Author: peterjang73

Functional magnetic resonance imaging, or fMRI, is one of the most advanced tools for understanding how we think. As a person in an fMRI scanner completes various mental tasks, the machine produces mesmerizing and colorful images of their brain in action.

Looking at someone’s brain activity this way can tell neuroscientists which brain areas a person is using but not what that individual is thinking, seeing or feeling. Researchers have been trying to crack that code for decades—and now, using artificial intelligence to crunch the numbers, they’ve been making serious progress. Two scientists in Japan recently combined fMRI data with advanced image-generating AI to translate study participants’ brain activity back into pictures that uncannily resembled the ones they viewed during the scans. The original and re-created images can be seen on the researchers’ website.

“We can use these kinds of techniques to build potential brain-machine interfaces,” says Yu Takagi, a neuroscientist at Osaka University in Japan and one of the study’s authors. Such future interfaces could one day help people who currently cannot communicate, such as individuals who outwardly appear unresponsive but may still be conscious. The study was recently accepted to be presented at the 2023 Conference on Computer Vision and Pattern Recognition.

The study has made waves online since it was posted as a preprint (meaning it has not yet been peer-reviewed or published) in December 2022. Online commentators have even compared the technology to “mind reading.” But that description overstates what this technology is capable of, experts say.

“I don’t think we’re mind reading,” says Shailee Jain, a computational neuroscientist at the University of Texas at Austin, who was not involved in the new study. “I don’t think the technology is anywhere near to actually being useful for patients—or to being used for bad things—at the moment. But we are getting better, day by day.”

The new study is far from the first that has used AI on brain activity to reconstruct images viewed by people. In a 2019 experiment, researchers in Kyoto, Japan, used a type of machine learning called a deep neural network to reconstruct images from fMRI scans. The results looked more like abstract paintings than photographs, but human judges could still accurately match the AI-made images to the original pictures.

Neuroscientists have since continued this work with newer and better AI image generators. In the recent study, the researchers used Stable Diffusion, a so-called diffusion model from London-based start-up Stability AI. Diffusion models—a category that also includes image generators such as DALL-E 2—are “the main character of the AI explosion,” Takagi says. These models learn by adding noise to their training images. Like TV static, the noise distorts the images—but in predictable ways that the model begins to learn. Eventually the model can build images from the “static” alone.

Released to the public in August 2022, Stable Diffusion has been trained on billions of photographs and their captions. It has learned to recognize patterns in pictures, so it can mix and match visual features on command to generate entirely new images. “You just tell it, right, ‘A dog on a skateboard,’ and then it’ll generate a dog on a skateboard,” says Iris Groen, a neuroscientist at the University of Amsterdam, who was not involved in the new study. The researchers “just took that model, and then they said, ‘Okay, can we now link it up in a smart way to the brain scans?’”

The brain scans used in the new study come from a research database containing the results of an earlier study in which eight participants agreed to regularly lay in an fMRI scanner and view 10,000 images over the course of a year. The result was a huge repository of fMRI data that shows how the vision centers of the human brain (or at least the brains of these eight human participants) respond to seeing each of the images. In the recent study, the researchers used data from four of the original participants.

To generate the reconstructed images, the AI model needs to work with two different types of information: the lower-level visual properties of the image and its higher-level meaning. For example, it’s not just an angular, elongated object against a blue background—it’s an airplane in the sky. The brain also works with these two kinds of information and processes them in different regions. To link the brain scans and the AI together, the researchers used linear models to pair up the parts of each that deal with lower-level visual information. They also did the same with the parts that handle high-level conceptual information.

“By basically mapping those to each other, they were able to generate these images,” Groen says. The AI model could then learn which subtle patterns in a person’s brain activation correspond to which features of the images. Once the model was able to recognize these patterns, the researchers fed it fMRI data that it had never seen before and tasked it with generating the image to go along with it. Finally, the researchers could compare the generated image to the original to see how well the model performed.

Many of the image pairs the authors showcase in the study look strikingly similar. “What I find exciting about it is that it works,” says Ambuj Singh, a computer scientist at the University of California, Santa Barbara, who was not involved in the study. Still, that doesn’t mean scientists have figured out exactly how the brain processes the visual world, Singh says. The Stable Diffusion model doesn’t necessarily process images in the same way the brain does, even if it’s capable of generating similar results. The authors hope that comparing these models and the brain can shed light on the inner workings of both complex systems.

As fantastical as this technology may sound, it has plenty of limitations. Each model has to be trained on, and use, the data of just one person. “Everybody’s brain is really different,” says Lynn Le, a computational neuroscientist at Radboud University in the Netherlands, who was not involved in the research. If you wanted to have AI reconstruct images from your brain scans, you would have to train a custom model—and for that, scientists would need troves of high-quality fMRI data from your brain. Unless you consent to laying perfectly still and concentrating on thousands of images inside a clanging, claustrophobic MRI tube, no existing AI model would have enough data to start decoding your brain activity.

