Got a ticket? Maybe you should get ChatGPT to write an explanation of why it should be revoked. CC-licensed photo by Charleston’s TheDigitelCharleston%27s TheDigitel on Flickr.
You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.
Next Friday there’ll be another post due at the Social Warming Substack, at about 0845 UK time.
A selection of 10 links for you. Overflowing with energy. I’m @charlesarthur on Twitter. Observations and links welcome.
FTX founder Sam Bankman-Fried arrested after US files criminal charges • CNBC
MacKenzie Sigalos and Rohan Goswami:
»
FTX founder Sam Bankman-Fried was arrested by Bahamian authorities this evening after the United States Attorney for the Southern District of New York shared a sealed indictment with the Bahamian government, setting the stage for extradition and US trial for the onetime crypto billionaire at the heart of the crypto exchange’s collapse.
Damian Williams, the US Attorney for the Southern District of New York, said on Twitter that the federal government anticipated moving to “unseal the indictment in the morning.”
Bahamas Attorney General Ryan Pinder said that the United States had filed unspecified criminal charges against Bankman-Fried and was “likely to request his extradition.”
In a statement, Bahamian Prime Minister Philip Davis said, “The Bahamas and the United States have a shared interest in holding accountable all individuals associated with FTX who may have betrayed the public trust and broken the law.”
«
This month’s least surprising headline. The question now is how long the extradition will take. Apropos of nothing, there’s an Ecuadorian embassy in Nassau.
unique link to this extract
Exclusive: US Justice Dept is split over charging Binance as crypto world falters, sources say • Reuters
By Angus Berwick, Dan Levine and Tom Wilson:
»
Splits between US Department of Justice prosecutors are delaying the conclusion of a long-running criminal investigation into the world’s largest cryptocurrency exchange Binance, four people familiar with the matter have told Reuters.
The investigation began in 2018 and is focused on Binance’s compliance with US anti-money laundering laws and sanctions, these people said. Some of the at least half dozen federal prosecutors involved in the case believe the evidence already gathered justifies moving aggressively against the exchange and filing criminal charges against individual executives including founder Changpeng Zhao, said two of the sources. Others have argued taking time to review more evidence, the sources said.
The inquiry involves prosecutors at three Justice Department offices: the Money Laundering and Asset Recovery Section, known as MLARS, the US Attorney’s Office for the Western District of Washington in Seattle and the National Cryptocurrency Enforcement Team. Justice Department regulations say that money laundering charges against a financial institution must be approved by the MLARS chief. Leaders from the other two offices, along with higher-level DOJ officials, would likely also have to sign off on any action against Binance, three of the sources said.
«
If Binance gets investigated.. that would pretty much be game over for a lot of the crypto market. And it feels very unlikely that its controls have been tight enough to prevent money laundering through the exchange.
Former top Twitter official forced to leave home due to threats amid ‘Twitter Files’ release • CNN Business
Donie O’Sullivan:
»
Twitter’s former head of trust and safety has fled his home due to an escalation in threats resulting from Elon Musk’s campaign of criticism against him, a person familiar with the matter told CNN on Monday.
Yoel Roth, who resigned from the social media company in November, has in recent weeks faced a storm of attacks and threats of violence following the release of the so-called “Twitter Files” — internal Twitter communications that new owner Musk has released through journalists including Matt Taibbi and Bari Weiss.
Roth’s position involved him working on sensitive issues including the suspension of then-President Donald Trump’s account in 2021. On Monday, Weiss posted a series of screenshots purported to show internal Twitter documents where Roth and others discussed whether to ban Trump’s account, with some employees questioning if the former president’s tweets violated the platform’s policies.
While Musk had initially been publicly supportive of Roth, that soon changed after he left the company. Roth has since been the subject of criticism and threats following the release of the Twitter Files. However, things took a dark turn over the weekend when Musk appeared to endorse a tweet that baselessly accused Roth of being sympathetic to pedophilia — a common trope used by conspiracy theorists to attack people online.
«
A common trope used by rightwing conspiracy theorists, which Musk either is, or seeks to encourage. And the difference is.. irrelevant.
unique link to this extract
So that fusion “net energy gain” means what for the future, exactly? •Thread Reader App
Wilson Ricks is a PhD candidate in large energy systems modelling at Princeton University:
»
The National Ignition Facility (NIF) has achieved net energy gain from fusion! This is incredibly exciting scientifically, but what does it mean for the future of energy?
In all likelihood, very little.
As I suspected yesterday, even if the lasers have generated more energy out than was put in, the trouble is that the lasers are only 1% efficient. So you don’t want 20% more power output; you need 100x or more. And that’s only the beginning of the power losses.
