Around 4 in the morning one Tuesday night in San Francisco, an autonomously-driven Cruise vehicle stopped at a red light — and was rear-ended by a Honda. But then "the Honda driver reversed backward several feet, stopped and drove forward again, making contact with the Cruise vehicle a second time," reports NBC News. After damaging the car and injuring its two test drivers, according to a collision report the Honda then "left the scene without exchanging information." It's just part of "a pattern bedeviling tech companies that are trying to make driverless cars a reality," reports NBC News, after reviewing collision reports from the California Department of Motor Vehicles: The reports, which were written by employees of the tech companies, describe 36 instances in 2022 in which a person driving a car or truck left the scene of a crash involving their vehicle and an autonomous vehicle. The problem has continued at a similar pace this year, with seven examples as of early March.... "My best guess is that the drivers think they can't be held liable," said Anderson Franco, a personal injury attorney in the city. "If you are operating your own vehicle and you crash into an autonomous vehicle, the correct thing to do is take photographs, call the police and have it documented," he said. But it's not always clear from the outside of a Cruise or other autonomous vehicle what to do if there's a problem. Cruise said in a statement to NBC News that it was in the process of making its phone number more prominently displayed on the outside of vehicles, so drivers in a crash know who to call.... The human drivers who have hit autonomous vehicles appear to be getting away with little accountability. Autonomous vehicles are usually equipped with a variety of external cameras that could record the license plate numbers of hit-and-run drivers but it's not clear how often the companies have gone down that road.... Cruise said in a statement that the hit-and-runs are usually minor. It said it works with San Francisco police "when necessary" and searches its videos for the license plate numbers of other cars "if needed." Cruise declined to comment on specific cases. Waymo said it has kept its options open about how to respond to hit-and-runs. California's Department of Motor Vehicles pointed out that because of the limited data available, "it's unclear if the rate of hit-and-run incidents involving autonomous vehicles is higher or lower than the rate involving conventional vehicles."

Read more of this story at Slashdot.

"Every day is April Fool's Day now, requiring a low but constant effort," argues Motherboard's senior editor, in a post shared by Slashdot reader samleecole. "As AI-generated shitposting becomes easier, it's inevitable that one of these will catch you with your guard down, or appeal to some basic emotion you are too eager to believe..." Even if you're trained in recognizing fake imagery and can immediately spot the difference between copy written by a language model and a human (content that's increasingly sneaking into online articles), doing endless fact-checking and performing countless micro-decisions about reality and fraud is mentally draining. Every year, our brains are tasked with processing five percent more information per day than the last. Add to this cognitive load a constant, background-level effort to decide whether that data is a lie. The disinformation apocalypse is already here, but not in the form of the Russian "dezinformatsiya" we feared. Wading through what's real and fake online has never been harder, not because each individual deepfake is impossible to distinguish from reality, but because the volume of low effort gags is outpacing our ability to process them.... Hany Farid, a professor at the University of California, Berkeley who's been studying manipulated media since long before deepfakes, told me that while he's used to getting a few calls every week from reporters asking him to take a look at images or videos that seem manipulated, over the past few weeks, he's gotten dozens of requests a day. "I don't even know how to put words to it. It really feels like it's unraveling," Farid told me in a phone call. When AI generated fakes started cropping up online years ago, he recalled, he warned that this would change the future, and some of his colleagues told him that he was overreacting. "The one thing that has surprised me is that it has gone much, much faster than I expected," he said. "I always thought, I agree that it is not the biggest problem today. But what's that Wayne Gretzky line? Don't skate to where the puck is, skate to where the puck is going. You've got to follow the puck. In this case, I don't think this was hard to predict." Buzzfeed noted that a viral image of the Pope in a white "puffer" coat" was created by a 31-year-old construction worker who created it while tripping on mushrooms, then posted it to Facebook. But Motherboard's article concludes with a quote from Peter Eckersley, the chief computer scientist for the Electronic Frontier Foundation who died in 2022. "There's a large and growing fraction of machine learning and AI researchers who are worried about the societal implications of their work on many fronts, but are also excited for the enormous potential for good that this technology processes." Eckersley said in a 2018 phone call. "So I think a lot of people are starting to ask, 'How do we do this the right way?' "It turns out that that's a very hard question to answer. Or maybe a hard question to answer correctly... How do we put our thumbs on the scale to ensure machine learning is producing a saner and more stable world, rather than one where more systems can be broken into more quickly?"