Even with those data, AI models are only good at tasks for which they’ve been explicitly trained, Jain explains. A model trained on how you perceive images won’t work for trying to decode what concepts you’re thinking about—though some research teams, including Jain’s, are building other models for that.

It’s still unclear if this technology would work to reconstruct images that participants have only imagined, not viewed with their eyes. That ability would be necessary for many applications of the technology, such as using brain-computer interfaces to help those who cannot speak or gesture to communicate with the world.

“There’s a lot to be gained, neuroscientifically, from building decoding technology,” Jain says. But the potential benefits come with potential ethical quandaries, and addressing them will become still more important as these techniques improve. The technology’s current limitations are “not a good enough excuse to take potential harms of decoding lightly,” she says. “I think the time to think about privacy and negative uses of this technology is now, even though we may not be at the stage where that could happen.”

A loose raft of brown seaweed spanning about twice the width of the U.S. is inching across the Caribbean. Currently, bucketloads of the buoyant algae are washing up on beaches on the eastern coast of Florida earlier in the year than usual, raising scientists’ concerns for what coming months will bring.

The seaweed is made up of algal species in the genus Sargassum. These species grow as a mat of glops of algae that stay afloat via little air-filled sacs attached to leafy structures. The algae form a belt between the Caribbean and West Africa in the Sargasso Sea in the North Atlantic Ocean and then ride the currents west. Scientists say that reports of a massive blob of seaweed slamming into coastlines are overblown because the Sargassum algae are scattered across the ocean, and much of the seaweed will never reach the coast’s sandy shores. But in recent years researchers have generally seen larger so-called Sargassum blooms. And once the seaweed begins washing up on beaches and rotting, it can cause serious problems, local communities say.

Among annual Sargassum censuses in the Atlantic Ocean, “2018 was the record year, and we’ve had several big years since,” says Brian Lapointe, an oceanographer at Florida Atlantic University, who has studied seaweed for decades. “This is the new normal, and we’re going to have to adapt to it.”

The seaweed “blob” has been dubbed the Great Atlantic Sargassum Belt, and though it’s sprawling, the algae in the belt cover only about 0.1 percent of the water’s surface, says Chuanmin Hu, an oceanographer at the University of South Florida, who has used satellites to study Sargassum for nearly 20 years.

Hu and his colleagues use data collected by NASA satellites, including Terra and Aqua, to estimate the total mass of Sargassum in the Atlantic every month, tracking a yearly cycle that typically peaks in June. Last year the seaweed broke the record for the highest amount ever recorded in the Atlantic, with some 22 million metric tons of the stuff found across the ocean, according to the team’s calculations.

Hu says the team estimated that the Atlantic contained about six million metric tons of Sargassum in February and that he’s confident March’s mass will be higher. “This month there should be more. There’s no doubt,” Hu says. “Even in the first two weeks, I have seen increased amounts.”

In the ocean, Hu says, the Sargassum is crucial habitat for fish and turtles, among other marine life. He calls the belt a “moving ecosystem.” And just a small portion of the seaweed present in the Atlantic will ever wash up on beaches, Hu adds.

But beaches in Fort Lauderdale and the Florida Keys are already reporting Sargassum deposits this year, Lapointe says, and it’s on beaches that the seaweed can be problematic. There, he says, the algae rot and release chemicals such as hydrogen sulfide gas, which smells like rotten eggs. When inhaled, the gas can also cause headaches and irritate a person’s eyes, nose and throat. People with asthma or other breathing problems may be more sensitive to the effect, according to the Florida Department of Health. The seaweed’s early arrival is raising concerns about what this summer might bring.

“This is pretty early in the Sargassum season to see that much coming in, so I think that’s also fueling some of the concern about what’s to come,” Lapointe says.

Hu says that Sargassum amounts can’t be forecast more than two or three months out, so this year’s seasonal peak in the summer is still too distant to predict. Researchers have expected this year might turn out to be heavy in seaweed, however, because even the winter lull saw higher amounts of the stuff than average.

And the Atlantic has been reliably producing much more Sargassum in recent decades than it has historically. Lapointe says that the high Sargassum levels of recent years are likely in part tied to nutrient-rich water running off land into rivers and out to the oceans, where it can fertilize the seaweed. But understanding and addressing the problem remains knotty, he adds.

“This has been going for over 10 years now, and we haven’t made a whole lot of progress in better understanding of all these nutrient and climate drivers,” he says. “It’s something we’re working on as scientists.”

