Despite having more than 500 million devices in homes, Amazon can’t get Alexa to turn a profit, and is cutting jobs. CC-licensed photo by ajay_suresh 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.
There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.
A selection of 11 links for you. Not that smart. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
Exclusive: Sam Altman’s ouster at OpenAI was precipitated by letter to board about AI breakthrough -sources • Reuters
Anna Tong, Jeffrey Dastin and Krystal Hu:
»
Ahead of OpenAI CEO Sam Altman’s four days in exile, several staff researchers sent the board of directors a letter warning of a powerful artificial intelligence discovery that they said could threaten humanity, two people familiar with the matter told Reuters.
The previously unreported letter and AI algorithm was a catalyst that caused the board to oust Altman, the poster child of generative AI, the two sources said. Before his triumphant return late Tuesday, more than 700 employees had threatened to quit and join backer Microsoft in solidarity with their fired leader.
The sources cited the letter as one factor among a longer list of grievances by the board that led to Altman’s firing. Reuters was unable to review a copy of the letter. The researchers who wrote the letter did not immediately respond to requests for comment.
OpenAI declined to comment.
According to one of the sources, long-time executive Mira Murati told employees on Wednesday that a letter about the AI breakthrough called Q* (pronounced Q-Star), precipitated the board’s actions.
The maker of ChatGPT had made progress on Q*, which some internally believe could be a breakthrough in the startup’s search for superintelligence, also known as artificial general intelligence (AGI), one of the people told Reuters. OpenAI defines AGI as AI systems that are smarter than humans.
«
Murati has been one of the people pumping the brakes on OpenAI’s headlong rush towards AGI, as has chief scientist Ilya Sutskever. Both were influential in chucking Altman out last week. Draw your own conclusions. Certainly one of the hot take guesses last weekend was “OpenAI has invented AGI but Altman didn’t tell the board or try to slow it down, ergo defenestration”. Which maybe was close to right.
Anyway, perhaps this will be a hinge moment in human history. I for one welcome our new etc.
unique link to this extract
“ChatGPT with voice” opens up to everyone on iOS and Android • Ars Technica
Ron Amadeo:
»
It may have been a chaotic week at OpenAI, but the company has somehow still found time to roll out a product. “ChatGPT with voice” is available to free users of the ChatGPT app. This feature launched for paying users in September, and if you haven’t heard, it’s a full-blown voice assistant. The feature is still slowly rolling out to devices; on my Android phone, I don’t have a “headphone” icon anywhere, even with a Plus subscription.
When you have the feature, you can open the app, press the “headphone” icon, ask a question, and a stilted robot voice will read out a reply. It’s just like the voice assistants from Apple, Amazon, or Google, but this one is powered by a large language model. ChatGPT’s voice model is purely a question-and-answer type of voice assistant, though. Usually, these things are handy for what they can do on your behalf—make a phone call, control a smart home, take a note, or make a calendar appointment—but this can only answer questions.
«
“Hey, ChatGPT, exactly why was Sam Altman fired a week ago?”
unique link to this extract
Amazon to cut jobs at Alexa unit to sharpen focus on generative AI • Computerworld
Charlotte Trueman:
»
The company’s emphasis on developing generative AI systems has resulted in the release of new offerings, like the tool rolled out in September to generate copy listings for users selling items on the company’s e-commerce platform. The application is fuelled by a large language model (LLM) that Amazon has been developing internally, originally to support the Alexa smart assistant.
But in a statement Monday, the company said its latest round of job cuts will impact staff working on Alexa-related efforts. “Several hundred roles are impacted, a relatively small percentage of the total number of people in the Devices business who are building great experiences for our customers.”
Amazon noted that there are more than half a billion Alexa devices in customers’ homes. “Our investments in generative AI are bringing our vision for an even more intuitive, intelligent, and useful Alexa closer than ever before,” the company said. “As we continue to invent, we’re shifting some of our efforts to better align with our business priorities, and what we know matters most to customers—which includes maximizing our resources and efforts focused on generative AI. These shifts are leading us to discontinue some initiatives, which is resulting in role eliminations.”
«
Half a billion devices in people’s homes, and yet can’t make money from them. That must be the definition of market failure.
unique link to this extract
Office landlords can’t get a loan anymore • WSJ
Konrad Putzier:
»
The office sector’s credit crunch is intensifying. By one measure, it’s now worse than during the 2008-09 global financial crisis.
