The Star Wars universe continues to expand on streaming television with the release of the first trailer for Star Wars: Skeleton Crew this weekend at Disney’s annual D23 Expo. The eight-episode standalone series is set in the same time frame as The Mandalorian and Ahsoka.
Executive producer Kathleen Kennedy was intrigued when series co-creator Jon Watts pitched a Star Wars series inspired by the 1985 film The Goonies. (Kennedy had co-produced that film and co-founded Amblin Entertainment.) She told co-creator Christopher Ford that The Goonies hadn’t been created specifically for kids, instead telling a story that just happened to be about kids going on an adventure. So Ford and Watts wrote Skeleton Crew with the same mindset: a show for everyone that just happened to feature kids as the central characters. Per the official premise:
Skeleton Crew follows the journey of four kids who make a mysterious discovery on their seemingly safe home planet, then get lost in a strange and dangerous galaxy, crossing paths with the likes of Jod Na Nawood, the mysterious character played by [Jude] Law. Finding their way home—and meeting unlikely allies and enemies—will be a greater adventure than they ever imagined.
Jude Law leads the cast as the quick-witted and charming (per Law) “Force-user” Jod Na Nawood. Ravi Cabot-Conyers plays Wim, Ryan Kiera Armstrong plays Fern, Kyriana Kratter plays KB, and Robert Timothy Smith plays Neil. Nick Frost will voice a droid named SM 33. The cast also includes Tunde Adebimpe, Kerry Condon, and Jaleel White in as yet undisclosed roles.
The trailer opens with our young protagonists at school, preparing to take a test that will set the course of their respective futures. At least one of them is bored with the daily routine and longs to do something more exciting. “What if we could go anywhere we want in the whole galaxy?” he asks. “A real adventure. No more pretend.” Naturally they find a mysterious “lost Jedi temple” buried in the woods and soon find themselves rocketing away on a spaceship—and getting lost. Can they survive all the dangers of space and find their way back to their home planet? With the help of Law’s Jedi, we like their chances.
Star Wars: Skeleton Crew premieres on December 3, 2024, on Disney+.
Enlarge/ In the experiments, people had to judge what constituted a fair monetary offer.
In many cases, AIs are trained on material that’s either made or curated by humans. As a result, it can become a significant challenge to keep the AI from replicating the biases of those humans and the society they belong to. And the stakes are high, given we’re using AIs to make medical and financial decisions.
But some researchers at Washington University in St. Louis have found an additional wrinkle in these challenges: The people doing the training may potentially change their behavior when they know it can influence the future choices made by an AI. And, in at least some cases, they carry the changed behaviors into situations that don’t involve AI training.
Would you like to play a game?
The work involved getting volunteers to participate in a simple form of game theory. Testers gave two participants a pot of money—$10, in this case. One of the two was then asked to offer some fraction of that money to the other, who could choose to accept or reject the offer. If the offer was rejected, nobody got any money.
From a purely rational economic perspective, people should accept anything they’re offered, since they’ll end up with more money than they would have otherwise. But in reality, people tend to reject offers that deviate too much from a 50/50 split, as they have a sense that a highly imbalanced split is unfair. Their rejection allows them to punish the person who made the unfair offer. While there are some cultural differences in terms of where the split becomes unfair, this effect has been replicated many times, including in the current work.
The twist with the new work, performed by Lauren Treimana, Chien-Ju Hoa, and Wouter Kool, is that they told some of the participants that their partner was an AI, and the results of their interactions with it would be fed back into the system to train its future performance.
This takes something that’s implicit in a purely game-theory-focused setup—that rejecting offers can help partners figure out what sorts of offers are fair—and makes it highly explicit. Participants, or at least the subset involved in the experimental group that are being told they’re training an AI, could readily infer that their actions would influence the AI’s future offers.
The question the researchers were curious about was whether this would influence the behavior of the human participants. They compared this to the behavior of a control group who just participated in the standard game theory test.
Training fairness
Treimana, Hoa, and Kool had pre-registered a number of multivariate analyses that they planned to perform with the data. But these didn’t always produce consistent results between experiments, possibly because there weren’t enough participants to tease out relatively subtle effects with any statistical confidence and possibly because the relatively large number of tests would mean that a few positive results would turn up by chance.
So, we’ll focus on the simplest question that was addressed: Did being told that you were training an AI alter someone’s behavior? This question was asked through a number of experiments that were very similar. (One of the key differences between them was whether the information regarding AI training was displayed with a camera icon, since people will sometimes change their behavior if they’re aware they’re being observed.)
The answer to the question is a clear yes: people will in fact change their behavior when they think they’re training an AI. Through a number of experiments, participants were more likely to reject unfair offers if they were told that their sessions would be used to train an AI. In a few of the experiments, they were also more likely to reject what were considered fair offers (in US populations, the rejection rate goes up dramatically once someone proposes a 70/30 split, meaning $7 goes to the person making the proposal in these experiments). The researchers suspect this is due to people being more likely to reject borderline “fair” offers such as a 60/40 split.
This happened even though rejecting any offer exacts an economic cost on the participants. And people persisted in this behavior even when they were told that they wouldn’t ever interact with the AI after training was complete, meaning they wouldn’t personally benefit from any changes in the AI’s behavior. So here, it appeared that people would make a financial sacrifice to train the AI in a way that would benefit others.
Strikingly, in two of the three experiments that did follow up testing, participants continued to reject offers at a higher rate two days after their participation in the AI training, even when they were told that their actions were no longer being used to train the AI. So, to some extent, participating in AI training seems to have caused them to train themselves to behave differently.
Obviously, this won’t affect every sort of AI training, and a lot of the work that goes into producing material that’s used in training something like a Large Language Model won’t have been done with any awareness that it might be used to train an AI. Still, there’s plenty of cases where humans do get more directly involved in training, so it’s worthwhile being aware that this is another route that can allow biases to creep in.
Enlarge/ US Consumer Product Safety Commission announced a recall of 1.1 million Samsung Slide-in Electric Ranges due to hundreds of reported fires.