Everyone wants in on AI these days, and for good reason. AI programs and services are pulling in millions of active users and dominating headlines. ChatGPT alone reached 100 million active users within two months. The latest players in the AI space are search engines, which want to transform traditional “googling” into educational conversations with artificial intelligence. DuckDuckGo is taking a slightly different approach, incorporating AI into search without the chat—instead, offering users clear, concise answers to their queries.

The search company is dubbing their new product “DuckAssist,” powered by natural language technology from OpenAI (creators of ChatGPT) and Anthropic. Unlike sophisticated and complex chatbots like Microsoft’s Bing or Google’s upcoming Bard, DuckAssist won’t offer you deeper learning through human-like conversations. Instead, the feature simply kicks in whenever you perform a search that could reasonably be answered by Wikipedia. When that happens, you’ll see DuckAssist appear with a blue “Ask” button: Choose it, and the AI will try to give you a clear answer to your query.

It works because DuckAssist is actually pulling its data from Wikipedia (as well as Britannica occasionally), and why not: If the AI can find the answer to your question in a series of cited articles on the free encyclopedia, it saves you the trip to Wikipedia yourself, as well as pouring through articles that sourced the same Wikipedia articles DuckAssist found. Plus, as Gizmodo’s Thomas Germain points out, limiting the information to these sources minimizes the issues other AI programs have had, including “hallucinations” and other bizarre behavior.

Because of how it’s currently designed, expect the best results from straightforward, fact-based questions, rather than subjective questions the AI would need to formulate an argument for. DuckDuckGo suggests questions like “what is a search engine index” over things like “what is the best search engine?” Good on them for not automatically saying “DuckDuckGo, of course!” In any case, all DuckAssist answers will provide a source, so you can see where it’s pulling the information from.

Like other AI models, DuckAssist isn’t actively connected to the internet. Rather, it’s pulling from a downloaded database of Wikipedia articles. DuckDuckGo says this database is “at most a few weeks old,” so don’t assume it will provide answers to questions that would need context more recent than early February 2023.

The best thing about DuckAssist, however, is that it’s available to everyone right now. Microsoft holds Bing’s AI chatbot behind a waitlist, while Google’s Bard is available to a limited pool of software testers. You can find the DuckAssist beta right now, however, in the iOS, Android, and Mac versions of the DuckDuckGo app, as well as the DuckDuckGo extensions for Firefox, Chrome, and Safari. It isn’t available for standard DuckDuckGo search yet, but there are still plenty of ways to try it out today.

If you use these forms of DuckDuckGo, DuckAssist will be on by default. You can disable DuckAssist, but only by disabling “Instant Answers” in settings, which will, as you’d expect, turn off Instant Answers in addition to the DuckAssist.

New research has found a possible link between having high levels of caffeine in your blood and a reduced risk of high body fat and type 2 diabetes. The authors say that clinical trials should be done to confirm whether calorie-free caffeinated drinks can help prevent these conditions.

Past studies have suggested a positive connection between coffee/caffeine consumption and reduced diabetes risk. But this research has largely relied on observational data, which can only point to a correlation between two factors, not show a direct cause-and-effect link. In this new study from scientists in Sweden and the UK, they decided to take a different approach, using a method called Mendelian randomization. The method tries to test whether having known genetic factors for one thing can directly affect the odds of the second factor.

In this case, the team focused on two common genetic variants that seem to slow down people’s metabolism of caffeine. As a result, these people tend to have higher blood caffeine levels, despite actually drinking less caffeinated beverages on average. Analyzing data from around 10,000 volunteers enrolled in other long-term studies, they tracked whether individuals carrying the variants were less likely to develop type 2 diabetes, cardiovascular disease, and other risk factors linked to both.

Overall, they found that people genetically predisposed to high blood caffeine levels were indeed less likely to have a high body mass index, high body fat, and type 2 diabetes. They also calculated that about 43% of this lower diabetes risk was attributable to being lower weight. However, they didn’t see any strong link between these variants and a changed risk of cardiovascular disease. The team’s findings were published Tuesday in the BMJ.

There are limitations to the findings, the authors acknowledge. For one, the sample comes almost entirely from people of European ancestry. The use of only two caffeine-related variants in the analysis also weakens any conclusions that can be drawn from it. And while Mendelian randomization is generally better at supporting a causal relationship between two factors than other types of studies, it’s still not a smoking gun —some studies using this same method haven’t found strong evidence between coffee consumption itself and lower diabetes risk.

That said, there are plausible mechanisms for how caffeine could lower diabetes risk. Caffeine is a stimulant, for instance, which can have short term effects on people’s appetite, and it might also increase people’s ability to burn fat or expend energy. At the very least, the authors say, it’s worth spending more resources to help settle this question.

“Randomized controlled trials are warranted to assess whether non-caloric caffeine-containing beverages might play a role in reducing the risk of obesity and type 2 diabetes,” they wrote.

