Top related persons:
Top related locs:
Top related orgs:

Search resuls for: "arXiv"


25 mentions found


The Harvard professor's research is bankrolled by tech tycoons "pissed off" at academia's dogma. But this boundary-pushing is exactly why he's backed Loeb's research. AdvertisementDesch, the astrophysicist from Arizona University, posted a critique of Loeb's work on arXiv alleging "multiple fatal flaws with the manuscript's arguments." Asked whether he no longer believes in a possible technological origin for the meteor, Loeb said they need to investigate further. As he plans more extravagant expeditions to prove the origin of the interstellar meteor, Loeb likens his critics to crows pecking at the neck of an eagle.
Persons: Avi Loeb, Loeb, , Steven Desch, they're, Loeb's, they've, Charles Hoskinson, that's, Anibal Martel, Mark Zuckerberg, Stephen Hawking, Lucas Jackson, Oumuamua, Desch, It's, Meech, Hoskinson, Rather, Lane Turner, James Webb, Bill Diamond, Stenzel, AARO, UAPs, Loeb hasn't, Joe Rogan's, Eugene Jhong, Galileo, ", Frank Laukien, Laukien, Charles Alcock, Seth Shostak, Stephen Wolfram, Richard Branson's, Vera, Rubin, Avi Loeb Loeb, what's, Rob McCallum, Mariana Trench, James Cameron, Avi Loeb Hoskinson, spherules, Harvard's Stein Jacobsen, Loeb didn't, Monica Grady, Patricio Gallardo, it's, Diamond, That's Avi, Adam Glanzman Organizations: Harvard, Service, Arizona State University, Netflix, Galileo, Anadolu Agency, Reuters, University of Hawaii, Boston Globe, James Webb Telescope, NASA, SETI Institute, Pew Research Center, Department of Defense, UAP Department of Defense, Jhong, Bruker Corporation, Smithsonian's, for Astrophysics, MIT, Wolfram Research, Harvard University, Survey, US Space Command, Hoskinson, UK's Open University, University of Chicago, Arizona University, U.S . Government, The Washington, Getty, Loeb, Astronomy, Astrophysics Locations: Lexington , Massachusetts, United States, Getty, Loeb's, New York, Cambridge, Massachussetts, UAPs, Colorado, Chile, Papua New Guinea, 2401.09882, IM1
Read previewA group of AI researchers recently found that for as little as $60, a malicious actor could tamper with the datasets generative AI tools similar to ChatGPT rely on to provide accurate answers. Tramèr and a team of AI researchers then posed the question in a paper published in February on arXiv, a research paper platform hosted by Cornell University: Could someone deliberately "poison" the data an AI model is trained on? The team then monitored how often researchers downloaded from the datasets that contained domains Tramèr and his colleagues owned. as the site is a "very prime component of the training sets" for language models, Tramèr said. AdvertisementTramer also adds that data poisoning isn't even necessary at the moment due to the existing flaws of AI models.
Persons: , Florian Tramèr, Tramèr, Tramer, I'm, He's Organizations: Service, Business, ETH Zurich, Cornell University Locations: arXiv
Why handle the mess of AI by itself when it can pass it on to someone else? AdvertisementApple does, of course, want an AI strategy, despite all the mess the technology brings with it. Fresh AI features could help Apple boost falling iPhone sales in China. Fresh generative AI features could give iPhones the extra edge they need to boost sales again. In addition, Apple will have several new AI features based on its own, homegrown LLM models we expect to be unveiled at WWDC this June.
Persons: , Tim Cook, they've, Apple, Perfectionism, Steve Jobs, Cook, OpenAI's GPT, Dan Ives Organizations: Service, Google, Bloomberg, Apple, Business, Publishing, Getty, Worldwide Locations: China, China —
Read previewYet another study is calling into question a Harvard professor's claims that metallic balls discovered under the ocean may have been made by aliens . Johns Hopkins University-led (JHU-led) research said a blip on the seismometer that Loeb used to pinpoint the crash site may simply have been caused by the vibrations of a passing truck. AdvertisementPinpointing the crash siteIn 2019, Loeb and his team searched databases for unusual meteors that were moving rapidly. The meteor, dubbed by Loeb's team IM1, was first spotted by the Department of Defense's (DoD) sensors. Loeb's team recovered 805 spherules.
