centralize power in the national state.
The Westminster system long has had American fans, most recently political commentator Matt Yglesias. These individuals praise the parliamentary system for giving government a chance to get things done, subject to a periodic, up-down democratic check.
What I am observing is that, contrary to common reputation, the UK political system is turning out to be more gridlocked than the American system.
Ever since reading Tim Urban's epic blog posts on artificial intelligence years ago I have keept a healthy interest in (and fear of) AI. This weekend OpenAI released version 3.5 of their Chat GPT AI.
It is breathtaking.
If you have been on Twitter the past few days you've seen it. If not, you need to. The tool is a bit of software that allows you to ask questions or give commands and get responses from the computer that are extremely human-like and impressive.
You don't have to be a techie to get enthralled with this. In fact I predict every person in the world will be interacting with some form of this technology in the next couple years.
I'm going to try to explain it to you but it is better to show than to tell so most of this edition will be screen shots of me and others interacting with the Chat GPT AI.
Scientists studying fusion energy at Lawrence Livermore National Laboratory in California announced on Tuesday that they had crossed a long-awaited milestone in reproducing the power of the sun in a laboratory.
That sparked public excitement as scientists have for decades talked about how fusion, the nuclear reaction that makes stars shine, could provide a future source of bountiful energy.
The result announced on Tuesday is the first fusion reaction in a laboratory setting that actually produced more energy than it took to start the reaction.
On November 30, OpenAI released ChatGPT, one of the most advanced conversational AI systems ever created. In just five days since its release, over 1 million people have interacted with the system.
I have applied the political typology test from Pew Research to ChatGPT to determine if its answers were skewed towards either pole of the political orientation spectrum.
Could I cool down the Earth by capturing a comet and dropping it in the ocean, like an ice cube in a glass of water? […]
No. In fact, it's honestly sort of impressive to find a solution that would actively make the problem worse in so many different ways.
For just a moment on December 6, nearly 9 in 10 people worldwide will simultaneously experience nighttime.
A few months ago, our number crunchers fact-checked (and partially confirmed) an internet claim that 99% of the world’s population gets sunlight at 11:15 UTC on July 8. This left us wondering: What about the opposite situation? At what precise instant do most people experience nighttime?
Trials rider and mountain biker Danny MacAskill is one of my long-running obsessions here — I first posted about him all the way back in 2009 and if there’s ever a kottke.org konference, you’d better believe MacAskill will be performing at it. Anyway, MacAskill recently visited San Francisco with Red Bull and explored some of that beautiful city’s most iconic locations on his bike. Wow, the tennis net ride at 2:45 — BONKERS.
A new analysis of Hubble data has clinched it: There's too much light in the space around the Solar System.
Not much extra light, to be sure. Just a subtle, ghostly glow, a faint excess that can't be accounted for in a census of all the light-emitting objects.
The strangest moon in the Solar System is bright yellow. The featured picture, an attempt to show how Io would appear in the "true colors" perceptible to the average human eye, was taken in 1999 July by the Galileo spacecraft that orbited Jupiter from 1995 to 2003. Io's colors derive from sulfur and molten silicate rock.
So what are some tantalizing locational mistakes that seemingly come pre-installed in American students’ minds that geography teachers wrestle to overcome?
So glad you asked! Here is a cherry-picked handful of examples that we’ll dive into…
The northiness of Africa
The northiness of Europe
The eastiness of South America
The idea that human affairs can be largely understood by looking at the relief map is not particularly in vogue at present, when the power of ideas is considered supreme. Yet geography continues to play a surprisingly resilient role in politics. This book is one of Kaplan's typical sweeping historical analyses, looking at how the contours of the world map continue to give us clues to how states will act. Love him or hate him, as always he is exceptionally erudite. One of the interesting arguments he makes, drawing from the geographer Harold Mackinder, is how the existence of liberalism is often dependent on the existence of large bodies of water surrounding a nation.
In this video from Vox (produced by none other than Estelle Caswell, who does the excellent Earworm series), scholar and musician Jake Blount runs us through a quick history of early Black folk music, using the banjo as a rough through-line.