Uber’s Record My Ride feature is expanding to even more cities across the United States. The feature gives both passengers and drivers the ability to easily record audio during a trip in an effort to boost safety for both parties involved.

What is Record My Ride?

Record My Ride is a safety feature from Uber that helps drivers and passengers document what’s happening during a trip if either party becomes uneasy or feels threatened. Using the Uber app, drivers and passengers can record audio to submit to the company in a safety report.

The feature was first piloted in 2019 in different countries in Latin America, before it was brought to select cities in the U.S. at the tail end of 2021. Uber claims that 70% of riders and drivers in Brazil’s Rio De Janeiro reported feeling safer while using the feature. While Uber can’t bring Record My Ride to the entirety of the U.S. since recording laws vary from state to state, the company expanded the feature to Cincinnati, Nashville, Phoenix, Salt Lake City, San Antonio and Tucson in November 2022.

At the same time, Uber revealed it was working on a dashcam-like video recording feature as part of a pilot program that included drivers in Cincinnati, Louisville, New York City, and Santos and João Pessoa in Brazil.

How do I access Record My Ride?

Record my Ride will only appear when you are a passenger or driver in an active ride. For passengers, tap the blue shield button in the bottom right corner of the map view during your ride to open the Safety Toolkit, and select Record Audio from the menu. Uber will immediately begin recording audio of your trip with your phone. For drivers, the blue shield icon is located in the bottom left of the map view.

Are there safety concerns with Record My Ride?

Uber says that the audio you record will remain on your phone and will stay encrypted until you file a report to Uber’s Safety Team, and add the audio as an attachment. Likewise, the company says that the passenger will get a notification when their driver begins recording audio, and vice versa, but the dashcam feature apparently sends no such notification. There is also the ongoing concern of a tech company collecting more data from its users to use in some way later on.

“Uber already has this treasure trove of highly personal data about people,” said Camille Fischer, a staff lawyer at the Electronic Frontier Foundation, in The New York Times. “When you pair surveillance during those trips, whether it’s over the driver or over the passenger, you are getting a more fine-tuned snapshot of people’s daily lives.”

Uber did not immediately return Gizmodo’s request for comment.

Regardless, Uber’s voyage into video and audio recording are born from reports of violence occurring in rides, in which drivers and passengers are often isolated from outside help. A driver in East Bay, California was brutally attacked by a passenger in early 2023, resulting in a $20,000 hospital bill. Uber is also the focus of a lawsuit waged against the company over reports of women being sexually assaulted by drivers as early as 2014.

Sometimes, the only thing that can get you through a long day of work is having a hot meal halfway through it. That’s not always an option if you’re stuck on a construction site, or even on the road, unless you’ve brought along Makita’s latest power tool: a cordless microwave that zaps instead of cuts.

If you thought your smartphone was a battery drainer, get ready for some depressing numbers. Electric motors, particularly those used in heavy duty power tools, can drain a normal rechargeable battery in mere minutes. As a result, there’s actually been a lot of innovation in power tool batteries to make them last as long as possible under heavy loads, and a recent push to make them compatible with other devices that usually rely on power outlets.

Companies like Dewalt now offer 60-volt rechargeable batteries that can work together to run even power-hungry table saws, while Makita has leveraged its own rechargeable battery tech to power everything from cordless coffee makers to jobsite wireless speakers.

Construction workers can’t run on empty, so Makita’s new cordless microwave is compatible with the company’s XGT system, allowing a pair of 40-volt rechargeable batteries to be clipped onto the back of the appliance for power. It’s not just targeted at construction sites, however; Makita’s new microwave could also ensure you don’t miss your hot morning coffee during a power outage, and seems like an ideal solution for camping (or glamping).Vanlifers living off the grid might also get a kick out of it.

With two fully-charged XGT batteries, Makita’s microwave can deliver 500-watts of power for eight minutes before stepping down to 350-watts until the batteries are dead. The batteries provide enough power to reheat about 11 lunches, or 20 hot drinks, but mileage will vary, particularly if starting with frozen meals. With just 0.28 cubic feet of capacity, space inside the microwave is also limited, but it should be large enough to handle single portion meals.

There’s a handle on top and an included strap which makes the 20lbs microwave (without batteries attached) easier to carry, but don’t expect to heat up a meal while it’s hanging off your shoulder. Not only does opening its door deactivate the appliance, but there’s a tilt sensor that also deactivates it when it’s being moved around.

For the time being, Makita has only announced availability for its new cordless microwave in Japan, where it will sell for ¥71,500, or around US $540. That’s definitely not cheap, and almost twice the cost of a well-equipped countertop model. But the joy of strapping this in to the passenger seat of your car and enjoying a piping hot breakfast burrito on your drive to work? Priceless.