Connect with us

Science

The BigScience group has launched the BLOOM model, or the BigScience Large Open-science Open-access Multilingual Language Model for machine learning.

Avatar

Published

on

AI
By Associated Press | Updated: 18 July 2022

The tech industry’s latest artificial intelligence constructs can be pretty convincing if you ask them what it feels like to be a sentient computer, or maybe just a dinosaur or squirrel. But they’re not so good — and sometimes dangerously bad — at handling other seemingly straightforward tasks.

Take, for instance, GPT-3, a Microsoft-controlled system that can generate paragraphs of human-like text based on what it’s learned from a vast database of digital books and online writings. It’s considered one of the most advanced of a new generation of AI algorithms that can converse, generate readable text on demand and even produce novel images and video.

Among other things, GPT-3 can write up most any text you ask for — a cover letter for a zookeeping job, say, or a Shakespearean-style sonnet set on Mars. But when Pomona College professor Gary Smith asked it a simple but nonsensical question about walking upstairs, GPT-3 muffed it.

“Yes, it is safe to walk upstairs on your hands if you wash them first,” the AI replied.

These powerful and power-chugging AI systems, technically known as “large language models” because they’ve been trained on a huge body of text and other media, are already getting baked into customer service chatbots, Google searches and “auto-complete” email features that finish your sentences for you. But most of the tech companies that built them have been secretive about their inner workings, making it hard for outsiders to understand the flaws that can make them a source of misinformation, racism and other harms.

“They’re very good at writing text with the proficiency of human beings,” said Teven Le Scao, a research engineer at the AI startup Hugging Face. “Something they’re not very good at is being factual. It looks very coherent. It’s almost true. But it’s often wrong.”

That’s one reason a coalition of AI researchers co-led by Le Scao — with help from the French government — launched a new large language model Tuesday that’s supposed to serve as an antidote to closed systems such as GPT-3. The group is called BigScience and their model is BLOOM, for the BigScience Large Open-science Open-access Multilingual Language Model. Its main breakthrough is that it works across 46 languages, including Arabic, Spanish and French — unlike most systems that are focused on English or Chinese.

It’s not just Le Scao’s group aiming to open up the black box of AI language models. Big Tech company Meta, the parent of Facebook and Instagram, is also calling for a more open approach as it tries to catch up to the systems built by Google and OpenAI, the company that runs GPT-3.

“We’ve seen announcement after announcement after announcement of people doing this kind of work, but with very little transparency, very little ability for people to really look under the hood and peek into how these models work,” said Joelle Pineau, managing director of Meta AI.

Competitive pressure to build the most eloquent or informative system — and profit from its applications — is one of the reasons that most tech companies keep a tight lid on them and don’t collaborate on community norms, said Percy Liang, an associate computer science professor at Stanford who directs its Center for Research on Foundation Models.

“For some companies this is their secret sauce,” Liang said. But they are often also worried that losing control could lead to irresponsible uses. As AI systems are increasingly able to write health advice websites, high school term papers or political screeds, misinformation can proliferate and it will get harder to know what’s coming from a human or a computer.

Meta recently launched a new language model called OPT-175B that uses publicly available data — from heated commentary on Reddit forums to the archive of US patent records and a trove of emails from the Enron corporate scandal. Meta says its openness about the data, code and research logbooks makes it easier for outside researchers to help identify and mitigate the bias and toxicity that it picks up by ingesting how real people write and communicate.

“It is hard to do this. We are opening ourselves for huge criticism. We know the model will say things we won’t be proud of,” Pineau said.

While most companies have set their own internal AI safeguards, Liang said what’s needed are broader community standards to guide research and decisions such as when to release a new model into the wild.

It doesn’t help that these models require so much computing power that only giant corporations and governments can afford them. BigScience, for instance, was able to train its models because it was offered access to France’s powerful Jean Zay supercomputer near Paris.

The trend for ever-bigger, ever-smarter AI language models that could be “pre-trained” on a wide body of writings took a big leap in 2018 when Google introduced a system known as BERT that uses a so-called “transformer” technique that compares words across a sentence to predict meaning and context. But what really impressed the AI world was GPT-3, released by San Francisco-based startup OpenAI in 2020 and soon after exclusively licensed by Microsoft.