Only one out of every three securitized office mortgages that expired during the first nine months of 2023 was paid off by the end of September, according to Moody’s Analytics.
That is the smallest share for the first nine months of any year since at least 2008 and well below the nadir reached in 2009, when 47% of these loans got paid off. That share is also well below the rate before the pandemic, when more than eight out of every 10 maturing securitized office mortgages were paid back in some years.
While the numbers cover only office mortgages packaged into bonds—so-called commercial mortgage-backed securities—they reflect a broader freeze in the lending market for office buildings.
Many office owners can’t pay back their old loans because they can’t get new mortgages. Remote work and rising vacancies have hit building profits, making it harder to pay interest. Higher interest rates have pushed debt costs up and building values down.
That combination is fueling a rise in defaults. The share of office CMBS loans that are delinquent has tripled over the past year to 5.75%, according to Trepp. It doesn’t help that many banks no longer issue new office loans and that many insurance companies and debt funds have become more cautious.
“People just don’t want to touch it,” said Alex Killick, managing director at CWCapital, a company that handles troubled CMBS loans.
«
Quietly, very quietly, the post-pandemic effect begins to be felt.
unique link to this extract
Personal data stolen in British Library cyber-attack appears for sale online • The Guardian
Harriet Sherwood:
»
The British Library has confirmed that personal data stolen in a cyber-attack has appeared online, apparently for sale to the highest bidder.
The attack was carried out on October 31 by a group known for such criminal activity, said the UK’s national library, which holds about 14m books and millions of other items. Since then, the library’s website has been shut down, with updates posted on X and emailed to members
This week, Rhysida, a known ransomware group, claimed it was responsible for the attack. It posted low-resolution images of personal information online, offering stolen data for sale with a starting bid of 20 bitcoins (about £596,000).
Rhysida said the data was “exclusive, unique and impressive” and that it would be sold to a single buyer. It set a deadline for bids of 27 November.
The images appear to show employment contracts and passport information. The library said it was “aware that some data has been leaked, which appears to be from files relating to our internal HR information”. It did not confirm that Rhysida was responsible for the attack, nor that the data offered for sale was information on personnel.
Academics and researchers who use the library have been told that disruption to the institution’s services after the serious ransomware attack was likely to continue for months.
This week, the library advised its users to change any logins also used on other sites as a precaution. It added: “We have taken targeted protective measures to ensure the integrity of our systems, and we continue to undertake an investigation with the support of the National Cyber Security Centre (NCSC), the Metropolitan police and cybersecurity specialists. As this investigation remains ongoing, we cannot provide further details at this time.”
«
Personal details about librarians? Sure that will attract the big bidders. The puzzling point is why the Library hasn’t got some sort of backup of its databases. But as is so often the case, it’s only after you get hit by ransomware that you get smart about this stuff.
unique link to this extract
Where did they all go? How Homo sapiens became the last human species left • The Guardian
Sarah Wild:
»
What we do know is that from about 40,000 years ago, Homo sapiens was the last human standing out of a large and diverse group of bipedal hominins. Hypotheses range from benign, such as H sapiens having better infant survival rates than other hominins, or climate changes pushing other species to the brink. Others suggest a more active role, such as H sapiens hunting other humans or interbreeding with them and assimilating their genetics.
About 300,000 years ago, the first H sapiens populations were springing up in Africa. They didn’t look like modern humans, but they are more similar to us than other Homo species. They had tall, rounded skulls with an almost vertical forehead. They didn’t have the glowering brows of Neanderthals (Homo neanderthalensis) or the protruding jaw of archaic-looking species such as Homo naledi. They also had chins; something that no other Homo species has had (although we don’t know why only H sapiens has the protuberance).
A study published in Nature this year exploded the idea that H sapiens originated from a single place in Africa in one great evolutionary leap. By analysing the genomes of 290 people, the researchers showed that H sapiens descended from at least two populations that lived in Africa for 1m years, before merging in several interactions.
Palaeoanthropologists continue to argue (quite vociferously) over who the last ancestor of H Sapiens was, but so far there is no conclusive evidence. Also, there is no single origin for H sapiens. There are ancient remains of early H sapiens in Jebel Irhoud in Morocco, Omo Kibish in Ethiopia and Florisbad in South Africa, suggesting that our species arose from multiple sites.