After hundreds of reports of fires causing dozens of injuries and several pet deaths, Samsung is recalling more than a million electric stoves sold in the US between 2013 and 2024.
In a press release, the US Consumer Product Safety Commission (CPSC) reported that the voluntary recall was due to “front-mounted knobs” on Samsung’s slide-in electric ranges. The faulty knobs “can be activated by accidental contact by humans or pets, posing a fire hazard”—particularly when people leave objects on the stove.
The stoves impacted by the recall were widely sold in Costco, Home Depot, Best Buy, Lowe’s, and other appliance stores nationwide. Their knobs can be easily triggered by accident, heating up the cooktop and increasing the risks of fires, the CPSC said. Since 2013, Samsung has received “over 300 reports of unintentional activation.” According to the CPSC:
“These ranges have been involved in approximately 250 fires. At least 18 fires caused extensive property damage. Approximately 40 injuries have been reported, eight of which required medical attention, and there have been reports of seven fires involving pet deaths.”
Luckily, there’s an easy solution recently devised that can prevent this safety hazard in homes across America, Samsung said. Customers concerned about unintentional activations can request free knob locks and covers that Samsung confirmed made it much harder to accidentally turn on the stove.
Whereas the problematic electric ranges’ knobs require users to push the knob and turn, “precision knobs” that Samsung rolled out in April introduce a new safety measure that requires users to pinch the knob before pushing and turning knobs to activate the stove.
“A simple pinching motion” releases a pin that otherwise would remain locked and prevent stoves from accidental activation when knobs are unintentionally bumped or perhaps twisted by a young child or knocked around by a pet, Samsung said.
Consumers who bought one of the 30 affected models listed here can contact Samsung online or by phone or email to receive free knob locks and covers and implement this new “pinching” safety measure, even if their warranty is expired. They can also check if their model has been affected here. If the serial number is no longer readable, customers should call or chat online with a Samsung agent.
Once Samsung receives a request for free knob locks and covers, repair kits “should arrive within five business days,” an FAQ said. And customers will receive tracking information once the knob locks and covers ship. Instructions to install will be provided and are also available online.
Until knob locks and covers are installed, customers can continue using their stoves, Samsung said. But the CPSC advised people to “keep children and pets away from the knobs,” “check the range knobs to ensure they are off before leaving the home or going to bed,” and avoid leaving “objects on the range when the range is not in use.”
Additionally, customers with Wi-Fi-enabled ranges can enable notifications in their Samsung SmartThings app to receive alerts when the stove is on.
Samsung noted that parents in particular seemed to appreciate the precision knobs, with one review calling it a “favorite feature,” because “we have two young girls in the house and not having to worry about one of them playing with the knobs and starting the stove… is a huge plus.”
Fire hazards go beyond Samsung—and can be fatal
Samsung’s recall is part of a worrying trend where front-mounted knobs on both gas and electric ranges from many different manufacturers have caused hundreds of fires in the US. In June, the CPSC’s Joint Gas and Electric Range Knob Working Group hosted a meeting with leading stove manufacturers and other stakeholders to confront the industry-wide problem.
During the meeting, the CPSC shared data showing that across 338 incidents between January 1, 2018, and May 30, 2024, stoves from “ten specific manufacturers” were involved in fires causing 31 injuries and two deaths. Additionally, the CPSC had recorded “two other fatal incidents where a range was accidentally turned on when a knob was bumped, but the manufacturer is unknown.”
According to the CPSC, manufacturers were “interested to learn the events that lead to the ranges accidentally activating, including whether pets were involved, unsupervised children were at fault, or there were unusual circumstances.” Companies said the CPSC data would help them “fully understand the issues” and “make sure that reasonable and foreseeable circumstances would be addressed” without impacting compliance with the Americans with Disabilities Act.
Samsung attended the meeting, saying that it joined other “major brands across the appliance industry” to “discuss how to revisit knob safety standards for all ranges to address inadvertent activation.”
The working group’s meetings are expected to continue, but a deadline to reconvene approximately a month after the June meeting has since passed without any further discussion.
A few months prior to the meeting, Samsung introduced the precision knobs as a novel solution in its ranges, as well as an additional safety feature now available in “its most premium Bespoke Slide-In electric and gas ranges,” which illuminates the knobs when they’re turned on. This provides a “visual cue when the knobs are activated,” Samsung said.
As manufacturers like Samsung continue to tweak knobs to improve safety, the CPSC this week issued a safety alert warning the public of fire hazards of gas and electric ranges.
The safety notice advises customers to use safety locks and covers to prevent accidental activation, keep kids and pets away from cooktops with front-mounted knobs, and take care when leaning over the stove to avoid bumping into knobs.
For anyone concerned about safety issues with a gas or electric range, the CPSC provides a database to search additional recalls or otherwise recommends contacting manufacturers directly.
“Consumers who have experienced or have concerns about accidental activation of the front-mounted control knobs on their cooktop should immediately contact the manufacturer of the range to ask if there is a solution or remedy available from the manufacturer,” the CPSC safety alert said.
Enlarge/ Dodge doesn’t want to be accused of building a boring EV.
Dodge
Electric motors have many advantages over internal combustion engines, including the fact that they don’t waste a lot of their power as sound energy. So quiet are electric vehicles, in fact, that federal vehicle safety regulations require EVs (and hybrids) to make a certain amount of noise at lower speeds to warn vulnerable road users like blind or visually impaired pedestrians of their presence.
Almost all of those cars end up sounding like a choir of depressed angels, a phrase memorably coined by either Richard Porter or Jonny Smith on the Smith and Sniff podcast. That’s not the case with the forthcoming electric Dodge Charger, however. When it first broke cover in March, we learned that the electric Charger would feature something called a “Fratzonic Chambered Exhaust” to give it some aural character. Now, thanks to the video from Dodge embedded here, we can all hear what that sounds like in practice:
Dodge knows that emotion is a big part of muscle car sales, so it’s made the electric Charger sound very emotional.