GPT-3 led to a boom in creative experimentation as AI researchers with paid access used it as a sandbox to gauge its performance — though without important information about the data it was trained on.

OpenAI has broadly described its training sources in a research paper, and has also publicly reported its efforts to grapple with potential abuses of the technology. But BigScience co-leader Thomas Wolf said it doesn’t provide details about how it filters that data, or give access to the processed version to outside researchers.

“So we can’t actually examine the data that went into the GPT-3 training,” said Wolf, who is also a chief science officer at Hugging Face. “The core of this recent wave of AI tech is much more in the dataset than the models. The most important ingredient is data and OpenAI is very, very secretive about the data they use.”

Wolf said that opening up the datasets used for language models helps humans better understand their biases. A multilingual model trained in Arabic is far less likely to spit out offensive remarks or misunderstandings about Islam than one that’s only trained on English-language text in the US, he said.

One of the newest AI experimental models on the scene is Google’s LaMDA, which also incorporates speech and is so impressive at responding to conversational questions that one Google engineer argued it was approaching consciousness — a claim that got him suspended from his job last month.

Colorado-based researcher Janelle Shane, author of the AI Weirdness blog, has spent the past few years creatively testing these models, especially GPT-3 — often to humorous effect. But to point out the absurdity of thinking these systems are self-aware, she recently instructed it to be an advanced AI but one which is secretly a Tyrannosaurus rex or a squirrel.

“It is very exciting being a squirrel. I get to run and jump and play all day. I also get to eat a lot of food, which is great,” GPT-3 said, after Shane asked it for a transcript of an interview and posed some questions.

Shane has learned more about its strengths, such as its ease at summarising what’s been said around the internet about a topic, and its weaknesses, including its lack of reasoning skills, the difficulty of sticking with an idea across multiple sentences and a propensity for being offensive.

“I wouldn’t want a text model dispensing medical advice or acting as a companion,” she said. “It’s good at that surface appearance of meaning if you are not reading closely. It’s like listening to a lecture as you’re falling asleep.”

Science

Elon Musk’s Neuralink Worth $5 Billion Based on Private Stock Trades

Avatar

Published

on

In Neuralink's last known fundraising in 2021, it raised $205 million (roughly Rs. 1,700 crore) at an approximately $2 billion (roughly Rs. 16,500 crore) valuation.
By Reuters | Updated: 5 June 2023

Elon Musk’s brain implant startup Neuralink, which was valued at close to $2 billion (roughly Rs. 16,500 crore) in a private fundraising round two years ago, is now worth around $5 billion (roughly Rs. 41,300 crore) based on privately executed stock trades described to Reuters by five sources with knowledge of the matter.

Some purchases by bullish investors boosted the valuation in recent months, ahead of Neuralink’s May 25 announcement that U.S. regulators had approved a human trial on its brain chip, the sources said.

Experts have said it could take several years for Neuralink to secure commercial use clearance. Kip Ludwig, former program director for neural engineering at the U.S. National Institutes of Health (NIH), said he “optimistically” expected Neuralink to take at least 10 more years to commercialize its brain implant. The company also faces other challenges that include federal probes into its handling of animal research.

Following the trial’s approval, however, Neuralink shares were marketed privately to investors in recent days at a $7 billion (roughly Rs. 57,900 crore) valuation, equivalent to $55 (roughly Rs. 4,500) per share, according to an email seen by Reuters. Reuters could not establish whether the seller found buyers for that price. The email cited the U.S. Food and Drug Administration’s (FDA) approval of the clinical trial as grounds for the deal being “sweeter.”

Neuralink executives and Musk did not respond to requests for comment.

Musk has expressed grand ambitions for Neuralink, saying its chip would allow healthy and disabled people alike to pop into neighbourhood facilities for speedy surgical insertions of devices to treat obesity, autism, depression and schizophrenia. He even sees them being used for web-surfing and telepathy. A Neuralink executive recently gave more modest short-term objectives, such as helping paralyzed patients communicate through computerized text without typing.