When H sapiens moved out of Africa is also the subject of debate. Genetic evidence suggests there was a big foray out of the continent between 80,000 and 60,000 years ago. But it was not the first expedition. A perplexing H sapiens skull in Apidima in Greece has been dated to being at least 210,000 years old.
«
Our ancestry gets more and more complicated. Once it was just Neanderthals. Now it’s loads of rivals.
unique link to this extract
Inside the operation to bring down Trump’s Truth Social • WIRED
David Gilbert:
»
The North Atlantic Fella Organization (NAFO) is an online activist group founded last year to combat pro-Russia propaganda related to the invasion of Ukraine. Last month, the group turned its attention to Trump’s social network and launched a campaign to take over the trending topics section on the website. The group says that the operation, which included 50 “NAFO commandos,” as members targeting Truth Social call themselves, was so successful that those running the campaign now have a long-term goal: Take down Truth Social completely.
“The goal we have in mind, which is lofty, is to help bring the platform down ahead of the 2024 election,” Rock Kenwell, the pseudonymous leader of the NAFO commandos, tells WIRED. “We know it’s going to be an aggregator for extremism and probably violence the way things are looking at this point.”
Describing the Truth Social platform’s current environment, Kenwell compared the challenge of combating the spread of pro-Trump messaging on the platform with “dealing with your racist uncle that nobody wants at the Thanksgiving dinner table because he’s just obnoxious and looking to fight with everybody.”
Truth Social was launched in early 2022 by Trump, who had been kicked off of mainstream platforms for inciting violence. Trump claimed that the network would challenge “Big Tech platforms” like Facebook, Instagram, and Twitter as a free speech platform open to everyone, but in the 18 months since it started, the site has failed to attract anyone outside of Trump sycophants and QAnon conspiracy groups, and has instead become the butt of late night comedy. Last week, a filing to the US Securities and Exchange Commission (SEC) showed that the platform had lost $31m since it launched.
“It’s a very easy platform to manipulate. It’s a very primitive, social media environment,” adds Kenwell.
«
NAFO aims to take over the trending topics section, and claims to have done so with a campaign launched at the end of October which, it claims, led to the suspension of app downloads (to prevent new account registration). The next target is fake ads (though whether these are paid-for fake ads, or just posts that look like ads, isn’t clear.)
One has to wonder though if increasing the (apparent) engagement on the site is really the way you kill it it.
unique link to this extract
Inside Linda Yaccarino’s X all-hands after Elon Musk sued Media Matters • Fortune
Kylie Robison:
»
At the staff meeting Monday, one anonymous employee asked Yaccarino what she would consider the best outcome of the lawsuit [against MEdia Matters]. Yaccarino responded that it “would be the validation that Media Matters, unfortunately manipulates, in this case, not just advertisers but people in general.”
“Sunlight is the best disinfectant. Hopefully this is a disinfectant that surfaces all the great work that all of us are doing at X,” she said.
Both Musk and Yaccarino have defended X on the social media platform in the wake of the report. Musk claimed on Saturday that the watchdog group “created an alternate account and curated the posts and advertising appearing on the account’s timeline to misinform advertisers about the placement of their posts.”
Yaccarino reiterated many of the details in Musk’s post on Monday, telling staff that “this was a contrived experience that could be curated—or this situation could be committed—on any platform today.”
“No platform is gonna get it right,” Yaccarino said. “So they basically gamed the system.”
The X chief executive also sought to calm worries about the advertiser exodus, claiming some brands still doing business with X have told her that they believe that X “is vital for the global community” but want Yaccarino to share more data so they can explain their positions to employees and key stakeholders.
“They know that I work hard on their behalf and that I am a truth teller, and I want them to hear it from me—everything that is going on at the company,” Yaccarino, who has been in the advertising business for several decades, told staff. “And I tell a lot of people this: I didn’t come to the company because I needed the company. I came to the company because I wanted to help lead X and be successful in what we’re trying to achieve here.”
As big brands have paused ads on X, numerous right-wing media companies and influencers have pledged to advertise on X in order to make up for lost revenue. One anonymous staffer asked what should “we do as employees to be more responsible … just to help offset anything that we might be seeing from a loss from advertisers?”
Yaccarino responded that staff should “be as fiscally responsible as possible,” including only expensing “critical and necessary travel.”
