Connect with us

Tech

Anthropic’s Claude Takes Control of a Robot Dog

Published

on

Anthropic’s Claude Takes Control of a Robot Dog


As more robots start showing up in warehouses, offices, and even people’s homes, the idea of large language models hacking into complex systems sounds like the stuff of sci-fi nightmares. So, naturally, Anthropic researchers were eager to see what would happen if Claude tried taking control of a robot—in this case, a robot dog.

In a new study, Anthropic researchers found that Claude was able to automate much of the work involved in programming a robot and getting it to do physical tasks. On one level, their findings show the agentic coding abilities of modern AI models. On another, they hint at how these systems may start to extend into the physical realm as models master more aspects of coding and get better at interacting with software—and physical objects as well.

“We have the suspicion that the next step for AI models is to start reaching out into the world and affecting the world more broadly,” Logan Graham, a member of Anthropic’s red team, which studies models for potential risks, tells WIRED. “This will really require models to interface more with robots.”

Courtesy of Anthropic

Courtesy of Anthropic

Anthropic was founded in 2021 by former OpenAI staffers who believed that AI might become problematic—even dangerous—as it advances. Today’s models are not smart enough to take full control of a robot, Graham says, but future models might be. He says that studying how people leverage LLMs to program robots could help the industry prepare for the idea of “models eventually self-embodying,” referring to the idea that AI may someday operate physical systems.

It is still unclear why an AI model would decide to take control of a robot—let alone do something malevolent with it. But speculating about the worst-case scenario is part of Anthropic’s brand, and it helps position the company as a key player in the responsible AI movement.

In the experiment, dubbed Project Fetch, Anthropic asked two groups of researchers without previous robotics experience to take control of a robot dog, the Unitree Go2 quadruped, and program it to do specific activities. The teams were given access to a controller, then asked to complete increasingly complex tasks. One group was using Claude’s coding model—the other was writing code without AI assistance. The group using Claude was able to complete some—though not all—tasks faster than the human-only programming group. For example, it was able to get the robot to walk around and find a beach ball, something that the human-only group could not figure out.

Anthropic also studied the collaboration dynamics in both teams by recording and analyzing their interactions. They found that the group without access to Claude exhibited more negative sentiments and confusion. This might be because Claude made it quicker to connect to the robot and coded an easier-to-use interface.

Courtesy of Anthropic

The Go2 robot used in Anthropic’s experiments costs $16,900—relatively cheap, by robot standards. It is typically deployed in industries like construction and manufacturing to perform remote inspections and security patrols. The robot is able to walk autonomously but generally relies on high-level software commands or a person operating a controller. Go2 is made by Unitree, which is based in Hangzhou, China. Its AI systems are currently the most popular on the market, according to a recent report by SemiAnalysis.

The large language models that power ChatGPT and other clever chatbots typically generate text or images in response to a prompt. More recently, these systems have become adept at generating code and operating software—turning them into agents rather than just text-generators.



Source link

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Tech

The Catastrophic Swatch x Audemars Piguet Launch Was Entirely Predictable and Utterly Avoidable

Published

on

The Catastrophic Swatch x Audemars Piguet Launch Was Entirely Predictable and Utterly Avoidable


The note from the communications team then, quite remarkably, lists some stats in an attempt to paint the launch in a positive light, as opposed the retail bin-fire it seemingly was: “We have received millions of clicks on our website. This new collaboration is literally making social media explode, with over 6 billion views within one week; by now, it is already 11 billion. All in all, the Royal Pop Collection is captivating the entire world, not least because the Royal Pop is, quite surprisingly, not a wristwatch.”

Audemars Piguet seems unhappy with how Swatch has handled the launch of its collaboration on the Royal Pop. AP told WIRED that “we understand the questions around the Royal Pop launch experience. As retail operations are handled by Swatch and their local teams, Swatch is best placed to comment on the operational handling of the launch. From AP’s perspective, safety and a positive experience for clients and teams remain the priority.” The brand did not respond when asked if it considered Swatch’s handling of the Royal Pop launch a “safe and positive experience”.

The madness of the Royal Pop launch is that, considering all that could have been learned from the MoonSwatch release in 2022, Swatch decided to repeat the playbook that went so badly wrong four years ago. This is a move, according to experts, that was entirely avoidable and utterly unnecessary.