The stock transactions at a valuation of around $5 billion (roughly Rs. 41,300 crore) have been carried out by shareholders such as employees and the company’s early backers, rather than Neuralink selling new shares to investors. Such so-called secondary trades are an imperfect gauge of a company’s value; their volume is thin and they lack the wider market consensus of a fundraising round or initial public offering (IPO).

Neuralink’s valuation jump in secondary trades is in sharp contrast to other startups. About 85percent of pre-IPO companies are currently valued in secondary trades at an average discount of 47 percent to their last funding round, according to data provider Caplight.

In Neuralink’s last known fundraising in 2021, it raised $205 million (roughly Rs. 1,700 crore) at an approximately $2 billion (roughly Rs. 16,500 crore) valuation, according to data provider Pitchbook.

Many of the recent stock sales have been to relatively small investors, who typically focus more on getting a slice of a company owned by Musk than scrutinizing its valuation. The maximum amount sought for the Neuralink shares marketed for sale at a $7 billion (roughly Rs. 57,900 crore) valuation was just $500,000, according to the email seen by Reuters.

Sim Desai, chief executive of Hiive, an online platform where the shares are traded, said demand for Neuralink stock has been “tremendous.” He pegged the valuation that buyers are willing to pay at around $4.5 billion (roughly Rs. 37,200 crore).

Some biomedical experts are skeptical. Arun Sridhar, a scientist and entrepreneur who specializes in neuromodulation, called Neuralink’s valuation “bonkers” based on how early the brain implant is in its clinical development.

“A study to assess safety and tolerability is in no shape or form valid to justify a $5 billion (roughly Rs. 41,300 crore) valuation,” said Sridhar, who helped launch Galvani Bioelectronics, a developer of implants backed by GSK Plc and Alphabet Inc’s Verily Life Sciences. Galvani is not a competitor of Neuralink because its implants under development will be installed in an artery to the spleen to help treat rheumatoid arthritis, rather than the brain.

Investigations
The FDA initially rejected Neuralink’s request for a human trial last year, citing safety reasons, Reuters has reported. Even after securing approval, the company faces several challenges.

Neuralink has come under scrutiny from U.S. lawmakers after Reuters reported in May that its animal-research board may have violated conflict-of-interest regulations. Neuralink employees who sat on that board, which oversees the welfare of the animals that were being tested, also stood to benefit from the implant’s quick development. Neuralink stock that some of the employees hold has jumped around 150 percent in value in just two years, based on the secondary trades.

The law enforcement arm of the U.S. Department of Agriculture has been investigating Neuralink for potential animal-welfare violations. Neuralink staff told Reuters last year that the company was rushing and botching surgeries on monkeys, pigs and sheep, resulting in far more animal deaths than necessary, as Musk pressured staff to receive FDA approval.

The Department of Transportation is separately probing whether Neuralink illegally transported dangerous pathogens on chips removed from monkey brains without proper containment measures.

Neither Musk nor Neuralink have responded to multiple requests for comment on the probes or the Reuters reports.

© Thomson Reuters 2023

Continue Reading

Science

NASA Panel Studying Unidentified Aerial Phenomena, UFO Sightings to Hold First Public Meeting Ahead of Report

Avatar

Published

on

A SpaceX Falcon 9, with NASA astronauts Doug Hurley and Bob Behnken in the Dragon crew capsule, lifts off from Pad 39A at the Kennedy Space Center in Cape Canaveral, Fla., on Saturday. David J. Phillip/AP
The panel was formed last June to examine unclassified UFO sightings and other data collected from various sectors.
By Reuters | Updated: 31 May 2023

A NASA panel formed last year to study what the government calls “unidentified aerial phenomena,” commonly termed UFOs, was due to hold its first public meeting on Wednesday, ahead of a report expected in coming weeks.

The 16-member body, assembling experts from fields ranging from physics to astrobiology, was formed last June to examine unclassified UFO sightings and other data collected from civilian government and commercial sectors.