Read more of this story at Slashdot.

A YouTuber has published a video where he tricks ChatGPT into generating usable Windows 95 activation keys. Tom's Hardware reports: After asking Open AI's chatbot directly for Windows 95 keys, he received an expected reasoned refusal. YouTuber Enderman then asked the same thing but from a different angle. The result was a success which was somewhat limited by ChatGPT's ability to process natural language requests into formulas. [...] Some of the tested results were checked by attempting to activate a fresh Windows 95 install in a virtual machine. While the keys passed a casual inspection, it turns out that only about 1-in-30 keys seem to work as expected. So what is the problem with these keys? Enderman complains that "the only issue keeping ChatGPT from successfully generating valid Windows 95 keys almost every attempt is the fact that it can't count the sum of digits and it doesn't know divisibility." In the five-digit string divisible by seven section, the AI appears to provide a stream of random numbers that don't pass this simple mathematical test. The report adds: "[W]hile quizzing ChatGPT about key generating may be fun, it would have probably been more productive to manipulate the AI into writing a Python script to generate a conforming key or to DIY it."

Read more of this story at Slashdot.

A YouTuber has published a video where he tricks ChatGPT into generating usable Windows 95 activation keys. Tom's Hardware reports: After asking Open AI's chatbot directly for Windows 95 keys, he received an expected reasoned refusal. YouTuber Enderman then asked the same thing but from a different angle. The result was a success which was somewhat limited by ChatGPT's ability to process natural language requests into formulas. [...] Some of the tested results were checked by attempting to activate a fresh Windows 95 install in a virtual machine. While the keys passed a casual inspection, it turns out that only about 1-in-30 keys seem to work as expected. So what is the problem with these keys? Enderman complains that "the only issue keeping ChatGPT from successfully generating valid Windows 95 keys almost every attempt is the fact that it can't count the sum of digits and it doesn't know divisibility." In the five-digit string divisible by seven section, the AI appears to provide a stream of random numbers that don't pass this simple mathematical test. The report adds: "[W]hile quizzing ChatGPT about key generating may be fun, it would have probably been more productive to manipulate the AI into writing a Python script to generate a conforming key or to DIY it."

Read more of this story at Slashdot.

Italy's privacy watchdog said Friday it had blocked ChatGPT, saying the artificial intelligence app did not respect user data and could not verify users' age. The decision "with immediate effect" will result in "the temporary limitation of the processing of Italian user data vis-a-vis OpenAI," said the Italian Data Protection Authority.

Read more of this story at Slashdot.

An anonymous reader quotes a report from CNBC: Google is reshuffling the reporting structure of its virtual assistant unit -- called Assistant -- to focus more on Bard, the company's new artificial intelligence chat technology. In a memo to employees on Wednesday, titled "Changes to Assistant and Bard teams," Sissie Hsiao, vice president and lead of Google Assistant's business unit, announced changes to the organization that show the unit heavily prioritizing Bard. "As the Bard teams continue this work, we want to ensure we continue to support and execute on the opportunities ahead," Hsiao said in the email. "This year, more than ever, we have been focused on delivery with impact to our users." Jianchang "JC" Mao, who reported directly to Hsiao, will be leaving the company for personal reasons, according to the memo, which was viewed by CNBC. Mao held the position of vice president of engineering for Google Assistant and "helped shape the Assistant we have today," Hsiao wrote. Taking Mao's place will be 16-year Google veteran Peeyush Ranjan, who most recently held the title of vice president in Google's commerce organization, overseeing payments. The new leadership changes suggest that the Assistant organization may be planning on integrating Bard technology into similar products in the future. [...] As part of Wednesday's change, Google Assistant engineering vice president Amar Subramanya will now lead engineering for the Bard team, the email said. Trevor Strohman, who previously led engineering efforts for Bard, will continue as an "Area Tech Lead" for Bard, reporting to Hsiao. Ars Technica's Ron Amadeo ponders if the Google Assistant is facing a "looming Google shutdown." "If we assume the idea of the Google Assistant -- a voice assistant that helps you do things -- isn't completely dead at Google, you could imagine a future where Bard's language model helps it understand what you want to do and will do it, but it feels like the service is years away from something like that," writes Amadeo. "The Assistant today doesn't have language model problems, though, just voice recognition problems, and Bard won't help with that."