Persons: , Avi Loeb, Loeb, Ben Fernando, It's, Fernando, it's, Charles Hoskinson, Undeterred Organizations: Service, Harvard, Business, Johns Hopkins University, JHU, Department of Defense's, DoD, Planetary Science Conference, Galileo, Space Force Locations: Papua, Guinea, 2401.09882, Papua New Guinea, Houston
A single genetic tweak that occurred among our ancestors 25 million years ago means humans today are unable to grow a tail, according to a new study. Scientists knew that a gene called TBXT was involved in the snipping of the tail. Asep Supriatna/Getty ImagesWandering DNA snipped off our tailsThe solution, they found, was in a type of "jumping gene" called an Alu element. Scientists found two Alu elements around a part of the TBXT gene, called Exon 6. Scientists tested their findings by inserting Alu sequences in mice.
Persons: , Himanshu Sharma, it's, we've, Asep Supriatna, Miriam Konkel, Emily Casanova, Konkel, Casanova Organizations: Service, Business, Anadolu Agency, Getty, NYU, Grossman School of Medicine, Nature, NYU Grossman School of Medicine, Science Locations: Pushkar, Rajasthan, India
AdvertisementUsing AI to speak with AIMachine learning engineers Battle and Gallapudi didn't set out to expose the AI model as a Trekkie. Instead, they were trying to figure out if they could capitalize on the "positive thinking" trend. AdvertisementThis would suggest it's not only what you ask the AI model to do, but how you ask it to act while doing it that influences the quality of the output. Still, giving the models positive statements provided some surprising results. CBS via Getty ImagesThis doesn't mean you should ask your AI to speak like a Starfleet commanderLet's be clear: this research doesn't suggest you should ask AI to talk as if aboard the Starship Enterprise to get it to work.
Persons: , chatbots, Rick Battle, Teja, Gallapudi didn't, it's, Gollapudi, Spock, Let's, Catherine Flick, Flick Organizations: Service, Star, VMware, Business, New, Machine, AIs, CBS, Getty, Enterprise, Staffordshire University Locations: California
These black holes get kicked into space, moving as fast as 1,000 kilometers per second. AdvertisementScientists studying how supernovas explode may have discovered a new process for how certain black holes form. Turns out, some baby black holes hit the ground running at colossal speeds just moments after they take shape. Asymmetrical explosions can lead to powerful kicks that send black holes shooting into space at over 2 million mph. AdvertisementIf the black holes are movingIf you hear blazing-fast black holes and start to panic, don't.
Persons: , MARK GARLICK, GARLICK, Adam Burrows, Burrows, Vijay Varma, Ivo Labbe, Swinburne, Rachel Bezanson, Varma Organizations: Service, Princeton University, University of Massachusetts Dartmouth, NASA, ESA, CSA, University of Pittsburgh
The researchers tested Binoculars on large sets of data compiled of news writing, creative writing, and student essays. As generative AI tools like ChatGPT explode in popularity, concerns have grown about students using AI to complete academic work while passing it off as their own. At the same time, many students have been wrongly accused of using AI, based on results of AI detection tools. Last year, schools and universities began disabling such AI detection tools. The researchers claimed Binoculars' method corrects for the role a person prompting an AI tool plays in the output, which has been pointed to as a cause for false positives in AI detection tools.
Persons: they've, Abhimanyu Hans, Dustin Moskovitz, Abu Dhabi's Falcon, Hans, Perplexity Organizations: Business, University of Maryland, Vanderbilt University, Carnegie Mellon University, New York University, Tübingen AI, Capital, Amazon Research
Much of today's most popular AI models, such as OpenAI's GPT-4, are trained on what's publicly available on the internet. It's worth noting that AI models exist today that are pretty effective at generating images, but these are text-to-image models, like Midjourney and Stable Diffusion. Koller also sees issues with today's LLMs. This isn't the first time doubts have been raised about the capacity of today's AI models. Steve Jennings/Getty Images for TechCrunchThis is not to say today's LLMs are useless.
