Group items matching

in title, tags, annotations or url

"But with all of the successes of AI, it's also important to pay attention to when, and how, it can go wrong, in order to prevent future errors. A recent paper by Roman Yampolskiy, director of the Cybersecurity Lab at the University of Louisville, outlines a history of AI failures which are "directly related to the mistakes produced by the intelligence such systems are designed to exhibit." According to Yampolskiy, these types of failures can be attributed to mistakes during the learning phase or mistakes in the performance phase of the AI system."

"No surprise, then, that Twitter is aglow with engineers and academics mocking Lemoine for falling into the seductive emptiness of his own creation. But while I agree that Lemoine has made a mistake, I don't think he deserves our scorn. His error is a good mistake, the kind of mistake we should want AI scientists to make."

"Plus, programming can teach valuable life lessons. However, in its storied past, coding wrought destruction as well. Instances of a little bit of bad code caused disaster on a major level. The following are 10 of the worst programming mistakes in history."

"According to tweets from DeepMind founder Demis Hassabis, however, this time AlphaGo really did make mistakes. The AI "thought it was doing well, but got confused on move 87," Hassabis said, later clarifying that it made a mistake on move 79 but only realized its error by 87. AlphaGo adjusts its playing style based on its evaluation of how the game is progressing."

""No-limit Texas Hold'em is a game of incomplete information where the AI must infer a human player's intentions and then act in ways that incorporate both the direct odds of winning and bluffing behaviour to try to fool the other player." The designers said their computer didn't "bluff" the human players. But by learning from its mistakes and practising its moves at night between games, the AI was working out how to defeat its human opponents."

"Scott is interested in conserving the stuff we have forgotten has value. Increasingly, our culture plays itself out on the internet, yet even now we have a tendency to view what we do on there as trivial. Or we make the mistake of assuming that digital means for ever."

"Unskilled manual laborers have felt the pressure of automation for a long time - but, increasingly, they're not alone. The last few years have been a bonanza of advances in artificial intelligence. As our software gets smarter, it can tackle harder problems, which means white-collar and pink-collar workers are at risk as well. Here are eight jobs expected to be automated (partially or entirely) in the coming decades. Call Center Employees call-center Telemarketing used to happen in a crowded call center, with a group of representatives cold-calling hundreds of prospects every day. Of those, maybe a few dozen could be persuaded to buy the product in question. Today, the idea is largely the same, but the methods are far more efficient. Many of today's telemarketers are not human. In some cases, as you've probably experienced, there's nothing but a recording on the other end of the line. It may prompt you to "press '1' for more information," but nothing you say has any impact on the call - and, usually, that's clear to you. But in other cases, you may get a sales call and have no idea that you're actually speaking to a computer. Everything you say gets an appropriate response - the voice may even laugh. How is that possible? Well, in some cases, there is a human being on the other side, and they're just pressing buttons on a keyboard to walk you through a pre-recorded but highly interactive marketing pitch. It's a more practical version of those funny soundboards that used to be all the rage for prank calls. Using soundboard-assisted calling - regardless of what it says about the state of human interaction - has the potential to make individual call center employees far more productive: in some cases, a single worker will run two or even three calls at the same time. In the not too distant future, computers will be able to man the phones by themselves. At the intersection of big data, artificial intelligence, and advanced

""Many predicted at the outset that Google would simply agree to most requests for removal to save time and money," said Barron. "That hasn't happened - 60% of removal requests have been rejected. Every single request is considered by our teams and there is no automation. If their judgment is challenged by the publisher we will reconsider and, if a mistake has been made, we will reinstate." "

"EFF's Legal Director Corynne McSherry offers five lessons to keep in mind: 1. (Lots of) mistakes will be made: copyright takedowns result in the removal of tons of legitimate content. 2. Robots won't help: automated filtering tools like Content ID have been a disaster, and policing copyright with algorithms is a lot easier than policing "bad speech." 3. These systems need to be transparent and have due process. A system that allows for automated instant censorship and slow, manual review of censorship gives a huge advantage to people who want to abuse the system. 4. Punish abuse. The ability to censor other peoples' speech is no joke. If you're careless or malicious in your takedown requests, you should pay a consequence: maybe a fine, maybe being barred form using the takedown system. 5. Voluntary moderation quickly becomes mandatory. Every voluntary effort to stem copyright infringement has been followed by calls to make those efforts mandatory (and expand them)."