The focus of Wednesday’s four-hour public session “is to hold final deliberations before the agency’s independent study team publishes a report this summer,” NASA said in announcing the meeting.

The panel represents the first such inquiry ever conducted under the auspices of the US space agency for a subject the government once consigned to the exclusive and secretive purview of military and national security officials.

The NASA study is separate from a newly formalised Pentagon-based investigation of unidentified aerial phenomena, or UAPs, documented in recent years by military aviators and analysed by US defense and intelligence officials.

The parallel NASA and Pentagon efforts — both undertaken with some semblance of public scrutiny — highlight a turning point for the government after decades spent deflecting, debunking and discrediting sightings of unidentified flying objects, or UFOs, dating back to the 1940s.

The term UFOs, long associated with notions of flying saucers and aliens, has been replaced in government parlance by “UAP.”

While NASA’s science mission was seen by some as promising a more open-minded approach to a topic long treated as taboo by the defense establishment, the US space agency made it known from the start that it was hardly leaping to any conclusions.

“There is no evidence UAPs are extraterrestrial in origin,” NASA said in announcing the panel’s formation last June.

In its more recent statements, the agency presented a new potential wrinkle to the UAP acronym itself, referring to it as an abbreviation for “unidentified anomalous phenomena.” This suggested that sightings other than those that appeared airborne may be included.

Still, NASA in announcing Wednesday’s meeting, said the space agency defines UAPs “as observations of events in the sky that cannot be identified as aircraft or known natural phenomena from a scientific perspective.”

US defense officials have said the Pentagon’s recent push to investigate such sightings has led to hundreds of new reports that are under examination, though most remain categorized as unexplained.

The head of the Pentagon’s newly formed All-domain Anomaly Resolution Office (AARO) has said the existence of intelligent alien life has not been ruled out but that no sighting had produced evidence of extraterrestrial origins.

© Thomson Reuters 2023

Continue Reading

Science

A Third of Milky Way’s Planets Orbiting Most Common Stars Could Hold Water, Harbour Life: Study

Avatar

Published

on

The most common stars in our galaxy are considerably smaller and cooler, sporting just half the mass of the Sun at most.
By Press Trust of India | Updated: 30 May 2023

One-third of the planets orbiting the most common stars across the Milky Way galaxy may hold onto liquid water and possibly harbour life, according to a study based on latest telescope data.

The most common stars in our galaxy are considerably smaller and cooler, sporting just half the mass of the Sun at most. Billions of planets orbit these common dwarf stars.

The analysis, published in the journal Proceedings of the National Academy of Sciences, shows that two-thirds of the planets around these ubiquitous small stars could be roasted by tidal extremes, sterilising them.

However, that leaves one-third of the planets—hundreds of millions across the galaxy—that could be in a goldilocks orbit close enough, and gentle enough, to be possibly habitable.

“I think this result is really important for the next decade of exoplanet research, because eyes are shifting towards this population of stars,” said Sheila Sagear, a doctoral student at the University of Florida (UF) in the US.

“These stars are excellent targets to look for small planets in an orbit where it’s conceivable that water might be liquid and therefore the planet might be habitable,” Sagear said in a statement.

Sagear and UF astronomy professor Sarah Ballard measured the eccentricity of a sample of more than 150 planets around M dwarf stars, which are about the size of Jupiter.

The more oval shaped an orbit, the more eccentric it is. If a planet orbits close enough to its star, at about the distance that Mercury orbits the Sun, an eccentric orbit can subject it to a process known as tidal heating.

As the planet is stretched and deformed by changing gravitational forces on its irregular orbit, friction heats it up. At the extreme end, this could bake the planet, removing all chance for liquid water.

“It’s only for these small stars that the zone of habitability is close enough for these tidal forces to be relevant,” Ballard said.

The researchers used data from NASA’s Kepler telescope, which captures information about exoplanets as they move in front of their host stars.

To measure the planets’ orbits, they focused especially on how long the planets took to move across the face of the stars. Their study also relied on new data from the Gaia telescope, which has measured the distance to billions of stars in the galaxy.