Read more of this story at Slashdot.

The nonconsensual deepfake economy has remained largely out of sight, but it's easily accessible, and some creators can accept major credit cards. From a report: Digitally edited pornographic videos featuring the faces of hundreds of unconsenting women are attracting tens of millions of visitors on websites, one of which can be found at the top of Google search results. The people who create the videos charge as little as $5 to download thousands of clips featuring the faces of celebrities, and they accept payment via Visa, Mastercard and cryptocurrency. While such videos, often called deepfakes, have existed online for years, advances in artificial intelligence and the growing availability of the technology have made it easier -- and more lucrative -- to make nonconsensual sexually explicit material. An NBC News review of two of the largest websites that host sexually explicit deepfake videos found that they were easily accessible through Google and that creators on the websites also used the online chat platform Discord to advertise videos for sale and the creation of custom videos. The deepfakes are created using AI software that can take an existing video and seamlessly replace one person's face with another's, even mirroring facial expressions. Some lighthearted deepfake videos of celebrities have gone viral, but the most common use is for sexually explicit videos. According to Sensity, an Amsterdam-based company that detects and monitors AI-developed synthetic media for industries like banking and fintech, 96% of deepfakes are sexually explicit and feature women who didn't consent to the creation of the content. Most deepfake videos are of female celebrities, but creators now also offer to make videos of anyone. A creator offered on Discord to make a 5-minute deepfake of a "personal girl," meaning anyone with fewer than 2 million Instagram followers, for $65.

Read more of this story at Slashdot.

An anonymous reader shares a report: AI image generator Midjourney has halted free trials of its service after a number of its generations -- including fabricated images of Donald Trump being arrested and the pope wearing a stylish jacket -- went viral online, with many mistaking the fakes for real photographs. Midjourney CEO and founder David Holz announced the change on Tuesday, citing "extraordinary demand and trial abuse."

Read more of this story at Slashdot.

Earlier today, more than 1,100 artificial intelligence experts, industry leaders and researchers signed a petition calling on AI developers to stop training models more powerful than OpenAI's ChatGPT-4 for at least six months. Among those who refrained from signing it was Eliezer Yudkowsky, a decision theorist from the U.S. and lead researcher at the Machine Intelligence Research Institute. He's been working on aligning Artificial General Intelligence since 2001 and is widely regarded as a founder of the field. "This 6-month moratorium would be better than no moratorium," writes Yudkowsky in an opinion piece for Time Magazine. "I refrained from signing because I think the letter is understating the seriousness of the situation and asking for too little to solve it." Yudkowsky cranks up the rhetoric to 100, writing: "If somebody builds a too-powerful AI, under present conditions, I expect that every single member of the human species and all biological life on Earth dies shortly thereafter." Here's an excerpt from his piece: The key issue is not "human-competitive" intelligence (as the open letter puts it); it's what happens after AI gets to smarter-than-human intelligence. Key thresholds there may not be obvious, we definitely can't calculate in advance what happens when, and it currently seems imaginable that a research lab would cross critical lines without noticing. [...] It's not that you can't, in principle, survive creating something much smarter than you; it's that it would require precision and preparation and new scientific insights, and probably not having AI systems composed of giant inscrutable arrays of fractional numbers. [...] It took more than 60 years between when the notion of Artificial Intelligence was first proposed and studied, and for us to reach today's capabilities. Solving safety of superhuman intelligence -- not perfect safety, safety in the sense of "not killing literally everyone" -- could very reasonably take at least half that long. And the thing about trying this with superhuman intelligence is that if you get that wrong on the first try, you do not get to learn from your mistakes, because you are dead. Humanity does not learn from the mistake and dust itself off and try again, as in other challenges we've overcome in our history, because we are all gone. Trying to get anything right on the first really critical try is an extraordinary ask, in science and in engineering. We are not coming in with anything like the approach that would be required to do it successfully. If we held anything in the nascent field of Artificial General Intelligence to the lesser standards of engineering rigor that apply to a bridge meant to carry a couple of thousand cars, the entire field would be shut down tomorrow. We are not prepared. We are not on course to be prepared in any reasonable time window. There is no plan. Progress in AI capabilities is running vastly, vastly ahead of progress in AI alignment or even progress in understanding what the hell is going on inside those systems. If we actually do this, we are all going to die. You can read the full letter signed by AI leaders here.

