Group items tagged

"So they trained a deep-learning neural net on tons of examples of deepfaked videos, and produced a model that's better than any previous automated technique at spotting hoaxery. (Their paper documenting the work is here.)"

"In Neural Photo Editing With Introspective Adversarial Networks, a group of University of Edinburgh engineers and a private research colleague describe a method for using "introspective adversarial networks" to edit images in realtime, which they demonstrate in an open project called "Neural Photo Editor" that "enhances" photos by predicting what should be under your brush."

""Robots are a great educational tool for children. It inspires them to learn about science and engineering," Sharkey told the Guardian in March. "But there are significant dangers in having robots mind our children. They do not have the sensitivity or understanding needed for childcare." "

" The observations made at Facebook are the latest in a long line of similar cases. In each instance, an AI being monitored by humans has diverged from its training in English to develop its own language. The resulting phrases appear to be nonsensical gibberish to humans but contain semantic meaning when interpreted by AI "agents." "

"People will admit more if they are alone than if others are in the room with them. However, on sensitive topics, every survey method will elicit substantial misreporting. People have no incentive to tell surveys the truth. How, therefore, can we learn what our fellow humans are really thinking and doing? Big data. Certain online sources get people to admit things they would not admit anywhere else. They serve as a digital truth serum. Think of Google searches. Remember the conditions that make people more honest. Online? Check. Alone? Check. No person administering a survey? Check."

""This study shows that if this algorithm were in place earlier, some materials could have conceivably been discovered years in advance," Jain said."

"Raza wasn't the only one in her class who felt concerned about new levels of surveillance. Another student in the class, who did not want to be named, said that in addition to privacy worries, they were concerned that they didn't even have enough RAM to run the Proctorio software. Worse, the tool's facial detection algorithm seemed to struggle to recognize them, so they needed to sit in the full light of the window to better expose the contours of their face, in their view an indication that the system might be biased. "

"In the future, the team predict the software could also learn the appropriate emotion to express at a given time. "It could be mostly empathic," said Huang, adding that a challenge would be to avoid the chatbot reinforcing negative feelings such as rage."

" "Why isn't my face being detected? We have to look at how we give machines sight," she said in a TED Talk late last year. "Computer vision uses machine-learning techniques to do facial recognition. You create a training set with examples of faces. However, if the training sets aren't really that diverse, any face that deviates too much from the established norm will be harder to detect.""

"Researchers from MIT and Google recently showed off a machine learning algorithm capable of automatically retouching photos just like a professional photographer. Snap a photo and the neural network identifies exactly how to make it look better-increase contrast a smidge, tone down brightness, whatever-and apply the changes in less than 20 milliseconds."

"Lexplore claims its technology is new-particularly the algorithm that separates typical from atypical readers. But the concepts it's based on aren't. Its tech draws from a deep well of previously-conducted research stretching back decades, which is generally supportive of using a combination of eye tracking and machine learning to screen for dyslexia. "Eye movements is one of the best ways to index reading ability at an incredibly in-depth level," says Julie Kirkby, a psychology professor at Bournemouth University in the United Kingdom, who has studied eye tracking and dyslexia for years."

"Computers once performed facial recognition rather imprecisely, by identifying people's facial features and measuring the distances among them - a crude method that did not reliably result in matches. But recently, the technology has improved significantly, because of advances in artificial intelligence. A.I. software can analyze countless photos of people's faces and learn to make impressive predictions about which images are of the same person; the more faces it inspects, the better it gets. Clearview is deploying this approach using billions of photos from the public internet. By testing legal and ethical limits around the collection and use of those images, it has become the front-runner in the field. "

"A natural alternative to symbolic AI came to prominence: Instead of modeling high-level reasoning processes, why not instead model the brain? After all, brains are the only things that we know for certain can produce intelligent behavior. Why not start with them?"

"But BERT, which is now being deployed in services like Google's internet search engine, has a problem: It could be picking up on biases in the way a child mimics the bad behavior of his parents. BERT is one of a number of A.I. systems that learn from lots and lots of digitized information, as varied as old books, Wikipedia entries and news articles."

"The gulf between how well facial recognition performs in academic settings vs. real world applications is vast."

"Researchers showed detectors can be defeated by inserting inputs called adversarial examples into every video frame. The adversarial examples are slightly manipulated inputs which cause artificial intelligence systems such as machine learning models to make a mistake. In addition, the team showed that the attack still works after videos are compressed."

"John became increasingly radicalised by an online barrage of far-right disinformation. "Posts of homeless British soldiers were set against Muslim families being given free homes. Now I know the posts were all fake, but the 15-year-old me didn't bother to fact-check." The worry is that John's contemporaries won't either. A surge of online extremism and disinformation has arrived at a time of lockdown-induced isolation, loneliness and home-schooling, creating what police call a "perfect storm". One British far-right group has even started pushing an alternative white-supremacist school curriculum for lockdown learning."

"The difference between skimming and reading with all our intelligence is the difference between fully activated reading brains and their short-circuited, screen-dulled versions."

"Creating an algorithm that discriminates or shows bias isn't as hard as it might seem, however. As a first-year graduate student, my advisor asked me to create a machine-learning algorithm to analyze a survey sent to United States physics instructors about teaching computer programming in their courses."

"One of the problems with The Social Dilemma is that it supposes that tech did what it claims it did - that these are actually such incredible geniuses that they figured out how to use machine learning to control minds. And that's the problem - the mind control thing they designed to sell you fidget spinners got hijacked to make your uncle racist. But there's another possibility, which is that their claims are rubbish. They just overpromised in their sales material, and that what actually happened with that growth of monopolies and corruption in the public sphere made people cynical, angry, bitter and violent. In which case the problem isn't that their tools were misused. The problem is that the structures in which those tools were developed are intrinsically corrupt and corrupting."