Ricks thinks that magnetic confinement fusion (as in ITER and JET) actually has the better chance of solving this. Full details are expected today, Tuesday, but basically it sounds like it’s back to square 100.
unique link to this extract
Abandoned: the human cost of neurotechnology failure • Nature
Lim Drew:
»
secret that has changed his life. Under the skin, nestled among the nerve fibres that allow him to feel and move his face, is a miniature radio receiver and six tiny electrodes. “I’m a cyborg,” he says, with a chuckle.
This electronic device lies dormant much of the time. But, when Möllmann-Bohle feels pressure starting to gather around his left eye, he retrieves a black plastic wand about the size of a mobile phone, pushes a button and fixes it against his face in a home-made sling. The remote vibrates for a moment, then launches high-frequency radio waves into his cheek.
In response, the implant fires a sequence of electrical pulses into a bundle of nerve cells called the sphenopalatine ganglion. By disrupting these neurons, the device spares 57-year-old Möllmann-Bohle the worst of the agonizing cluster headaches that have plagued him for decades. He uses the implant several times a day. “I need this device to live a good life,” he says.
«
It was made by an American company called ATI. But:
»
by the end of 2019, ATI had collapsed. The company’s closure left Möllmann-Bohle and more than 700 other people alone with a complex implanted medical device. People using the stimulator and their physicians could no longer access the proprietary software needed to recalibrate the device and maintain its effectiveness. Möllmann-Bohle and his fellow users now faced the prospect of the battery in the hand-held remote wearing out, robbing them of the relief that they had found. “I was left standing in the rain,” Möllmann-Bohle says.
«
Drew pulls together plenty of examples: there’s such a huge tension between the investment needed to come up with new technologies, the cost to healthcare systems of deploying them, and the need of companies to make a profit.
unique link to this extract
NYC Mayor Eric Adams took three paychecks in crypto. How’s that going for him? • Slate
Alexander Sammon:
»
On Jan. 21, according to reporting in the Verge, Adams received the first of his biweekly paychecks— $5,900, according to the New York Post—and flowed that money into the crypto exchange Coinbase. Adams’ office confirmed that he did indeed go on to convert three paychecks into cryptocurrency, splitting the money between Bitcoin and Ethereum.
Some back-of-the envelope math gives a rough sense of how much money that decision has cost hizzoner. At time of writing, Bitcoin sits at $16,811.40; Ethereum is $1,229.74. Maybe they’ll have gone up since then; it’s possible, by the time you’re reading this, they’ll be lower still. That means that Adams has lost roughly 53%, 60%, and 57% on his first three paychecks.
Check my work: I calculated those figures based on the daily average of each currency on each of the three paydays. I assumed that the money was parceled evenly between the two separate coins. Cryptocurrencies can fluctuate substantially in value even on a minute-by-minute basis (critics would say this is just one reason they seem not to be especially well-suited as currencies), so it’s possible Adams arbitraged his purchases so perfectly that he beat the daily average. It’s possible too that he loaded up on Bitcoin, which is only down 64% this year, rather than Ethereum, which is down 66%.
…Meanwhile, Adams’ other November commitment, support for a New York City–specific cryptocurrency, has gone even worse. NYCCoin is down almost 94% since it was introduced with the mayor’s support in February. MiamiCoin, which began trading in August of 2021 and has been pushed by that city’s similarly crypto-zealous mayor, Francis Suarez, is down 98% for the year, trading at $0.000458.
«
The critique of crypto remains the same: what is it better at doing than fiat currency? “Being transferred” is a good answer, but it’s not sufficient while the price yo-yos around.
unique link to this extract
GitHub – f/awesome-chatgpt-prompts: This repo includes ChatGPT prompt curation to use ChatGPT better • Github
»
The ChatGPT model is a large language model trained by OpenAI that is capable of generating human-like text. By providing it with a prompt, it can generate responses that continue the conversation or expand on the given prompt.
In this repository, you will find a variety of prompts that can be used with ChatGPT. We encourage you to add your own prompts to the list, and to use ChatGPT to generate new prompts as well.
«
So we focused for a bit on magic spells for pictures; now it’s incantations for text to make it act as an interviewer, a football commentator, a Javascript console, a plagiarism checker, a character from a movie or book,..
unique link to this extract
Disputing a parking fine with ChatGPT • Notes by Lex
Lex Toumbourou:
»
Recently, on holidays in Far North Queensland, my wife and I parked our rental car in a paid parking lot to visit a restaurant.