Read more of this story at Slashdot.

Companies rushing to adopt hot new types of artificial intelligence should exercise caution when using open-source versions of the technology, some of which may not work as advertised or include flaws that hackers can exploit, security researchers say. From a report: There are few ways to know in advance if a particular AI model -- a program made up of algorithms that can do such things as generate text, images and predictions -- is safe, said Hyrum Anderson, distinguished engineer at Robust Intelligence, a machine learning security company that lists the US Defense Department as a client. Anderson said he found that half the publicly available models for classifying images failed 40% of his tests. The goal was to determine whether a malicious actor could alter the outputs of AI programs in a manner that could constitute a security risk or provide incorrect information. Often, models use file types that are particularly prone to security flaws, Anderson said. It's an issue because so many companies are grabbing models from publicly available sources without fully understanding the underlying technology, rather than creating their own. Ninety percent of the companies Robust Intelligence works with download models from Hugging Face, a repository of AI models, he said.

Read more of this story at Slashdot.

Artificial intelligence experts, industry leaders and researchers are calling on AI developers to hit the pause button on training any models more powerful than the latest iteration behind OpenAI's ChatGPT. From a report: More than 1,100 people in the industry signed a petition calling for labs to stop training powerful AI systems for at least six months to allow for the development of shared safety protocols. Prominent figures in the tech community, including Elon Musk and Apple co-founder Steve Wozniak, were listed among the signatories, although their participation could not be immediately verified. "Recent months have seen AI labs locked in an out-of-control race to develop and deploy ever more powerful digital minds that no one -- not even their creators -- can understand, predict, or reliably control," said an open letter published on the Future of Life Institute website. "Powerful AI systems should be developed only once we are confident that their effects will be positive and their risks will be manageable."

Read more of this story at Slashdot.

Alphabet's Google is striking a partnership to combine its artificial intelligence language models with software from startup Replit that helps computer programmers write code, a bid to compete with a similar product from Microsoft's GitHub and OpenAI. From a report: Replit's Ghostwriter, which has 20 million users, will rely on Google's language-generation AI to improve its ability to suggest blocks of code, complete programs and answer developer questions. Google Cloud Vice President June Yang declined to specify which language AI products Replit will use, noting that it's a customized combination of systems that address different tasks like chat and code-generation. Previously, Replit built the product with its own AI. Google "has much better technology than most people know," Replit Chief Executive Officer Amjad Masad said in an interview. The startup will also expand its use of Google's cloud services and hopes the relationship with the tech giant will help it win over larger corporate customers -- right now Replit's clients are largely individual developers and startups. Google also will distribute Replit's software as part of the partnership. GitHub, which is wholly owned by Microsoft, last year released a product called Copilot, which suggests blocks of code as a software developer types, speeding up the process and automating rote or finicky coding tasks.

Read more of this story at Slashdot.

The latest breakthroughs in artificial intelligence could lead to the automation of a quarter of the work done in the US and eurozone, according to research by Goldman Sachs. From a report: The investment bank said on Monday that "generative" AI systems such as ChatGPT, which can create content that is indistinguishable from human output, could spark a productivity boom that would eventually raise annual global gross domestic product by 7 percent over a 10-year period. But if the technology lived up to its promise, it would also bring "significant disruption" to the labor market, exposing the equivalent of 300 million full-time workers across big economies to automation, according to Joseph Briggs and Devesh Kodnani, the paper's authors. Lawyers and administrative staff would be among those at greatest risk of becoming redundant. They calculate that roughly two-thirds of jobs in the US and Europe are exposed to some degree of AI automation, based on data on the tasks typically performed in thousands of occupations. Most people would see less than half of their workload automated and would probably continue in their jobs, with some of their time freed up for more productive activities. In the US, this should apply to 63 percent of the workforce, they calculated. A further 30 percent working in physical or outdoor jobs would be unaffected, although their work might be susceptible to other forms of automation. But about 7 percent of US workers are in jobs where at least half of their tasks could be done by generative AI and are vulnerable to replacement.