The Fratzonic Chambered Exhaust—the name refers to a Dodge logo—combines various chambers underneath the Charger’s body with some woofers and mid-range speakers, all driven by a dedicated 600 W amplifier. The system is also designed to transmit vibrations into the chassis through elastomeric bushings, mimicking an internal combustion engine and its motor mounts.
“We know our Dodge enthusiasts want that visceral feel you get when you drive a Dodge muscle car, and the Charger’s new Fratzonic system delivers the adrenaline-pumping spirit that they expect. It reacts to specific inputs and driving events, giving the driver a direct connection to their new Charger. Simply put, when you hear it and feel it, you will know it’s a Dodge Charger Daytona,” said Matt McAlear, Dodge brand CEO at Stellantis.
As you can hear in the video, the Fratzonic system makes more than one sound. Dodge says it “intensifies a suite of dynamic vehicle events, including power up/power down, idle/rev, acceleration/throttle, powershot [like a boost function], and deceleration/regenerative braking.”
Dodge is not the first automaker to apply a degree of automotive skeuomorphism in an effort to appeal to the more traditionally minded car enthusiast. Last year, we tested a Toyota fitted with a fake manual transmission, and later this month, we’ll be conducting a more extensive test of the Hyundai Ioniq 5 N, which sonically replicates gearshifts, winning over just about every skeptic who has driven it.
The two-door electric Charger Daytona goes on sale later this year and will cost $59,595 for the Charger Daytona R/T and $73,190 for the Charger Daytona Scat Pack.
Enlarge/ Amina Abakarova allegedly spreading mercury on her rival’s chess board.
Russia is no stranger to unique poisonings. State agents have been known to use everything from polonium-laced tea to the deadly nerve agent “novichok” when making assassination attempts against both defectors in the UK and internal political rivals like Alexei Navalny. But a new “first” in the long history of poisonings was opened this month in the Russian republic of Dagestan, where a 40-something chess player named Amina Abakarova attempted to poison a rival by depositing liquid mercury on and around her chess board.
Malcolm Pein, the English Chess Federation’s director of international chess, told the UK’s Telegraph that he had “never seen anything like this before… This is the first recorded case of somebody using a toxic substance, to my knowledge, in the history of the game of chess.” Usually, he said, chess rivals confine themselves to “psychological” tactics.
Oliver Carroll, a Ukraine war correspondent for The Economist, summed up the situation with some social media snark: “I know that on the standards of Russian doping it’s perhaps only a 7 out of 10. But still…”
Mercury near the Caspian Sea
The strange story began on August 2, when a regional chess tournament was taking place in Makhachkala, a Russian town on the Caspian Sea just north of Azerbaijan. According to the Telegram channel of the Ministry of Internal Affairs of the Republic of Dagestan, emergency services were called after 30-year-old Umayganat Osmanova fell ill during a chess match.
Osmanova said she had seen some tiny gray or silver “beads” rolling out from beneath her side of the chess board, but this apparently didn’t seem odd until she began to feel unwell. A chess.com story translated some of Osmanova’s remarks about what happened. “I still feel bad,” she said. “In the first minutes, I felt a lack of air and a taste of iron in my mouth. I had to spend about five hours on this board. I don’t know what would have happened to me if I hadn’t seen it earlier.”
Such symptoms are consistent with exposure to elemental mercury, the liquid or “quicksilver” version of mercury sometimes used in thermometers. According to the Cleveland Clinic, this form of mercury is “usually harmless if you touch or swallow it because its slippery texture won’t absorb into your skin or intestines.” But if you breathe in any of it, watch out—symptoms occur “immediately” and can include coughing, breathing trouble, nausea, bleeding gums, and a “metallic taste in your mouth.”
Tournament officials consulted security camera footage, where they saw Abakarova walking through a nearly empty room of chess tables some 20 minutes before play was to begin. (In one news account, Abakarova had been asking casually beforehand whether there were any such cameras in the venue.) In the footage, Abakarova walked over to one particular table, pulled a small vial from her bag, and appeared to smear something on the pieces and the table itself.
The security camera footage was soon released onto the Internet, and you can now watch it on YouTube.
Sazhid Sazhidov, the minister of sports of the Republic of Dagestan, posted a note to Telegram after footage of the incident began circulating, saying that “a multiple winner of these competitions, Amina Abakarova from Makhachkala, treated the table at which her opponent—the no less titled European Champion Umayganat Osmanova from Kaspiysk—was to play with an unknown substance which, as it later turned out, was mercury compounds.” (One news outlet claims that Abakarova admitted to police that the mercury she used had come from thermometers.)
Sazhidov noted that he was “perplexed by what happened, and the motives of such an experienced athlete as Amina Abakarova are also unclear to me. The actions she took could have led to the most tragic outcome, they threatened the lives of everyone who was in the chess house, including herself. Now she will have to answer for what she did before the law.”
The president of the Russian Chess Federation, Andrey Filatov, temporarily suspended Abakarova from competitions until the conclusion of the case against her, at which point she could face a lifetime ban from competitive chess.
The source of Abakarova’s dispute with Osmanova remains unclear. One Russian news outlet said that the two had known each other for years but had recently fought. In this version of the story, Abakarova showed up to one recent match with a phone, which is against the rules. Osmanova was upset but did not tell the judges. “She should have been grateful to me that I didn’t make a fuss and forgave her,” Osmanova said. “Instead, Amina refused to shake my hand during the competition last week.”
Chessbase said the dispute was over a recent match between the two in which “both chess players scored the same number of points, but the victory was awarded to Osmanova, based on additional factors.”
Another Telegram channel says that the issue was about negative statements made by Osmanova about Abakarova and her family members.
Whatever the source of the conflict, Abakarova now faces the prospect of several years in prison—and the world of chess just got meaner.