«
From my experience at multiple organisations, when the chief exec tells the rank and file to be very careful with the travel expenses, things are financially going very badly.
unique link to this extract
Commercial flights are experiencing ‘unthinkable’ GPS attacks and nobody knows what to do • Vice
Matthew Gault:
»
In late September, multiple commercial flights near Iran went astray after navigation systems went blind. The planes first received spoofed GPS signals, meaning signals designed to fool planes’ systems into thinking they are flying miles away from their real location. One of the aircraft almost flew into Iranian airspace without permission. Since then, air crews discussing the problem online have said it’s only gotten worse, and experts are racing to establish who is behind it.
OPSGROUP, an international group of pilots and flight technicians, sounded the alarm about the incidents in September and began to collect data to share with its members and the public. According to OPSGROUP, multiple commercial aircraft in the Middle Eastern region have lost the ability to navigate after receiving spoofed navigation signals for months. And it’s not just GPS—fallback navigation systems are also corrupted, resulting in total failure.
According to OPSGROUP, the activity is centered in three regions: Baghdad, Cairo, and Tel Aviv. The group has tracked more than 50 incidents in the last five weeks, the group said in a November update, and identified three new and distinct kinds of navigation spoofing incidents, with two arising since the initial reports in September.
While GPS spoofing is not new, the specific vector of these new attacks was previously “unthinkable,” according to OPSGROUP, which described them as exposing a “fundamental flaw in avionics design.” The spoofing corrupts the Inertial Reference System (IRS), a piece of equipment often described as the “brain” of an aircraft that uses gyroscopes, accelerometers, and other tech to help planes navigate.
«
Nobody knows who’s doing it (Syria is a suspect) nor why. Not very encouraging; the sort of thing that could accidentally cause a serious international incident if crews aren’t careful.
unique link to this extract
Behold the Ozempic effect on business • FT
Rana Foroohar:
»
Let’s start with the pharmaceutical firms themselves. If you don’t have an Ozempic knock-off in the development pipeline, your share price may take a hit. Novo Nordisk now has a market capitalisation that is higher than the entire GDP of Denmark, and Eli Lilly’s share price is up 40% since it rolled out its own weight-loss copycat Mounjaro. But both Pfizer and Moderna — neither of which have a successful semaglutide on the market — have seen their share prices plummet in recent months.
And it’s not only companies in the weight-loss business that are being affected. In early October, when Novo Nordisk announced that Ozempic was so effective against kidney disease that it was stopping a trial early, shares in some dialysis providers tanked.
Now, healthcare analysts say that the $250bn cardiovascular disease market could be reduced by 10% by 2050, and hundreds of billions-worth of additional business in treatments for diabetes, kidney and liver disease and other weight related illnesses could be disrupted.
The Ozempic effect doesn’t stop there. Analysts have downgraded doughnut maker Krispy Kreme recently amid worries that Americans on semaglutides just won’t reach for as many sweet treats as they have in the past.
…The new weight-loss drugs will also disrupt the US healthcare system — the only question is how. Semaglutides are expensive, but so is obesity.
One study found that obesity adds $1,861 in annual healthcare costs per American. But if the government decided that Medicare should reimburse for weight-loss drugs (it currently doesn’t) that would create huge costs, too. Insurance companies have long complained about obesity-related costs, but also don’t like the idea of tens of millions of Americans suddenly going on semaglutides.
«
From Napoleon to Elon Musk • The Bluestocking
Helen Lewis has put a transcript on her excellent Substack (you surely must subscribe) of a conversation she had with David Runciman, professor of politics at Cambridge University, about his new book The Handover which is about “the similarities between states, corporations and AIs”. It’s all excellent, though this passage jumped out at me:
»
David Runciman: The great appeal of AI systems, particularly for states and corporations, is their efficiency—and what they mean by “efficiency” is they strip out human error.
But in warfare, mutually assured destruction is a game of chicken. The way you bypass that in a game of chicken is you strip out the human beings and make sure that it’s a robot that presses the button, because the robot won’t have qualms. China and America are currently involved in an arms race which is premised on this. The Chinese state is massively investing in AI weapons technology, precisely because in this great game of chicken that may be played over Taiwan, you want to signal to the other side that you’ve got fewer humans involved in your process than they do.
Helen Lewis: There was a famous thought experiment once that said the nuclear code should be embedded in the heart of a bodyguard, who walks around with the president all the time. And if he wants to launch a nuclear weapon that will kill 80,000 people, maybe he first of all has to carve into the heart of the guy that’s been walking around with him.
David Runciman: You think Donald Trump wouldn’t do that?
«
| • 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 