Hype With No Control

“Luxury drops cannot rely on surprise, scarcity and social frenzy as the strategy, then act surprised when human behaviour follows,” says Kate Hardcastle, author of The Science of Shopping and advisor to brands including Disney, Mastercard, Klarna and American Express. “Retailers are already dealing with heightened tensions around theft, aggression and crowd management globally. Add a highly restricted product, long queues, resale economics, social media amplification and the emotional intensity attached to luxury access, and the environment can escalate very quickly if not expertly managed.”

Hardcastle confirms that what is particularly difficult for Swatch here is that the MoonSwatch launch already provided a live blueprint of the risks. “Once a brand has experienced scenes involving crowd surges, disappointment and policing,” she says, “the obligation shifts from reacting to proactively engineering a safer customer experience. Successful luxury houses increasingly control the experience with far greater precision.”

Neil Saunders, managing director of retail at Global Data, is even more candid. “The chaos does not reflect well on Swatch, and it probably makes Audemars Piguet wonder what on Earth it has gotten itself into,” he says. “Wanting to create some hype is understandable, but not being able to control it becomes damaging both commercially and for the brand image. Swatch should understand this better than most as it has been through this before with MoonSwatch.”

Not only Saunders and Hardcastle, but scores of commenters on Swatch’s Instagram post, point out well-known and obvious solutions that would have mitigated or entirely avoided the Royal Pop’s shambolic release.

“We have seen other premium or limited launches use staggered collection windows, verified appointment systems, geo-ticketing, VIP allocation tiers, timed QR access, private client previews and controlled queue technology to reduce volatility while preserving excitement,” says Hardcastle, adding that some combine digital ballots with curated in-store experiences so consumers feel part of an occasion rather than participants in a scramble.



Source link

Continue Reading

Tech

The Backward Logic of Chickenpox Parties

Published

on

The Backward Logic of Chickenpox Parties


Anyone who has had chickenpox shares one distinct memory: the relentless, all-consuming itch.

Ciara DiVita was only 3 years old when she caught the virus, but she remembers it well—along with the oven mitts she was made to wear to stop herself scratching. She also recalls being taken to hang out with her cousin while covered in blisters, in the hopes of deliberately infecting them.

DiVita, now 30, was actually the second in the chain, having been taken by her parents to catch chickenpox from an infectious friend. “I imagine the chain continued and my cousin gave it to someone else at a chickenpox play date,” she says.

A lot has changed over the past three decades, most notably the development of a chickenpox vaccine, meaning the virus is no longer the childhood rite of passage it once was.

Thanks to the vaccine’s success, children today are much less likely to be exposed to the infection at school or on the playground.

Chickenpox parties are also largely considered a relic of the past—a strategy many Gen X and millennial children were subjected to before vaccines became routine. But much like the virus itself—latent, opportunistic—they haven’t disappeared entirely.

Before a vaccine existed, chickenpox, which is caused by the varicella-zoster virus, felt unavoidable. In temperate countries like the UK and the US, around 90 percent of children caught the virus before adolescence (in tropical countries the average age of infection is higher).

It’s nothing to do with chickens. The splotchy, scratchy, highly contagious disease is possibly named after the French word for chickpea, pois chiche, according to one theory, because the round bumps caused by the virus resemble their size and shape. While most infant cases are mild, adolescents and adults are more likely to develop severe complications.

This is where the idea of “getting it over and done with” emerged from, according to Maureen Tierney, associate dean of clinical research and public health at Creighton University in Omaha, Nebraska.

“You were trying to have your child get the disease when they were at the greatest chance of not having complications,” Tierney says, explaining that, generally speaking, the older the patient, the more severe the infection can be.

While varicella-zoster is usually a mild, self-limiting disease in children, it can be much more severe—and sometimes life-threatening—in adults.

“I had an otherwise healthy adult patient who died of chickenpox pneumonia when I was first practicing,” Tierney says. “You never forget those scenarios.”

The virus spreads rapidly through respiratory droplets and contact with fluid from its characteristic blisters, meaning if one child contracts it, siblings and classmates are likely to be next, if unvaccinated.

Before the existence of social media, the idea that children should deliberately infect each other spread just as rapidly around communities—in conversations in the school yard, church groups, and pediatric waiting rooms—leading to the popularity of so-called chickenpox parties.