Humane AI Pins were returned at faster rate than they were sold between May and August, according to a report from The Verge on Wednesday. The AI gadget released in April to abysmal reviews, and Humane is now reportedly dealing with over $1,000,000 worth of returned product.
The AI Pin is a lapel pin that markets numerous features—like an AI voice assistant, camera, and laser projector—which its creators claim will replace smartphones as a go-to gadget. It costs $700 and requires a subscription that costs $24 per month, not including taxes and fees, for cloud storage, cellular data, and a number.
In June, The New York Times, citing two anonymous sources, reported that Humane had sold 10,000 of its AI devices. But today, only 7,000 sold units have not been returned, The Verge reported yesterday, citing someone “with direct knowledge.” The Verge said it viewed internal sales data showing returns outpacing device/accessory sales of about $9,000,000. Internal data also reportedly revealed that 1,000 AI Pin orders were canceled before they even shipped.
Humane didn’t respond to Ars Technica’s request for comment. Company spokesperson Zoz Cuccias told The Verge that there were inaccuracies in The Verge’s report, “including the financial data.” However, Cuccias declined to share specifics with the publication, saying that Humane has “nothing else to provide as we do not comment on financial data and will refer it to our legal counsel.”
Reportedly exacerbating the problem is that there is currently no way to refurbish and resell the pins. That would mean that thousands of AI Pins are currently sitting as e-waste until the problem is addressed. According to The Verge, problems stem from the pins’ connection to T-Mobile service, which prevents Humane from reassigning returned pins. T-Mobile hasn’t commented on the issue, but an anonymous source told The Verge that Humane is holding on to returned pins in hopes of “eventually” finding a solution.
As a new device category, there was already concern about AI gadgets like the AI Pin or Rabbit R1 becoming e-waste. Worries about the ability of the devices’ companies to last and questions over whether these gadgets would be better as apps suggest that even if Humane found a way to reassign thousands of returned devices, we could still eventually be dealing with a massive pile of obsolete AI Pins.
And there’s plenty of reason to be concerned about Humane’s survival.
Horrible reviews from the start
Humane had hoped to sell about 100,000 units during the device’s first year of availability, an anonymous source told the NYT in June. The alarming sales and return figures reported by the Verge come after the company’s founders, two former Apple employees, accrued a reported $240 million in funding.
As detailed by the NYT in June, sources close to the AI Pin claimed that Humane’s cofounders ignored poor internal reviews and forced the product’s release despite concerns about heat and battery life. In June, Humane warned users against using the pin’s charging case due to a fire risk. Speaking to The Verge this week, Cuccias acknowledged that Humane “knew we were at the starting line, not the finish line” when it released the AI Pin. The company rep noted software updates that have come out in response to negative feedback.
Elon Musk’s X Corp. today sued the World Federation of Advertisers and several large corporations, claiming they “conspired, along with dozens of non-defendant co-conspirators, to collectively withhold billions of dollars in advertising revenue” from the social network formerly known as Twitter.
“We tried peace for 2 years, now it is war,” Musk wrote today, a little over eight months after telling boycotting advertisers to “go fuck yourself.”
X’s lawsuit in US District Court for the Northern District of Texas targets a World Federation of Advertisers initiative called the Global Alliance for Responsible Media (GARM). The other defendants are Unilever PLC; Unilever United States; Mars, Incorporated; CVS Health Corporation; and Ørsted A/S. Those companies are all members of GARM. X itself is still listed as one of the group’s members.
“This is an antitrust action relating to a group boycott by competing advertisers of one of the most popular social media platforms in the United States… Concerned that Twitter might deviate from certain brand safety standards for advertising on social media platforms set through GARM, the conspirators collectively acted to enforce Twitter’s adherence to those standards through the boycott,” the lawsuit said.
The lawsuit seeks treble damages to be calculated based on the “actual damages in an amount to be determined at trial.” X also wants “a permanent injunction under Section 16 of the Clayton Act, enjoining Defendants from continuing to conspire with respect to the purchase of advertising from Plaintiff.”
The lawsuit came several weeks after Musk wrote that X “has no choice but to file suit against the perpetrators and collaborators in the advertising boycott racket,” and called for “criminal prosecution.” Musk’s complaints were buoyed by a House Judiciary Committee report claiming that “the extent to which GARM has organized its trade association and coordinates actions that rob consumers of choices is likely illegal under the antitrust laws and threatens fundamental American freedoms.”
Yaccarino claims “illegal boycott” is stain on industry
We contacted all of the organizations named as defendants in the lawsuit and will update this article if any provide a response.
An advertising industry watchdog group called the Check My Ads Institute, which is not involved in the lawsuit, said that Musk’s claims should fail under the First Amendment. “Advertisers have a First Amendment right to choose who and what they want to be associated with… Elon Musk and X executives have the right, protected by the First Amendment, to say what they want online, even when it’s inaccurate, and advertisers have the right to keep their ads away from it,” the group said.
X CEO Linda Yaccarino posted an open letter to advertisers claiming that the alleged “illegal boycott” is “a stain on a great industry, and cannot be allowed to continue.”
“The illegal behavior of these organizations and their executives cost X billions of dollars… To those who broke the law, we say enough is enough. We are compelled to seek justice for the harm that has been done by these and potentially additional defendants, depending what the legal process reveals,” Yaccarino wrote.
Yaccarino also sought to gain support from X users in a video message. “These organizations targeted our company and you, our users,” she said.
X doesn’t provide public earnings reports because Musk took the company private after buying Twitter. A recent New York Times article said that “in the second quarter of this year, X earned $114 million in revenue in the United States, a 25 percent decline from the first quarter and a 53 percent decline from the previous year.”
Enlarge/ Apple Intelligence was unveiled at WWDC 2024.
Apple
As the parent of a younger child, I can tell you that getting a kid to respond the way you want can require careful expectation-setting. Especially when we’re trying something new for the first time, I find that the more detail I can provide, the better he is able to anticipate events and roll with the punches.