“The distance is really the key piece of information we were missing before that allows us to do this analysis now,” Sagear said.

The team found that stars with multiple planets were the most likely to have the kind of circular orbits that allow them to retain liquid water.

Stars with only one planet were the most likely to see tidal extremes that would sterilise the surface, according to the researchers.

Since one-third of the planets in this small sample had gentle enough orbits to potentially host liquid water, that likely means that the Milky Way has hundreds of millions of promising targets to probe for signs of life outside our solar system, they added.

Continue Reading

Science

New Low-Cost Smartphone Attachment, Custom App to Monitor Blood Pressure at User’s Fingertip

Avatar

Published

on

The researchers estimate that the cost of the clip could be as low as 10 cents (Rs. 0.7) apiece when manufactured at scale.

By Press Trust of India | Updated: 30 May 2023

Scientists have developed a simple, low-cost clip that uses a smartphone’s camera and flash to monitor blood pressure at the user’s fingertip. The clip developed by researchers at the University of California (UC) San Diego, US, works with a custom smartphone app and currently costs about 80 cents (Rs. 5.6) to make.

The researchers estimate that the cost could be as low as 10 cents (Rs. 0.7) apiece when manufactured at scale.

The technology, described in the journal Scientific Reports, could help make regular blood pressure monitoring easy, affordable and accessible to people in resource-poor communities, they said.

It could benefit older adults and pregnant women, for example, in managing conditions such as hypertension, according to the researchers.

“We have created an inexpensive solution to lower the barrier to blood pressure monitoring,” said study first author Yinan Xuan, a Ph.D. student at UC San Diego.

“Because of their low cost, these clips could be handed out to anyone who needs them but cannot go to a clinic regularly,” said study senior author Edward Wang, a professor at UC San Diego and director of the Digital Health Lab.

“We have created an inexpensive solution to lower the barrier to blood pressure monitoring,” said study first author Yinan Xuan, a Ph.D. student at UC San Diego.

“Because of their low cost, these clips could be handed out to anyone who needs them but cannot go to a clinic regularly,” said study senior author Edward Wang, a professor at UC San Diego and director of the Digital Health Lab.

“Our is a calibration-free system, meaning you can just use our device without touching another blood pressure monitor to get a trustworthy blood pressure reading,” Wang said.

To measure blood pressure, the user simply presses on the clip with a fingertip. A custom smartphone app guides the user on how hard and long to press during the measurement.

The clip is a 3D-printed plastic attachment that fits over a smartphone’s camera and flash. It features an optical design similar to that of a pinhole camera. When the user presses on the clip, the smartphone’s flash lights up the fingertip.

That light is then projected through a pinhole-sized channel to the camera as an image of a red circle. A spring inside the clip allows the user to press with different levels of force.

The harder the user presses, the bigger the red circle appears on the camera.

The smartphone app extracts two main pieces of information from the red circle. By looking at the size of the circle, the app can measure the amount of pressure that the user’s fingertip applies.

By looking at the brightness of the circle, the app can measure the volume of blood going in and out of the fingertip.

An algorithm converts this information into systolic and diastolic blood pressure readings.

The researchers tested the clip on 24 volunteers from the UC San Diego Medical Center. Results were comparable to those taken by a blood pressure cuff.

“Using a standard blood pressure cuff can be awkward to put on correctly, and this solution has the potential to make it easier for older adults to self-monitor blood pressure,” said study co-author Alison Moore, from UC San Diego School of Medicine.

Continue Reading

Science

China to Send First Civilian Into Space on Tuesday as Part of Crewed Mission to Its Space Station

Avatar

Published

on

Until now, all Chinese astronauts sent into space have been part of the People's Liberation Army.
By Agence France-Presse | Updated: 29 May 2023

China will send its first civilian astronaut into space as part of a crewed mission to the Tiangong space station on Tuesday, its Manned Space Agency announced, as Beijing pushes ahead with its extra-terrestrial ambitions.

The world’s second-largest economy has invested billions of dollars into its military-run space programme, trying to catch up with the United States and Russia after years of belatedly matching their milestones.