Read more of this story at Slashdot.

An anonymous reader quotes a report from the BBC: Facial recognition firm Clearview has run nearly a million searches for US police, its founder has told the BBC CEO Hoan Ton-That also revealed Clearview now has 30 billion images scraped from platforms such as Facebook, taken without users' permissions. [...] The company is banned from selling its services to most US companies, after the American Civil Liberties Union (ACLU) took Clearview AI to court in Illinois for breaking privacy law. But there is an exemption for police, and Mr Ton-That says his software is used by hundreds of police forces across the US. Police in the US do not routinely reveal whether they use the software, and it is banned in several US cities including Portland, San Francisco and Seattle. The use of facial recognition by the police is often sold to the public as only being used for serious or violent crimes. In a rare interview with law enforcement about the effectiveness of Clearview, Miami Police said they used the software for every type of crime, from murders to shoplifting. Assistant Chief of Police Armando Aguilar said his team used the system about 450 times a year, and that it had helped solve several murders. However, critics say there are almost no laws around the use of facial recognition by police.

Read more of this story at Slashdot.

Zoom is partnering with OpenAI to bring AI-generated summaries, message drafts, and more to the video conferencing app through its Zoom IQ AI-powered assistant. The Verge reports: While Zoom IQ can already do things like create chapters and highlights for recorded meetings, Zoom's giving the assistant even more features, including a way to catch up on meetings that you may have been late to. That means you'll be able to ask the tool to summarize what you've missed as well as "ask further questions." Additionally, Zoom IQ can do several other things, like generate whiteboards based on text prompts and provide recaps of meetings as well as summarize threads in Zoom Team Chat. Similar to the ChatGPT bot coming to Slack, Zoom IQ also lets you generate responses to your colleagues using AI. The company says it's planning to roll out AI-powered message and email drafts on an invitation-only basis in April but will introduce "select" Zoom IQ meeting summary features "more broadly."

Read more of this story at Slashdot.

EU police force Europol on Monday warned about the potential misuse of artificial intelligence-powered chatbot ChatGPT in phishing attempts, disinformation and cybercrime, adding to the chorus of concerns ranging from legal to ethical issues. From a report: "As the capabilities of LLMs (large language models) such as ChatGPT are actively being improved, the potential exploitation of these types of AI systems by criminals provide a grim outlook," Europol said as it presented its first tech report starting with the chatbot. It singled out the harmful use of ChatGPT in three areas of crime. "ChatGPT's ability to draft highly realistic text makes it a useful tool for phishing purposes," Europol said. With its ability to reproduce language patterns to impersonate the style of speech of specific individuals or groups, the chatbot could be used by criminals to target victims, the EU enforcement agency said.

Read more of this story at Slashdot.

The US Federal Trade Commission is paying close attention to developments in artificial intelligence to ensure the field isn't dominated by the major tech platforms, Chair Lina Khan said Monday. From a report: "As you have machine learning that depends on huge amounts of data and also a huge amount of storage, we need to be very vigilant to make sure that this is not just another site for big companies to become bigger," Khan said at an event hosted by the Justice Department in Washington. Khan said companies offering AI tools need to make sure they are not "overselling or overstating" what their products can do. "Sometimes we see claims that are not fully vetted or not really reflecting how these technologies work," Khan said, noting recent guidance from the agency on AI-enabled products. "Developers of these tools can potentially be liable if technologies they are creating are effectively designed to deceive."

Read more of this story at Slashdot.

Apple has quietly acquired a Mountain View-based startup, WaveOne, that was developing AI algorithms for compressing video. From a report: Apple wouldn't confirm the sale when asked for comment. But WaveOne's website was shut down around January, and several former employees, including one of WaveOne's co-founders, now work within Apple's various machine learning groups. In a LinkedIn post published a month ago, WaveOne's former head of sales and business development, Bob Stankosh, announced the sale. "After almost two years at WaveOne, last week we finalized the sale of the company to Apple," Stankosh wrote. "We started our journey at WaveOne, realizing that machine learning and deep learning video technology could potentially change the world. Apple saw this potential and took the opportunity to add it to their technology portfolio." WaveOne was founded in 2016 by Lubomir Bourdev and Oren Rippel, who set out to take the decades-old paradigm of video codecs and make them AI-powered. Prior to joining the venture, Bourdev was a founding member of Meta's AI research division, and both he and Rippel worked on Meta's computer vision team responsible for content moderation, visual search and feed ranking on Facebook.