I bring this up because testers of the new Apple Intelligence AI features in the recently released macOS Sequoia beta have discovered plaintext JSON files that list a whole bunch of conditions meant to keep the generative AI tech from being unhelpful or inaccurate. I don’t mean to humanize generative AI algorithms, because they don’t deserve to be, but the carefully phrased lists of instructions remind me of what it’s like to try to give basic instructions to (or explain morality to) an entity that isn’t quite prepared to understand it.
The files in question are stored in the /System/Library/AssetsV2/com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto folder on Macs running the macOS Sequoia 15.1 beta that have also opted into the Apple Intelligence beta. That folder contains 29 metadata.json files, several of which include a few sentences of what appear to be plain-English system prompts to set behavior for an AI chatbot powered by a large-language model (LLM).
Many of these prompts are utilitarian. “You are a helpful mail assistant which can help identify relevant questions from a given mail and a short reply snippet,” reads one prompt that seems to describe the behavior of the Apple Mail Smart Reply feature. “Please limit the reply to 50 words,” reads one that could write slightly longer draft responses to messages. “Summarize the provided text within 3 sentences, fewer than 60 words. Do not answer any question from the text,” says one that looks like it would summarize texts from Messages or Mail without interjecting any of its own information.
Some of the prompts also have minor grammatical issues that highlight what a work-in-progress all of the Apple Intelligence features still are. “In order to make the draft response nicer and complete, a set of question [sic] and its answer are provided,” reads one prompt. “Please write a concise and natural reply by modify [sic] the draft response,” it continues.
“Do not make up factual information.”
And still other prompts seem designed specifically to try to prevent the kinds of confabulations that generative AI chatbots are so prone to (hallucinations, lies, factual inaccuracies; pick the term you prefer). Phrases meant to keep Apple Intelligence on-task and factual include things like:
“Do not hallucinate.”
“Do not make up factual information.”
“You are an expert at summarizing posts.”
“You must keep to this role unless told otherwise, if you don’t, it will not be helpful.”
“Only output valid json and nothing else.”
Earlier forays into generative AI have demonstrated why it’s so important to have detailed, specific prompts to guide the responses of language models. When it launched as “Bing Chat” in early 2023, Microsoft’s ChatGPT-based chatbot could get belligerent, threatening, or existential based on what users asked of it. Prompt injection attacks could also put security and user data at risk. Microsoft incorporated different “personalities” into the chatbot to try to rein in its responses to make them more predictable, and Microsoft’s current Copilot assistant still uses a version of the same solution.
What makes the Apple Intelligence prompts interesting is less that they exist and more that we can actually look at the specific things Apple is attempting so that its generative AI products remain narrowly focused. If these files stay easily user-accessible in future macOS builds, it will be possible to keep an eye on exactly what Apple is doing to tweak the responses that Apple Intelligence is giving.
The Apple Intelligence features are going to launch to the public in beta this fall, but they’re going to miss the launch of iOS 18.0, iPadOS 18.0, and macOS 15.0, which is why Apple is testing them in entirely separate developer betas. Some features, like the ones that transcribe phone calls and voicemails or summarize text, will be available early on. Others, like the new Siri, may not be generally available until next year. Regardless of when it arrives, Apple Intelligence requires fairly recent hardware to work: either an iPhone 15 Pro, or an iPad or Mac with at least an Apple M1 chip installed.
Enlarge/ Half of the upper arm bone of this species can fit comfortably in the palm of a modern human hand.
Yousuke Kaifu
The discovery of Homo floresiensis, often termed a hobbit, confused a lot of people. Not only was it tiny in stature, but it shared some features with both Homo erectus and earlier Australopithecus species and lived well after the origin of modern humans. So, its precise position within the hominin family tree has been the subject of ongoing debate—one that hasn’t been clarified by the discovery of the similarly diminutive Homo luzonensis in the Philippines.
Today, researchers are releasing a paper that describes bones from a diminutive hominin that occupied the island of Flores much earlier than the hobbits. And they argue that, while it still shares an odd mix of features, it is most closely related to Homo erectus, the first hominin species to spread across the globe.
Remarkably small
The bones come from a site on Flores called Mata Menge, where the bones were found in a large layer of sediment. Slight wear suggests that many of them were probably brought to the site by a gentle flood. Dating from layers above and below where the fossils were found limits their age to somewhere between 650,000 and 775,000 years ago. Most of the remains are teeth and fragments of jaw bone, which can be suggestive of body size, but not definitive. But the new finds include a fragment of the upper arm bone, the humerus, which is more directly proportional to body size.
The researchers argue that the bone is broken at roughly the mid-point of the humerus, meaning that the full-sized bone was twice its length. Based on the relationship between humerus length and body size, they estimate that the individual it came from was only a bit above a meter tall.
They also took a slice from the center of the sample and imaged the cells present in the bone when it fossilized. These suggest that the fossil came from a fully mature adult. That makes its dimensions, including the diameter of the bone, the smallest yet found. It is, to quote the paper, “smaller than LB1 (H. floresiensis) and any other adult individuals of small-bodied fossil hominins (Australopithecus and H. naledi.” So, even by the standards of small species, the new fossils belong to an extremely small individual.
As for what these individuals are related to, the answers are (once again) complicated. The morphology of the humerus is most closely related to the H. floresiensis individuals who resided on Flores hundreds of thousands of years later. Beyond that, it’s most similar to H. naledi. From there, its shape appears to be equally distant from various species, including both H. erectus and various species of Australopithecus. The teeth show a variety of affinities but are generally closest to members of the Homo genus.
So, the authors make two arguments. One is that the fossils come from the ancestors of the hobbits and belong to the same species, indicating that they inhabited Flores for at least half a million years. The second is that it’s a branch off the population of H. erectus, a species that was similar in stature to modern humans. The population would have evolved a shorter stature once isolated on Flores.
Nothing makes a lot of sense
That’s the argument, at least. There will undoubtedly be different opinions among paleontologists, however. Some had already argued that H. floresiensis was an offshoot of H. erectus and will be happy to accept this as new evidence. But the species is such a hodge-podge of features of earlier and contemporary species that it has been easy for others to make contrary arguments.