Until now, all Chinese astronauts sent into space have been part of the People’s Liberation Army.

“Payload expert Gui Haichao is a professor at Beijing University of Aeronautics and Astronautics,” China Manned Space Agency Spokesperson Lin Xiqiang told reporters Monday.

Gui will be “mainly responsible for the on-orbit operation of space science experimental payloads”, Lin said.

The commander is Jing Haipeng — on his fourth mission into space, according to state media — and the third crew member is engineer Zhu Yangzhu.

They are set to take off from the Jiuquan Satellite Launch Centre in northwest China on Tuesday at 9:31 am (0131 GMT), the Manned Space Agency said.

Gui’s university, known as Beihang University in English, said he hailed from an “ordinary family” in western Yunnan province.

He “first felt the attraction of aerospace” listening to the news of China’s first man in space, Yang Liwei, on campus radio in 2003, the university said in a post on social media.

‘Space dream’

Under President Xi Jinping, plans for China’s “space dream” have been put into overdrive.

China is planning to build a base on the Moon and the country’s National Space Administration said it aims to launch a crewed lunar mission by 2029.

The final module of the T-shaped Tiangong — whose name means “heavenly palace” — successfully docked with the core structure last year.

The station carries a number of pieces of cutting-edge science equipment, state news agency Xinhua reported, including “the world’s first space-based cold atomic clock system”.

Once finished, Tiangong is expected to remain in low Earth orbit at between 400 and 450 kilometres (250 and 280 miles) above the planet for at least 10 years — realising an ambition to maintain a long-term human presence in space.

It will be constantly crewed by rotating teams of three astronauts, who will conduct scientific experiments and help test new technologies.

While China does not plan to use Tiangong for global cooperation on the scale of the International Space Station, Beijing said it is open to foreign collaboration.

It is not yet clear how extensive that cooperation will be.

China has been effectively excluded from the International Space Station since 2011, when the United States banned NASA from engaging with the country.

Continue Reading

Science

Elon Musk’s Neuralink Says It Has FDA Approval for Study of Brain Implants in Humans

Avatar

Published

on

The FDA nod "represents an important first step that will one day allow our technology to help many people," Neuralink said in a tweet.
By Reuters | Updated: 26 May 2023

Elon Musk’s brain-implant company Neuralink on Thursday said the US Food and Drug Administration (FDA) had given the green light to its first-in-human clinical trial, a critical milestone after earlier struggles to gain approval.

The FDA nod “represents an important first step that will one day allow our technology to help many people,” Neuralink said in a tweet. It did not elaborate on the aims of the study, saying only that it was not recruiting yet and more details would be available soon.

Neuralink and the FDA did not immediately respond to Reuters requests for comment.

Musk envisions brain implants could cure a range of conditions including obesity, autism, depression, and schizophrenia as well as enabling web browsing and telepathy. He made headlines late last year when he said he was so confident in the devices’ safety that he would be willing to implant them in his children.

On at least four occasions since 2019, Musk predicted Neuralink would begin human trials. But the company only sought FDA approval in early 2022 and the agency rejected the application, seven current and former employees told Reuters in March.

The FDA had pointed out several concerns to Neuralink that needed to be addressed before sanctioning human trials, according to the employees. Major issues involved the lithium battery of the device, the possibility of the implant’s wires migrating within the brain, and the challenge of safely extracting the device without damaging brain tissue.

Neuralink, founded in 2016, has been the subject of several federal probes.

In May, US lawmakers urged regulators to investigate whether the makeup of a panel overseeing animal testing at Neuralink contributed to botched and rushed experiments.

The Department of Transportation is separately probing whether Neuralink illegally transported dangerous pathogens on chips removed from monkey brains without proper containment measures.

Neuralink is also under investigation by the US Department of Agriculture’s Office of Inspector General for potential animal-welfare violations. This probe has also been looking at the USDA’s oversight of Neuralink.

Neuralink has not responded to requests for comment on the probes.

© Thomson Reuters 2023

Continue Reading

Trending