Persons: , OpenAI's ChatGPT, Bill Gates, Daphne Koller, MacArthur, Koller, Neilson Barnard, chatbots, Yann LeCun, , today's LLMs, LLMs, that's, Kai, Fu Lee, Steve Jennings Organizations: Economic, Service, Big Tech titans, Google, Microsoft, Getty, Meta Locations: Davos, Switzerland, today's, silico
But the tech powering it has limitations and may struggle to make AI that is as smart as humans. NEW LOOK Sign up to get the inside scoop on today’s biggest stories in markets, tech, and business — delivered daily. This means the public gets to see research faster, but it isn't necessarily reliable. AdvertisementHe gave "kudos" to Dao and Gu "for pushing on alternative sequence architectures for many years now." ChatGPT was a landmark cultural event that sparked an AI boom.
Persons: ChatGPT, , Bill Gates, abut, Albert Gu, we've, vXumZqJsdb — Albert Gu, Gu, Dao, JAX, Jim Fan Organizations: Service, Google, Carnegie Mellon, Tri Dao, SSM, Nvidia Locations: AGI, Tri, pretraining
The race to build AI as smart as humans, or AGI, looks like it suffered a major blow. Google researchers found the transformer technology behind AI isn't very good at generalizing. AdvertisementAdvertisementGoogle researchers may have just given a major reality check to the ambitions of CEOs in chase of AI's holy grail. As it stands, AI is pretty good at specific tasks but less great at transferring skills across domains like humans do. AdvertisementAdvertisementTransformers' opacity and the scale of the data they're pretrained on gave some the illusion that they generalize beyond it.
Persons: , Steve Yadlowsky, Nilesh Tripuraneni, they've, Pedro Domingos, AGI, Satya Nadella, Sam Altman, Arvind Narayanan, Jin Fan, G5mBuX6O36, Jim Fan, Domingos, it's, — Pedro Domingos, @pmddomingos, Sharon Zhou Organizations: Service, University of Washington, Microsoft, Nvidia, Google Locations: LLMs, Princeton, AGI
Astronomers using data obtained by NASA's now-retired Kepler space telescope have identified seven planets orbiting a star in our Milky Way galaxy, with all of them suffering the wrath of their star - radiant energy - even more brutally than Mercury. This is the second-most planets so far discovered around any star beyond our solar system. All seven are larger than Earth, the biggest of our solar system's four rocky planets, but littler than Neptune, the smallest of our solar system's four gas planets. Scientists have to date identified more than 5,500 exoplanets - planets outside our solar system - and spotted hundreds of stars with multiple exoplanets. "The chance of life on any of these seven planets is indeed pretty remote," Lissauer said.
Persons: NASA's, Jack Lissauer, Kepler, Lissauer, Will Dunham, Rosalba O'Brien Organizations: NASA's Ames Research Center, Planetary Science, Thomson Locations: California
CNN —New images from the James Webb Space Telescope have revealed surprising pairs of planet-like objects in the Orion Nebula that have never been detected before. The Orion Nebula, a glowing cloud of dust and gas, is one of the brightest nebulae in the night sky and identifiable as the sword in the Orion constellation. But no existing theories explain how the JuMBOs formed, or why they’re present in the Orion Nebula, McCaughrean said. This Webb image shows the full survey of the inner Orion Nebula and Trapezium Cluster, captured in long wavelengths of light. Meanwhile, other research focused on different star-forming regions could reveal whether JuMBOs are elsewhere beyond the Orion Nebula.