Even if those arguments were settled, there’s the issue of how it got there. Even at times of significantly lower sea levels, Flores would have required a significant ocean crossing from what is now Java, where H. erectus is known to have been present, and which was connected to Asia at the time. There’s no indication that any species that came before modern humans had developed boating technology, and some have suggested that the population was established on Flores after being swept there on tsunami debris. Once present, the island environment could have selected for a smaller body size.
But then there’s the issue of Homo luzonensis, which shared a similar body size but inhabited a very different island. That would seem to require a second event that was also unlikely: either a second ocean passage involving individuals from Flores or another ocean trip by H. erectus followed by similar evolution of smaller body size, despite a potentially different environment.
It’s clear that, while the new finds tell us something about the Flores population, they’re not going to settle any arguments.
Google Chrome’s long, long project to implement a new browser extension platform is seemingly going to happen, for real, after six years of cautious movement.
One of the first ways people are seeing this is if they use uBlock Origin, a popular ad-blocking extension, as noted by Bleeping Computer. Recently, Chrome users have seen warnings pop up that “This extension may soon no longer be supported,” with links asking the user to “Remove or replace it with similar extensions” from Chrome’s Web Store. You might see a similar warning on some extensions if you head to Chrome’s Extensions page (chrome://extensions).
What’s happening is Chrome preparing to make Manifest V3 required for extensions that want to run on its platform. First announced in 2018, the last word on Manifest V3 was that V2 extensions would start being nudged out in early June on the Beta, Dev, and Canary update channels. Users will be able to manually re-enable V2 extensions “for a short time,” Google has said, “but over time, this toggle will go away as well.” The shift for enterprise Chrome deployments is expected to be put off until June 2025.
Google has said that its new extension platform was built for “improving the security, privacy, performance, and trustworthiness of the extension ecosystem.” The Electronic Frontier Foundation (EFF) disagrees most strongly with the security aspect, and Firefox-maker Mozilla, while intending to support V3 extensions for cross-browser compatibility, has no plans to cut off support for V2 extensions, signaling that it doesn’t see the big improvement.
Perhaps the biggest point of friction is with ad blockers. Google has said it “isn’t killing ad blockers” but “making them safer,” in an explanatory blog post. Google noted in November 2023 that Manifest V3 allowed for a greater number, and more dynamic updating, of content-blocking rules in extensions, specifically ad blockers.
But one of the biggest changes is in disallowing “remotely hosted code,” which includes the filtering lists that ad blockers keep regularly updated. Ad blockers that want to update their filtering lists, perhaps in response to pivots by platforms like Google’s YouTube and ad servers, will have to do so through the Chrome Web Store’s review process. Ad-blocking coders see it as an intentional gatekeeping and slowing.
Google said before the initial May push toward V3 that 85 percent of actively maintained extensions in its store had Manifest V3 versions ready. Raymond Hill wrote on uBlock Origin’s GitHub page Friday that there will not be a full version of uBlock Origin that works with Manifest V3, but instead a “Lite” version that is “a pared-down version of uBO with a best effort at converting filter lists used by uBO into a Manifest V3-compliant approach.”
Welcome to Edition 7.05 of the Rocket Report! The Federal Aviation Administration grounded SpaceX’s Falcon 9 rocket for 15 days after a rare failure of its upper stage earlier this month. The FAA gave the green light for Falcon 9 to return to flight July 25, and within a couple of days, SpaceX successfully launched three missions from three launch pads. There’s a lot on Falcon 9’s to-do list, so we expect SpaceX to quickly return to form with several flights per week.
As always, we welcome reader submissions. If you don’t want to miss an issue, please subscribe using the box below (the form will not appear on AMP-enabled versions of the site). Each report will include information on small-, medium-, and heavy-lift rockets as well as a quick look ahead at the next three launches on the calendar.
Big delay for a reusable rocket testbed. The French space agency, CNES, has revealed that the inaugural test flight of its Callisto reusable rocket demonstrator will not take place until late 2025 or early 2026, European Spaceflight reports. CNES unveiled an updated website for the Callisto rocket program earlier this month, showing the test rocket has been delayed from a debut launch later this year to until late 2025 or early 2026. The Callisto rocket is designed to test techniques and technologies required for reusable rockets, such as vertical takeoff and vertical landing, with suborbital flights from the Guiana Space Center in South America.
Cooperative action … Callisto, which stands for Cooperative Action Leading to Launcher Innovation in Stage Toss-back Operations, is a joint project between CNES, German space agency DLR, and JAXA, the Japanese space agency. It will stand 14 meters (46 feet) tall and weigh about 4 metric tons (8,800 pounds), with an engine supplied by Japan. Callisto is one of several test projects in Europe aiming to pave the way for a future reusable rocket. (submitted by EllPeaTea and Ken the Bin)
The easiest way to keep up with Eric Berger’s space reporting is to sign up for his newsletter, we’ll collect his stories in your inbox.
Small step for Themis. Another European project established to demonstrate reusable rocket tech is making slow progress toward a first flight. The Themis project, funded by the European Space Agency, is similar in purpose to the Callisto testbed discussed above. This week, the German aerospace manufacturing company MT Aerospace announced it has begun testing a demonstrator of the landing legs that will be used aboard the Themis reusable booster, European Spaceflight reports. The landing legs for Themis are made of carbon fiber-reinforced plastic composites, and the initial test demonstrated good deployment and showed it would withstand the impact energy of landing.
