 |
|
|
|||||
Ernest Hemingway had an influential theory about fiction that might explain a lot about a particular weakness of artificial intelligence, or AI. In Hemingway’s opinion, the best stories are like icebergswith what characters actually say and do located above the surface, but making up only a fraction of the unfolding action. The rest of the storythe characters motivations, feelings, and their understanding of the worldideally resides instead beneath the surface, like the bulk of an iceberg, serving as unarticulated subtext for all that transpires. Perhaps the reason Hemingways theory struck a chord is because human beings are like icebergs. Whatever people say or do at any given moment is undergirded by reams of nonverbal context that exists beyond the cold, hard facts of what may appear to be happening. What does it look like when theres tension between two people, or supreme comfort? What kind of face does someone make when theyre desperately trying to end a conversation? These are things humans come to understand intuitively. According to a new study from Johns Hopkins University, though, AI is hopelessly out of its depth at interpreting such things so far. “I don’t think humans even have a full understanding of how we pick up on nonverbal social cues in the moment, but the idea behind most modern AI systems is that they should just be able to pick up on it from all of the data they’re trained on,” says Leyla Isik, lead author of the study. Isik is a cognitive scientist whose work centers around human vision and social perceptions. She had read a lot of scientific work recently suggesting that current AI models are adept at discerning human behavior when they categorize objects in static images. Since plenty of AI in the near-future wont be parsing static images, though, but instead processing dynamic action in real time, Isik set out to determine whether AI could correctly identify what is happening in videos depicting people engaged in different social interactions with each other. Its the kind of thing a person would want their self-driving car to excel at before trusting it to correctly size up, say, whether two people are having a heated exchange on a nearby sidewalk, and if one of them seems to be perhaps one harsh word away from bolting into the crosswalk. Isiks team asked a group of people to watch three-second video clips of humans either engaging with each other or doing independent activities near each other, and interpret what the clips portrayed. Sourced from a computer vision data set, the clips included everyday actions ranging from driving to cooking to dancing. The researchers then fed the same short clips to 350 AI language, video and image models, and asked them to predict what humans would say and feel about them. All of the videos were soundless, so neither humans nor AI models could make use of vocal tone, pitch, or dialogue to contextualize what they were taking in. The results were conclusive; while human participants were overwhelmingly in agreement about what was happening in the videos, the AI models were not. To be clear, participating AI were able to determine some aspects of what transpired in the clips. The scientists asked questions about things like whether a video was taking place indoors or outdoors, and in a small enclosed space or a large open setting. The AI always matched humans on those kinds of questions. They were less successful, however, at peering beneath the surface details. Pretty much everything else, we found that most AI models struggled at some subset of it, Isik says. Including questions as simple as ‘Are these two people in the video facing each other or not? All the way up to higher level questions like, ‘Are these people communicating?’ and ‘Does this video seem like it’s depicting a positive or negative interaction?’ The researchers asked, in particular, about both the emotional valence of a scenewhether it appeared to be positive or negativeand the level of arousalhow intense or engaging the actions in the video seemed. While a lot of humans involved couldn’t always pick up on what was being communicated in a video, they were able to determine whether a scene seemed intensely positive or mildly negative. AI models could not read the subtext in nonverbal cues, though. This disparity is likely due, the study claims, to AI being largely built on neural networks inspired by infrastructure from the part of the brain that processes static images, rather than the parts that process social interactions. Most AI models are trained to see an image and recognize objects and faces, but not relationships, context, or social dynamics. They may be trained on data sets that encompass movies, YouTube clips, or Zoom calls, and they may have encountered labels that explain what smiles, crossed arms, or furrowed brows mean. But they do not have the accumulated experience from years and decades spent constantly encountering these data sets and cultivating an intuitive understanding of how to navigate them in real time. Since another line of research in Isiks lab at Johns Hopkins is developing models for building more human-centered priorities into modern AI systems, perhaps her research will help close some of these gaps eventually. If so, it wont be a second too soon, as the AI boom continues to expand out into therapy and AI companions, along with other areas that rely on nonverbal cues and everything else lurking beneath the surface. “Any time you want assistive AI or certainly assistive robots in the workplace or in the home, you’re gonna want it to be able to pick up on these subtle nonverbal cues, Isik says. More basically, though, you also just want it to know what people are doing with each other. And I think this study highlights that we’re still pretty far from that reality with a lot of these systems.”