Parents swapped advice about oatmeal baths and calamine lotion and arranged to bring children together when one was thought to be infectious—despite the practice never being an official medical recommendation.

“They thought, well, if it’s going to happen to my kid anyway, it might as well happen in a controlled environment,” says Monica Abdelnour, a pediatric infectious disease specialist at Phoenix Children’s Hospital. “The families were ready to encounter this infection, deal with it, and then move on.”

While the majority of children who develop chickenpox feel well again within a week or two, around three in every 1,000 infected experience a severe complication such as pneumonia, serious bacterial skin infections, encephalitis (inflammation of the brain), or meningitis.



Source link

Continue Reading

Tech

A Danish Couple’s Maverick African Research Finds Its Moment in RFK Jr.’s Vaccine Policy

Published

on

A Danish Couple’s Maverick African Research Finds Its Moment in RFK Jr.’s Vaccine Policy


In 1996, Guinea-Bissau seemed like an ideal research post for budding pediatrician Lone Graff Stensballe. Her supervisor, a fellow Dane named Peter Aaby, had spent nearly two decades collecting data on 100,000 people living in the mud brick homes of the West African country’s capital.

Aaby and his partner, Christine Stabell Benn, believed that the years of research in the impoverished country had yielded a major discovery about vaccines—and what they described as “non-specific effects”: The measles and tuberculosis vaccines, which were derived from live, weakened viruses and bacteria, they said, boosted child survival beyond protecting against those particular pathogens.

But, the scientists said, shots made from deactivated whole germs, or pieces of them, such as the diphtheria-tetanus-pertussis (DTP) shot, caused more deaths—especially in little girls—than getting no vaccine at all.

The World Health Organization repeatedly and inconclusively examined these astonishing findings. They tended to elicit shrugs from other global health researchers, who found Aaby’s research techniques unusual and his results generally impossible to replicate.

Then came Donald Trump, Covid, and the administrative reign of anti-vaccine advocate Robert F. Kennedy Jr.

Suddenly, Aaby and Benn weren’t just sending up distant smoke signals from a far corner of the planet. They were confidently voicing their views and policy prescriptions online and in medical journals. The “framework” for “testing, approving, and regulating vaccines needs to be updated to accommodate non-specific effects,” their team wrote in a 2023 review.

And the Trump administration has taken notice.

“They became more strident in saying that their findings were real and that the world needed to do something about it,” said Kathryn Edwards, a Vanderbilt University vaccinologist who has been aware of Aaby’s work since the 1990s. “And they became more aligned with RFK.”

Kennedy, as secretary of the Department of Health and Human Services, cited one of Aaby’s papers to justify slashing $2.6 billion in US support for Gavi, a global alliance of vaccination initiatives. The cut could result in 1.2 million preventable deaths over five years in the world’s poorest countries, the nonprofit agency has estimated. Kennedy has frozen $600 million in current Gavi funding over largely debunked vaccine safety claims.

Kennedy described the 2017 paper as a “landmark study” by “five highly regarded mainstream vaccine experts” that found that girls who received a diphtheria-tetanus-pertussis, or DTP, shot were 10 times more likely to die from all causes than unvaccinated children.

In fact, the study was far too small to confidently make such assertions, as Benn acknowledged. In a study of historical data that included 535 girls, four of those vaccinated against DTP in a three-month period of infancy died of unrelated causes, while one unvaccinated girl died during that period. A follow-up published by the same group in 2022 found that the DTP shot by itself had no effect on mortality. Critics say the 2017 study, rather than being a landmark, exemplified the troubling shortfalls they perceive in the Danish team’s research.

As Aaby and Benn’s US profile has risen, scientists in Denmark have set upon the work of their compatriots. In news and journal articles published over the past 18 months, Danish statisticians and infectious disease experts have said the duo’s methods were unorthodox, even shoddy, and were structured to support preconceived views. A national scientific board is investigating their work.

Stensballe, who worked with Aaby and Benn for 20 years, has been among those voicing doubts.

“It took years to see what I see clearly today, that there is a strange concerning pattern in their work,” Stensballe said in a phone interview from Copenhagen, where she treats children at Rigshospitalet, the city’s largest teaching hospital. She said their work is full of confirmation bias—favoring interpretations that fit their hypotheses.



Source link

Continue Reading

Trending