Also delayed … Like Callisto, Themis is facing delays in getting to the launch pad. ArianeGroup, the ESA-selected Themis prime contractor, had been expected to conduct an initial hop test of the demonstrator before the end of 2024. However, officials have announced the initial hop tests won’t happen until sometime next year. The Themis booster is intended to eventually become the first stage booster for an orbital-class partially reusable rocket being developed by MaiaSpace, a subsidiary of ArianeGroup. (submitted by Ken the Bin)
Falcon 9 is flying again. A SpaceX Falcon 9 rocket returned to flight on July 27, barely two weeks after an upper stage failure ended a streak of more than 300 consecutive successful launches, Ars reports. By some measures this was an extremely routine mission—it was, after all, SpaceX’s 73rd launch of this calendar year. And like many other Falcon 9 launches this year, the “Starlink 10-9” mission carried 23 of the broadband Internet satellites into orbit. However, after a rare failure earlier this month, this particular Falcon 9 rocket was making a return-to-flight for the company and attempting to get the world’s most active booster back into service.
Best part is no part … The Falcon 9 successfully deployed its payload of Starlink satellites about an hour after lifting off from NASA’s Kennedy Space Center in Florida. Later in the weekend, SpaceX launched two more Starlink missions on Falcon 9 rockets from Florida and California, notching three flights in less than 28 hours. The launch failure on the previous Falcon 9 launch was caused by a liquid oxygen leak on the upper stage, which led to a “hard start” on the upper stage engine when it attempted to reignite in flight. Engineers and technicians were quickly able to pinpoint the cause of the leak, a crack in a “sense line” for a pressure sensor attached to the vehicle’s liquid oxygen system.
Atlas V’s NSSL era is over. United Launch Alliance delivered a classified US military payload to orbit Tuesday for the last time with an Atlas V rocket, ending the Pentagon’s use of Russian rocket engines as national security missions transition to all-American launchers, Ars reports. This was the 101st launch of an Atlas V rocket since its debut in 2002, and the 58th and final Atlas V mission with a US national security payload since 2007. The Atlas V is powered by an RD-180 main engine made in Russia, and with a little prodding from SpaceX (via a lawsuit) and Congress, the Pentagon started making moves to end its reliance on the RD-180 a decade ago.
Other options available … The RD-180 never failed on a National Security Space Launch (NSSL) mission using the Atlas V rocket, but its use became politically untenable after Russia’s annexation of Crimea in 2014, which predated Russia’s full-scale invasion of Ukraine eight years later. SpaceX began launching US military missions in 2018, and ULA debuted its new Vulcan rocket in January. Assuming a successful second test flight of Vulcan in September, ULA’s next-generation rocket has a good shot at launching its first national security mission by the end of the year. The Space Force’s policy is to maintain at least two independent launch vehicles capable of flying military payloads into orbit. Vulcan and SpaceX’s Falcon rocket family fulfill that requirement, so the military no longer needs the Atlas V. However, 15 more Atlas V rockets remain in ULA’s inventory for future commercial flights.
Crackdown at the Cape. While this week’s landmark launch of the Atlas V rocket is worthy of celebration, there’s a new ULA policy that deserves ridicule, Ars reports. Many of the spectacular photos of rocket launches shared on social media come from independent photographers, who often make little to no money working for an established media organization. Instead, they rely on sales of prints to recoup at least some of their expenses for gas, food, and camera equipment needed to capture these images, which often serve as free publicity for launch providers like ULA. Last month, ULA announced it will no longer permit these photographers to set up remote cameras at their launch pads if they sell their images independently. This new policy was in place for the Atlas V launch from Cape Canaveral, Florida, on Tuesday morning.
But why? … “ULA will periodically confirm editorial publication for media participating in remote camera placement,” ULA stated in an email distributed to photographers last month. “If publication does not occur, or photos are sold outside of editorial purposes, privileges to place remote cameras may be revoked.” To the photographers who spend many hours preparing their equipment, waiting to set up and remove cameras, and persevering through scrubs and more, it seemed like a harsh judgment. And nobody knows why it happened. ULA has offered no public comment about the new policy, and the company did not respond to questions from Ars about the agreement.
Astroscale achieves a first in orbit. There are more than 2,000 mostly intact dead rockets circling the Earth, but until this year, no one ever launched a satellite to go see what one looked like after many years of tumbling around the planet, Ars reports. A Japanese company named Astroscale launched a small satellite in February to chase down the derelict upper stage from a Japanese H-IIA rocket. Astroscale’s ADRAS-J spacecraft arrived near the H-IIA upper stage in April, and the company announced this week that its satellite has now completed two 360-degree fly-arounds of the rocket. This is the first time a satellite has maneuvered around an actual piece of space junk, and it offers an unprecedented snapshot of how an abandoned rocket holds up to 15 years in the harsh environment of space.
Prepping for the future … Astroscale’s ADRAS-J mission is partially funded by the Japan Aerospace Exploration Agency (JAXA). Astroscale and JAXA also have a contract for a follow-up mission named ADRAS-J2, which will attempt to link up with the same H-IIA rocket and steer it on a trajectory to burn up in Earth’s atmosphere. This would be the first demonstration of active debris removal, a concept pursued by Astroscale and other companies to help clear space junk out of low-Earth orbit.
An update on Ariane 6. The European Space Agency has released its first update on the results from the first flight of the Ariane 6 rocket since its launch July 9. Europe’s new flagship rocket had a mostly successful inaugural test flight. Its first stage, solid rocket boosters, and upper stage performed as expected for the first phase of the flight, delivering eight small satellites into an on-target orbit. The launch pad at the Guiana Space Center in South America also held up to the violent environment of launch, ESA said.
Still investigating … However, the final phase of the mission didn’t go according to plan. The upper stage’s Vinci engine was supposed to reignite for a third time on the test flight to deorbit the rocket, which would have released two small reentry capsules on technology demonstration missions to test heat shield technologies. This didn’t happen. An Auxiliary Propulsion Unit, which is a small engine to provide additional bursts of thrust and pressurize the upper stage’s propellant tanks, shut down shortly after startup ahead of the third burn of the primary Vinci engine. “This meant the Vinci engine’s third boost could not take place,” ESA said. “Analysis of the APU’s behavior is ongoing and further information will be made available as soon as possible, while the next task force update is expected in September.” (submitted by Ken the Bin)
Room to grow at Starbase. SpaceX has since launched Starship four times from its launch site in South Texas, known as Starbase, and is planning a fifth launch within the next two months, Ars reports. However, as it continues to test Starship and make plans for regular flights, SpaceX will need a higher flight rate. This is especially true as the company is unlikely to activate additional launch pads for Starship in Florida until at least 2026. To that end, SpaceX has asked the FAA for permission for up to 25 flights a year from South Texas, as well as the capability to land both the Starship upper stage and Super Heavy booster stage back at the launch site.