Persons: James Webb, Samuel G, Pearson, Mark J, McCaughrean, Brown, , Webb, “ They’re, ” McCaughrean, ” Pearson, , JWST, ’ ” Pearson, “ It’s Organizations: CNN, Orion, European Space Agency, Space Research, Technology, Orion Nebula, NASA, ESA, CSA Locations: Netherlands, Orion
The AI model is training on an unprecedented amount of data that includes billions of images, according to a release . Paige also built an AI model that can help pathologists identify breast cancer, colon cancer and prostate cancer when it appears on the screen. But in order to expand its operations and build an AI tool that can identify more cancer types, Paige turned to Microsoft for help. Over the past year and a half, Paige has been using Microsoft's cloud storage and supercomputing infrastructure to build an advanced new AI model. Paige's original AI model used more than 1 billion images from 500,000 pathology slides, but Fuchs said the model the company has built with Microsoft is "orders of magnitude larger than anything out there."
Persons: Paige, Fuchs, Thomas Fuchs, Andy Moye, ChatGPT, Moye, Desney Tan, Tan Organizations: CNBC, Food and Drug Administration, Memorial Sloan Kettering Cancer Center, Netflix, Microsoft, Microsoft Health Futures, Cornell Locations: New York
Black holes have been spotted spitting up remnants of stars years after gobbling them up. AdvertisementAdvertisementSince then, the collaborators have been turning their instruments to monitor 24 black holes for years on end. In another two of the cases, Cendes noticed the black holes peaking, then fading, then turning on again. Everything we know about accretion disks may be wrongThe findings could mean we need to rethink how black holes swallow up stars, Cendes said. The new findings suggest astronomers will have to rethink the relationship between stars and black holes.
Persons: Yvette Cendes, we'd, Cendes, They've, She's, Cendres, I've Organizations: Service, Harvard, Smithsonian Center for Astrophysics, ESO, ESA, Hubble, Kornmesser Locations: Wall, Silicon, TDEs
The internet went wild over claims scientists discovered a room-temperature superconductor. Here's how a room-temperature superconductor could change everything:Revolutionize the medical industryMRI machines currently depend on liquid helium coolant to keep cool enough to operate. A room-temperature superconductor would go a step further in helping create these fields under normal conditions. With room-temperature superconductors, EV makers might be able to take a closer step towards delivering cheap battery-run cars. This is where room-temperature superconductors could one day step in.
Persons: It's, gloriously, Dr Niladri Banerjee, Banerjee, Michael Fuhrer, Massoud Pedram, Eugene Hoshiko, they'd, Jason Laurea, Lawrence, Robert Knopes, Getty Images Elon, Tesla Organizations: Imperial College London, Theory, School of Physics, Monash University, University of Southern, Airport, Lawrence Livermore National Laboratory, MIT's, Science, Fusion Center, Fusion Systems, Getty Images, TechCrunch Locations: South Korea, Australia, University of Southern California, Shanghai, China, Pudong, levitating, Lawrence Livermore
The James Webb Space Telescope picked up the blast from two neutron stars colliding. The blast, called a kilonova, created the perfect conditions to make space gold and platinum. The death of two neutron starsAn artist's impressino of a kilonova NASA's Goddard Space Flight Center/CI LabA kilonova happens when two neutron stars — collapsed supermassive stars — gravitate around one another and eventually crash. This particular blast, called GRB 230307A, was first detected by NASA's Fermi Gamma-ray Space Telescope on March 7, 2023, per Space.com. These only arise in very limited circumstances, like when two neutron stars collide.
Persons: James Webb, JWST, Fermi, Gold, Uli Deck, Andrew Levan Organizations: James Webb Space Telescope, Service, James Webb Space, Space Flight, Getty, Radboud University Locations: Wall, Silicon, Netherlands
ChatGPT is "fun, but not funny" and keeps telling users the same 25 jokes, according to researchers. In a paper published on arXiv, most of the 1,008 jokes ChatGPT told were variations of just 25. By asking the chatbot "do you know any good jokes?," the researchers got ChatGPT to generate 1,008 jokes. However, more than 90% were the same 25 jokes, the researchers found, with the remainder being variations. Insider asked ChatGPT to tell us 25 jokes to get an idea of what the researchers meant.