I paid using the EasyPark App per the council’s instruction on various signs throughout the lot.
When we returned, they had slapped a fine on our Toyota anyway [for “failing to display a valid ticket in the prescribed manner”].
I double-checked what I had entered into the app. I mistyped the number plate by one letter. Oops.
Since we have never received a parking fine before, and I had proof of payment, I knew there was a good chance the council would let me off if I sent a letter of explanation. We had already been experimenting with ChatGPT, and the letter seemed a good test case.
ChatGPT: Write a letter to Cairns Regional Council requesting to dispute parking fine: I paid in EasyPark but accidentally mistyped the number plate.The first response was close but longer than I would like. Plus, I didn’t tell it that I was planning to attach a photo of the fine and proof of payment.
That is good, but make it shorter. Also include that I have attached fine and proof of payment.«
And he gets a response that the Penalty Infringement Notice has been withdrawn; the language used, to my eyes at least, is quite similar to that spouted by ChatGPT. Only me?
unique link to this extract
Recruited for Navy SEALs, many sailors wind up scraping paint • The Japan Times
Dave Philipps:
»
A sailor fresh out the elite Navy SEAL selection course slung his gear over his broad shoulder and clomped down a steel ladder into the guts of a Navy ship to execute a difficult, days-long mission specifically assigned to him: scrubbing the stinking scum out of the ship’s cavernous bilge tank.
Hardly the stuff of action movies, but it’s how many would-be SEALs end up.
The Navy attracts recruits for the SEALs using flashy images of warriors jumping from planes or rising menacingly from the dark surf. But very few make it through the harrowing selection course, and those who don’t still owe the Navy the rest of their four-year enlistments. So they end up doing whatever Navy jobs are available — often, menial work that few others want.
The recruits are almost all hypermotivated overachievers, often with college degrees, who have passed a battery of strength and intelligence tests. But many find themselves washing dishes in cramped galleys, cleaning toilets on submarines or scraping paint on aircraft carriers.
Unlike civilian workers, they cannot quit. To walk away would be a crime.
…Relegating promising candidates who don’t quite clear the bar to years of drudgery would be a harsh arrangement even if the SEAL selection course were running as designed. But lately, classes that were always hard became dangerous. A number of sailors were hospitalized. Others were forced to quit if they wanted medical care. And in February, one sailor died.
…On average, about 70% of each class over the last decade has rung the bell. But the rate suddenly soared in 2021, reaching as high as 93%.
«
And the US Navy doesn’t quite know why.
unique link to this extract
ChatGPT, Galactica, and the progress trap • WIRED
Abeba Birhane and Deborah Raji are Fellows in Trustworthy AI at the Mozilla Foundation:
»
Among the most celebrated AI deployments is that of BERT—one of the first large language models developed by Google—to improve the company’s search engine results. However, when a user searched how to handle a seizure, they received answers promoting things they should not do—including being told inappropriately to “hold the person down” and “put something in the person’s mouth.” Anyone following the directives Google provided would thus be instructed to do exactly the opposite of what a medical professional would recommend, potentially resulting in death.
The Google seizure error makes sense, given that one of the known vulnerabilities of LLMs is their failure to handle negation, as Allyson Ettinger demonstrated years ago with a simple study. When asked to complete a short sentence, the model would answer 100% correctly for affirmative statements (“a robin is …”) and 100% incorrectly for negative statements (“a robin is not …”). In fact, it became clear that the models could not actually distinguish between the two scenarios and provided the exact same responses (using nouns such as “bird”) in both cases.
Negation remains an issue today and is one of the rare linguistic skills to not improve as the models increase in size and complexity. Such errors reflect broader concerns linguists have raised about how such artificial language models effectively operate via a trick mirror—learning the form of the English language without possessing any of the inherent linguistic capabilities that would demonstrate actual understanding.
Additionally, the creators of such models confess to the difficulty of addressing inappropriate responses that “do not accurately reflect the contents of authoritative external sources.” Galactica and ChatGPT have generated, for example, a “scientific paper” on the benefits of eating crushed glass (Galactica) and a text on “how crushed porcelain added to breast milk can support the infant digestive system” (ChatGPT).
«
The negation point explains a great deal about what we’ve seen.
unique link to this extract
| • Why do social networks drive us a little mad? • Why does angry content seem to dominate what we see? • How much of a role do algorithms play in affecting what we see and do online? • What can we do about it? • Did Facebook have any inkling of what was coming in Myanmar in 2016? Read Social Warming, my latest book, and find answers – and more. |
Errata, corrigenda and ai no corrida: none notified
The Overspill: when there's more that I want to say 