"This type of approach can speed up learning times and improve the efficiency of algorithms, says Max Jaderberg at Google's AI company DeepMind. The company used a similar technique last year to teach an AI to explore a virtual maze. Its algorithm learned much more quickly than conventional reinforcement learning approaches. "Our agent is far quicker and requires a lot less experience from the world to train, making it much more data efficient," he says."

"Chinese delivery firm is moving to embrace automation.Chinese delivery firm is moving to embrace automation. Orange robots at the company's sorting stations are able to identify the destination of a package through a code-scan, virtually eliminating sorting mistakes. Shentong's army of robots can sort up to 200,000 packages a day, and are self-charging, meaning they are operational 24/7."

"But that was not before the falsely accused man had his name and image widely shared. The alert sent by Citizen contained a photo and was seen by more than 861,000 people. It read: "Citizen is offering a $30,000 reward to anyone who provides information that leads to the arrest of the arson suspect." Citizen told the Guardian in a statement it offered the cash reward "without formal coordination with the appropriate agencies". "Once we realized this error, we immediately retracted the photo and reward offer," it said. "We are actively working to improve our internal processes to ensure this does not occur again. This was a mistake we are taking very seriously.""

"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."

"Dubbed typosquatting, the idea is simple (if devious): A hacker registers a domain that is close enough to a real site, like yourbanknarne.com, and puts up a clone of yourbankname.com. The unsuspecting victim goes to the wrong site by mistake, and enters their personal banking information. In doing so, they have inadvertently handed the digital keys to their account to a hacker. "

"This all presents big questions for which we don't yet have answers. "At what point should kids know better?" asked David Dockterman, a lecturer at the Harvard Graduate School of Education. "When should a person's 'permanent digital record' start recording, if ever? To what extent should social media be a space for trial-and-error exploration around identity and social behavior?" "These are fantastically difficult moral dilemmas for teenagers who act impulsively, using tools that are not fully under their control, leading to consequences that perhaps none of us can anticipate," said Sonia Livingstone, professor of social psychology at the London School of Economics and Political Science. "This is the first time we've had a society in which almost by default, everything is recorded and shared and aggregated in ways that create a lifelong profile. Children should have the right to make mistakes.""

"One of the beautiful things about the internet is the sheer amount of knowledge it contains: if you're interested in any topic, you can find a surfeit of information about it in an instant. But this can also have a downside. Search engines can end perpetuating bias, for example. And research by Adrian Ward from the University of Texas, Austin suggests that we can mistake information we've searched for as our own knowledge. Now, in a new paper in the Journal of Consumer Psychology, Ward and colleagues have found that sharing information online also makes us feel that our knowledge has increased - even if we haven't read it."

"What is startling about the images in Furey's platform is that they contain mistakes so egregious and easy to spot that it makes one wonder how no one caught the issues, or if, alternatively, Furey believes the typical Toronto resident does in fact have three arms. An increasing number of political figures, particularly on the political right have been using generative AI images in campaigns. Florida governor and Republican presidential candidate Ron Desantis released what look like AI-generated images of his competition, former president Donald Trump, hugging former chief medical advisor Anthony Fauci. The Trump campaign had weeks earlier released a video mocking Desantis' wobbly twitter spaces campaign launch using AI-generated voices."

""I would reverse the permanent ban," Musk said on Tuesday, speaking via video link at a car industry conference organised by the Financial Times. "I do think it was not correct to ban Donald Trump," he said. "I think that was a mistake. It alienated the country and did not result in Donald Trump not having a voice."

""We submitted another two examples of real-life Amharic-language hate speech to them a week later," said Global Witness. "Both ads were, again, accepted by Facebook for publication within a matter of hours.""

"A fake, AI-generated image of Pope Francis stepping out in a stylish white puffer jacket and bejewelled crucifix racked up millions of views over the weekend - with many mistaking it for a real image. Experts fear the rapidly developing technology behind the image could soon undermine our ability to distinguish fake photos, which can be generated in seconds, from reality."