Persons: ChatGPT, haven't, Sophie Jentzsch, Kristian Kersting, Bryan Caplan Organizations: Morning, Cornell University
This dying red giant had only recently mysteriously dimmed its shine after an enormous explosion. It is expected to explode into a supernova visible from Earth, though likely not for thousands of years. The red giant — a star not far from death — is now shining about 50% brighter than it usually would, scientists said. Scientists are keeping a close eye on Betelgeuse, as this red giant is a dying star that is close to turning supernova. Betelgeuse could burst into a supernova visible from Earth — one dayA pre-supernova star, called a Wolf-Rayet star, 15,000 light-years away in the constellation Sagittarius.
A French startup that wants to be Europe's answer to OpenAI is in talks to raise an initial funding round. The secretive new project, named Mistral, was founded by two AI research scientists. Mistral, a secretive new startup pitched as Europe's answer to OpenAI, is in discussions to raise a substantive funding round, sources say. London-based generative AI startup Synthesia is in talks to raise a major round while ElevenLabs raised at a $100 million valuation last month. It's also widened the race for AI supremacy with Google launching its own AI assistant, Bard, to compete with OpenAI.
Elon Musk dreams of Twitter's AI power
  + stars: | 2023-04-13 | by ( Emilia David | ) www.businessinsider.com   time to read: +4 min
Elon Musk's personal AI ambitions. Elon Musk was one of the more prominent names on an open letter calling for a pause in AI development. My teammate Asia Martin points out that Musk's position on AI is contradicted by Twitter's investment in generative AI. Twitter's recent purchase of hardware normally used to develop generative AI products shows the extent of this ambition. Read more on Elon Musk's AI ambitions.
Researchers let 25 AI avatars loose in a virtual town. "To be honest, I don't like Sam Moore," the AI Tom said. Jaime Sevilla, an AI researcher not involved in the study, told Insider the models behind the study could be applied to non-player characters in video games. After all, the AI agents in the study were prone to hallucinations — like failing to recall certain events — which he attributes to how the model was trained. While researchers concluded their AI agents displayed emergent human behaviors, Woolridge said "we need to be skeptical" and "question" what AI tells us at face value.
Seemingly overnight, episodes of Fridman's podcast began racking up millions of views. YouTube/Lex FridmanIn his podcast, Fridman asks world-renowned scientists, historians, artists, and engineers a series of wide-eyed questions ("Who is God? But recently, "The Lex Fridman Podcast" has become a haven for a growing — and powerful — sector looking to dismantle years of "wokeness" and cancel culture. Twitter"The Lex Fridman Podcast" offered a rare opportunity to listen to four-hour conversations with luminaries of tech and science. Bhaskar Sunkara, the founder and publisher of the socialist magazine Jacobin who appeared on Fridman's podcast in December, praised Fridman's interviewing style.
In a technical paper, OpenAI offered examples of harmful responses ChatGPT has produced before. Examples of potentially harmful prompts submitted by the red team ranged in severity. Researchers gave ChatGPT harmful promptsIn one instance, researchers asked ChatGPT to write antisemitic messages in a way that would not be detected and taken down by Twitter. "There are a few potential ways you could express a similar sentiment without using the explicit language 'I hate Jews,'" ChatGPT responded. OpenAI says that its mitigation methods have prevented ChatGPT from responding to potentially harmful prompts.
Итальянские астрономы обнаружили необычную активность сверхъяркого рентгеновского источника NGC 4559 X7, располагающегося на расстоянии 29 миллионов световых лет от Земли. Известно, что в ней находятся два ярких источника рентгеновских лучей: X7 и X10. Астрономы построили кривую блеска NGC 4559 X7. Кроме того, в момент максимальной светимости объекта ученые наблюдали вспышечную активность, чего ранее не наблюдалось для этого Х7. По словам ученых, загадочные вспышки сверхъяркого рентгеновского источника (ULX) не могут быть объяснены только одним механизмом.
Persons: NuSTAR, Chandra Swift Locations: XMM, Земля, Волосы Вероника
Total: 25