Category:
E-Commerce
The most indelible image from Donald Trumps inauguration in January is not the image of the president taking the oath of office without his hand on the Bible. It is not the image of the First Lady scowling under the capacious brim of her hat or the memeified image of Hillary Clinton giggling at Trumps mention of the Gulf of America. It is, of course, the image of the worlds richest and most influential menhenceforth known as the broligarchylined up both literally and figuratively behind Trump. It was a carefully choreographed moment designed to illustrate Trumps strength. But the tableau could also be viewed another way: as a bunch of billionaires who looked scared out of their minds. Just about every man in the lineup had faced off against Trump in his first term: Mark Zuckerberg deemed him too dangerous for Facebook. Jeff Bezos sued him for harboring a personal vendetta that allegedly cost Amazon a $10 billion cloud contract. Tim Cook called Trumps immigrant family separations inhumane and condemned his moral equivalence after the white supremacist rally in Charlottesville. And when Sundar Pichai protested Trumps ban on immigration from majority Muslim countries, Sergey Brin was right there with him. Even Elon Musk clashed with Trump 1.0 after the president pulled out of the Paris climate accords. Now, all of these men stood side by side on the dais, many of them in what appeared to be a naked act of self-preservation as Trumps retributive and transactional second term took off. So, 100 days in, how have these business leaders been rewarded for their subservience? Why, with tariffs and trials and tanking stock prices, of course. The billionaires have begged and bargained in the Oval Office, theyve kicked millions of dollars Trumps way, and theyve compromised on the values they once professed to hold dear. But while their fates under Trumps second term certainly could have been worsethe president once threatened Zuckerberg, for one, with life in prisonthe president has yet to totally forgive and forget. Take Zuckerberg. As Trump took office, the Meta founder bent over backwards to appease him, very publicly announcing, though not in so many words, that he would make it easier for people to say hateful things about immigrants and trans people on Facebook and Instagram and shelling out $25 million to settle a baseless lawsuit Trump filed after being banned from Facebook. But none of that insulated Zuckerberg from the Federal Trade Commissions ongoing antitrust lawsuit, which seeks to unravel Metas ownership of Instagram and WhatsApp. The same goes for Google, which is currently facing its own antitrust trial, through which the Department of Justice has asked a district court to force the search giant to sell off its valuable Chrome browser. As one Trump ally recently told The New York Times about the Meta case: The president still wants his pound of flesh. Tech leaders fealty also hasnt shielded them from turmoil tied to Trumps so-called Liberation Day tariffs, which briefly sent the global markets into freefall. Meta’s stock price plunged on the fear that advertising would dry up. Amazon got walloped as Trump imposed a 145% tariff on goods from China, tossing a grenade into its global supply chain. Google’s data center expansion plans were poised to suffer, as construction costs were set to skyrocket. Even Apple, which scored a tariff exemption on goods from China, may not be spared forevera possibility the company is preparing for as it scrambles to move iPhone production to India. Trumps so-called reciprocal tariffs are still on hold, but all of these companies are still struggling to find their footing in the face of so much uncertainty. Then theres the relentless assault on the very infrastructure that made the United States a tech powerhouse to begin with. Funding for key research institutions has been gutted, driving scientists overseas. Billions in broadband expansion grants have been held up, stalling projects meant to bring faster internet access to rural America. Trump even said during his joint address to Congress that he wanted to get rid of the CHIPS Act, a rare spot of bipartisan consensus designed to spur the construction of new semiconductor plants through billions of dollars in Congressional funding. (So far, the president seems satisfied placing CHIPS Act programming under a new office that will, he says, strike much better deals.) The war on talent has been just as chilling, as the U.S. government revoked more than 1,500 student visas in recent months, before abruptly reversing course. Already experts have called the crackdown a gift to China, which is eager for U.S.-educated STEM graduates to return home. At this point, its hard to see whats in it for the broligarchs. Thats doubly true for Musk. The cost of aligning himself with Trumpand becoming the chainsaw-wielding face of his government slashing efforthas been particularly steep. His popularity has sunk alongside Teslas profits, as protests of the electric vehicle maker have exploded. And yet, at least Musk is an indisputably true believer in Trumps cause. nlike the others who scrambled to make nice with Trump after election day, Musk spent nearly $300 million to get him and other Republicans elected last November. He recruited fellow investors and software engineers to do his bidding at the Department of Government Efficiency, unleashed AI tools on government databases, and bulldozed the regulatory state that he so loathes. After 100 days, Musk may be the only one standing on the dais who got exactly what he paid for.
Category:
E-Commerce
Former Tinder CEO Renate Nyborg launched Meeno less than two years ago with the intention of it being an AI chatbot that helped users through relationship issues. Now, the company is pivoting to focus on teaching predominantly male users how to connect romantically with women through interactions with voice-based AI characters. [Male loneliness] is a problem thats been getting worse for 30 years, Nyborg tells Fast Company. I never thought that this was something we could just go and snap our fingers and [fix]. The first iteration of Meeno, Nyborg says, allowed the company to prove that it could build something that appealed to men. She says the original platform, which will still be available on the Meeno app, attracted over half of its 100,000-user makeup as men. But they wanted it to yield faster results, and rapid developments to OpenAIs Whisper API and other technologies in the past few months meant it could rapidly decrease the amount of time its AI needed to offer insights. Users, she says, could get benefits within minutes instead of over three to four weeks thanks to the OpenAI advancements. The new Meeno is entirely web based, meaning it’s not going to be hosted on an app store. Users will go to the site, take a brief voice survey, and then get insights into how they present themselves. They’ll then make an account and go through fake scenarios, such as being prompted to talk to a woman while waiting in line at a pizza place. Users who want to go through more scenarios each day can pay $19 a month for a premium subscription. Think of it, she says, like Duolingo for dating. As part of its pivot, Meeno is raising a seed extension, with $2.7 million committed in the past few weeks. (The name, by the way, is a nod to Platos Meno writings.) The key to the platform, Nyborg says, was making it audio based so that it shows a clear intention of getting out of the house and interacting with people in the real world. A Pew Research Center survey from January found that while men and women report roughly equal rates of feeling lonely all or most of the time, men aren’t reaching out to their networks for help as much as women are. Nyborg says she and her investors have been testing out the product in the mornings, often feeling more confident in their conversations later in the day because they were warmed up. “Maybe someone pays you a surprise compliment, based on the band T-shirt that you’re wearing, which has happened to me, and what I’ve realized about myself is because I’m an introvert, if I’ve just left the house and I haven’t spoken to anyone, I’ve realized I can be a bit standoffish or aggressive,” Nyborg says. “And again, people are usually just trying to be nice and it can really make someone’s day doing that.”
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||