The answer is probably yes … On Monday, the FAA signaled that it is inclined to grant this request. The agency released a draft assessment indicating that its extensive 2022 analysis of Starship launch activities on the environment, wildlife, local communities, and more was sufficient to account for SpaceX’s proposal for more launches. There is more to do for this conclusion to become official, including public meetings and a public comment period this month.
SpaceX eyes Australia. SpaceX is in talks with US and Australian officials to land and recover one of its Starship rockets off Australia’s coast, a possible first step toward a bigger presence for Elon Musk’s company in the region as the two countries bolster security ties, Reuters reports. At the end of SpaceX’s fourth Starship test flight in June, the rocket made a controlled splashdown in the Indian Ocean hundreds of miles off the northwest coast of Australia. The discussions now underway are focused on the possibility of towing a future Starship vehicle from its splashdown point in the ocean to a port in Australia, where SpaceX engineers could inspect it and learn more about how it performed.
Eventually, it’ll come back to land … On the next Starship flight, currently planned for no earlier than late August, SpaceX plans to attempt to recover Starship’s giant Super Heavy booster using catch arms on the launch pad tower in Texas. On Sunday, Elon Musk told SpaceX and Tesla enthusiasts at an event called the “X Takeover” that it will take a few more flights for engineers to get comfortable returning the Starship itself to a landing onshore. “We want to be really confident that the ship heat shield is super robust and lands at the exact right location,” he said. “So before we try to bring the ship back to the launch site, we probably want to have at least three successful landings of the ship [at sea].” (submitted by Ken the Bin)
Next three launches
August 2: Electron | “Owl for One, One for Owl” | Mahia Peninsula, New Zealand | 16: 39 UTC
August 3: Falcon 9 | NG-21 | Cape Canaveral Space Force Station, Florida | 15: 28 UTC
August 4: Falcon 9 | Starlink 11-1 | Vandenberg Space Force Base, California | 07: 00 UTC
A small study in Texas suggests that human bird flu cases are being missed on dairy farms where the H5N1 virus has taken off in cows, sparking an unprecedented nationwide outbreak.
The finding adds some data to what many experts have suspected amid the outbreak. But the authors of the study, led by researchers at the University of Texas Medical Branch in Galveston, went further, stating bluntly why the US is failing to fully surveil, let alone contain, a virus with pandemic potential.
“Due to fears that research might damage dairy businesses, studies like this one have been few,” the authors write in the topline summary of their study, which was posted online as a pre-print and had not been peer-reviewed.
The study authors, led by Gregory Gray, were invited to two undisclosed dairy farms in Texas that experienced H5N1 outbreaks in their herds starting in early and late March, respectively. The researchers had a previously approved research protocol to study novel respiratory viruses on dairy farms, easing the ability to quickly begin the work.
Rare study
“Farm A” had 7,200 cows and 180 workers. Illnesses began on March 6, and nearly 5 percent of the herd was estimated to be affected during the outbreak. “Farm B” had 8,200 cows and 45 workers. After illnesses began on March 20, an estimated 14 percent of the herd was affected.
The researchers first visited Farm A on April 3 and Farm B on April 4, collecting swabs and samples at each. Based on the previously approved protocol, they were limited to taking nasal swabs and blood samples from no more than 10 workers per farm. On Farm A, 10 workers provided nasal swabs and blood samples. On Farm B, only seven agreed to give nasal swabs, and four gave blood samples.
While swabs from cows, milk, a dead bird, and a sample of fecal slurry showed signs of H5N1, all of the nasal swabs from the 14 humans were negative. However, when researchers looked for H5N1-targeting antibodies in their blood—an indicator that they were previously infected—two of the 14, about 14 percent, were positive.
Both of the workers with previous infections, a man and a woman, were from Farm A. And both reported having flu-like symptoms. The man worked inside cattle corrals, close to the animals, and he reported having a cough at the time the samples were taken. The woman, meanwhile, worked in the cafeteria on the farm and reported recently recovering from an illness that included fever, cough, and sore throat. She noted that other people on the farm had similar respiratory illnesses around when she did.
The finding suggests human cases of H5N1 are going undetected. Moreover, managing to find evidence of two undetected infections in a sample of just 14 workers suggests it may not be hard to find more. The Centers for Disease Control and Prevention estimates that around 200,000 people work with livestock in the US.
Known infections in humans have all been mild so far. But experts are anxious that with each new infection, the wily H5N1 virus is getting new opportunities to adapt further to humans. If the virus evolves to cause more severe disease and spread from human to human, it could spark another pandemic.
Federal officials are also worried about this potential threat. In a press briefing Tuesday, Nirav Shah, the CDC’s principal deputy director, announced a $5 million effort to vaccinate farm workers—but against seasonal flu.
Shah explained that the CDC is concerned that if farm workers are infected with H5N1 and the seasonal flu at the same time, the viruses could exchange genetic segments—a process called reassortment. This could give rise to the pandemic threat experts are worried about. By vaccinating the workers against the seasonal flu, it could potentially prevent the viruses from comingling in one person, Shah suggested.
The US does have a bird flu-specific vaccine available. But in the briefing, Shah said that the use of that vaccine in farm workers is not planned for now, though there’s still active discussion on the possibility. The lack of severe disease and no documented human-to-human transmission from H5N1 infections both argue against deploying a new vaccine, Shah said. “There has to be a strong and compelling case,” he added. Shah also suggested that the agency expects vaccine uptake to be low among farm workers.