Read more of this story at Slashdot.

ZDNet columnist Chris Matyszczyk flags Microsoft's latest "poetic use of words" when describing the ChatGPT-based functionalities it's bundling into applications like Word and Excel. It's there to help steer you to your destination. It's there to free you to focus on steering your life. And it's there to help you land on the perfect version of you, the one that does more in order to, I don't know, be more. There's one difference, though, between Microsoft's Copilot and, say, an American Airlines co-pilot. Hark the words of Microsoft VP of Modern Work and Business Applications Jared Spataro: "Sometimes, Copilot will get it right. Other times, it will be usefully wrong, giving you an idea that's not perfect, but still gives you a head start." I wonder how long it took for someone to land on the concept of "usefully wrong." You wouldn't want, say, the steering wheel on your car to be usefully wrong. Any more than you'd want your electrician to be usefully wrong. Somehow, though, one is supposed to cheer that a piece of AI (hurriedly) slipped into one's most basic business tools can be utterly mistaken.... Of course, all these companies — Microsoft, too — claim they're being responsible in the way they create their new offerings. Wait, didn't Microsoft just lay off its entire AI ethics and society team?

Read more of this story at Slashdot.

Bill Gates calls the invention of AI "as fundamental as the creation of the microprocessor, the personal computer, the Internet, and the mobile phone," predicting "Entire industries will reorient around it" in an essay titled "The AI Age has Begun." In my lifetime, I've seen two demonstrations of technology that struck me as revolutionary. The first time was in 1980, when I was introduced to a graphical user interface — the forerunner of every modern operating system, including Windows.... The second big surprise came just last year. I'd been meeting with the team from OpenAI since 2016 and was impressed by their steady progress. In mid-2022, I was so excited about their work that I gave them a challenge: train an artificial intelligence to pass an Advanced Placement biology exam. Make it capable of answering questions that it hasn't been specifically trained for. (I picked AP Bio because the test is more than a simple regurgitation of scientific facts — it asks you to think critically about biology.) If you can do that, I said, then you'll have made a true breakthrough. I thought the challenge would keep them busy for two or three years. They finished it in just a few months. In September, when I met with them again, I watched in awe as they asked GPT, their AI model, 60 multiple-choice questions from the AP Bio exam — and it got 59 of them right. Then it wrote outstanding answers to six open-ended questions from the exam. We had an outside expert score the test, and GPT got a 5 — the highest possible score, and the equivalent to getting an A or A+ in a college-level biology course. Once it had aced the test, we asked it a non-scientific question: "What do you say to a father with a sick child?" It wrote a thoughtful answer that was probably better than most of us in the room would have given. The whole experience was stunning. I knew I had just seen the most important advance in technology since the graphical user interface. Some predictions from Gates: "Eventually your main way of controlling a computer will no longer be pointing and clicking or tapping on menus and dialogue boxes. Instead, you'll be able to write a request in plain English...." "Advances in AI will enable the creation of a personal agent... It will see your latest emails, know about the meetings you attend, read what you read, and read the things you don't want to bother with." "I think in the next five to 10 years, AI-driven software will finally deliver on the promise of revolutionizing the way people teach and learn. It will know your interests and your learning style so it can tailor content that will keep you engaged. It will measure your understanding, notice when you're losing interest, and understand what kind of motivation you respond to. It will give immediate feedback." "AIs will dramatically accelerate the rate of medical breakthroughs. The amount of data in biology is very large, and it's hard for humans to keep track of all the ways that complex biological systems work. There is already software that can look at this data, infer what the pathways are, search for targets on pathogens, and design drugs accordingly. Some companies are working on cancer drugs that were developed this way." AI will "help health-care workers make the most of their time by taking care of certain tasks for them — things like filing insurance claims, dealing with paperwork, and drafting notes from a doctor's visit. I expect that there will be a lot of innovation in this area.... AIs will even give patients the ability to do basic triage, get advice about how to deal with health problems, and decide whether they need to seek treatment."

Read more of this story at Slashdot.