ChatGPT’s creators have attempted to get the system to explain itself. They found that while they had some success, they ran into some issues – including the fact that artificial intelligence may be using concepts that humans do not have named for, or understanding of. Researchers at OpenAI, which developed ChatGPT, used the most recent version of its model known as GPT-4 to try and explain the behaviour of GPT-2, an earlier version. It is an attempt to overcome the so-called black box problem with large language models such as GPT. While we have a relatively good understanding of what goes into and comes out of such systems, the actual work that goes on inside remains largely mysterious. That is not only a problem because it makes things difficult for researchers. It also means that there is little way of knowing what biases might be involved in the system, or if it is providing false information to people using it, since there is no way of knowing how it came to the conclusions it did. Engineers and scientists have aimed to resolve this problem with “interpretability research”, which seeks find ways to look inside the model itself and better understand what is going on. That has often required looking at the “neutrons” that make up such a model: just like in the human brain, an AI system is made up of a host of so-called neutrons that represent parts of the data it uses. Finding those is difficult, however, since humans have had to pick through the neurons and manually inspect them to find out what they represent. But some systems have hundreds of billions of parameters and so actually getting through them all with people is impossible. Now, researchers at OpenAI have looked to use GPT-4 to automate that process, in an attempt to more quickly pick through the behaviour. They did so by attempting to create an automated process that would allow the system to provide natural language explanations of the neuron’s behaviour – and apply that to another, earlier language model. That worked in three steps: looking at the neuron in GPT-2 and having GPT-4 try and explain it, then simulating what that neuron would, and finally scoring that explanation by comparing how the simulated activation worked with the real one. Most of those explanations went badly, and GPT-4 scored itself poorly. But researchers said that they hoped the experiment showed that it would be possible to use the AI technology to explain itself, with further work. The creators came up against a range of “limitations”, however, that mean the system as it exists now is not as good as humans at explaining the behaviour. Part of the problem may be that explaining how the system is working in normal language is impossible – because the system may be using individual concepts that humans cannot name. “We focused on short natural language explanations, but neurons may have very complex behavior that is impossible to describe succinctly,” the authors write. “For example, neurons could be highly polysemantic (representing many distinct concepts) or could represent single concepts that humans don’t understand or have words for.” It also runs into problems because it is focused on specifically what each neuron does individually, and not how that might affect things later on in the text. Similarly, it can explain specific behaviour but not what mechanism is producing that behaviour, and so might spot The system also uses a lot of computing power, the researchers note. Read More Google to unveil major new AI AI robots figure out how to play football in shambolic footage White House asks hackers to break ChatGPT White House reveals plan to ‘protect’ citizens from danger of AI DeepMind boss says human-level AI is just a few years away Regulator to probe use of artificial intelligence such as ChatGPT

HILLIARD, Ohio--(BUSINESS WIRE)--Aug 15, 2023--

FC Barcelona President Joan Laporta has hailed the creation of Barca Games.

The introduction of robots into businesses actually slows – at least at first, according to a new study. Researchers found that the introduction of robots bring down profit margins. But as they technology becomes better integrated, it will start to rise again, the study found. Researchers believe that U-shaped curve comes about because of reduced costs, new processes and innovative products. When companies first adopt robots with the aim of reducing costs, competitors are able to do the same, and so profit margins do not initially grow. The real profits come, however, when the robots are properly adopted and integrated into the company’s processes, and they can use that new innovation to develop new products, the researchers suggest. Those are the findings from a new study from the University of Cambridge and published in IEEE Transactions on Engineering Management. While robots are known to increase productivity when looking at a whole industry or country, it is less clear whether it helps with profit margins. The researchers set out to answer that question and see whether companies were using robots to improve processes within companies. And they were also looking to understand whether it had followed the same perhaps unexpected trajectory as when computers were first introduced into businesses. “If you look at how the introduction of computers affected productivity, you actually see a slowdown in productivity growth in the 1970s and early 1980s, before productivity starts to rise again, which it did until the financial crisis of 2008,” said co-author Professor Chander Velu from Cambridge’s Institute for Manufacturing. “It’s interesting that a tool meant to increase productivity had the opposite effect, at least at first. We wanted to know whether there is a similar pattern with robotics.” To find out, researchers gathered data for 25 European countries that showed industry-level trends between 1995 and 2017. That data did not include specific companies but did allow them to see whole sectors. They then gathered robotics data from the International Federation of Robotics (IFR). By setting the two against each other, they were able to see how the adoption of robotics changed profit margins. There they found that U-shaped curve: that the adoption of robotics drove down profits, initially, even if it came back eventually. “Initially, firms are adopting robots to create a competitive advantage by lowering costs,” said Velu. “But process innovation is cheap to copy, and competitors will also adopt robots if it helps them make their products more cheaply. This then starts to squeeze margins and reduce profit margin.” Read More Apple planning new version of AirTags – but not for some time Algorithm finds 600-foot, ‘potentially hazardous’ asteroid near Earth Reddit closes Place after obscene protests Apple planning new version of AirTags – but not for some time Algorithm finds 600-foot, ‘potentially hazardous’ asteroid near Earth Reddit closes Place after obscene protests

Apple has released a new iPhone update that ensures they won’t break when being charged in the car. The iOS 17.1.1 update includes a range of bug fixes for some unusual and high-profile issues with Apple’s new iPhones. It is available to download now as usual. When the iPhone 15 was released, some users of specific cars – particularly those made by BMW and Toyota – reported that charging their phones in the car would break them. In some cases, when people used the built-in wireless charging then Apple Pay would stop working, for instance. That is now fixed in the new update, according to Apple’s release notes. It fixes an issue where “In rare circumstances, Apple Pay and other NFC features may become unavailable on iPhone 15 models after wireless charging in certain cars”, it says. The new operating system update also fixes an issue where “Weather Lock Screen widget may not correctly display snow”, the release notes say. That refers to a problem where users would instead see a file icon, which had been widely reported. Apple has also released an update for the Watch, numbered 10.1.1. That fixes another prominent issue that led to unexpected battery drain on the Watch, with some reporting that their devices lost all their charge in just a few hours. The Mac and HomePod also have their own bug updates. They can all be installed as usual: through the Settings app on the iPhone or Mac, or through the devoted apps for the Watch and HomePod, though all devices should eventually prompt the user to update automatically. The update came after a Bloomberg report that indicated Apple had taken some engineers off work on upcoming major updates to ensure that the current operating system versions were as free of bugs as possible. Recent weeks have seen a range of unexpected bugs in Apple software, including widespread reports that the iPhone 15 would get uncomfortably hot, which was later fixed with another software update. Read More The Apple Watch has a major issue but Apple is working on a fix Something is happening with Apple’s Mac and iPads Apple just revealed a new MacBook Pro – with a new colour

Foxconn says it is exiting an ambitious project to help build one of India's first chip factories.

Elon Musk has admitted that his endorsement of an antisemitic conspiracy theory on his social media site X, previously known as Twitter, was a “mistake” and “one of the most foolish” things he had done on the platform. The world’s richest man said he had “no problem being hated” but also blamed members of the media for much of the outrage surrounding the post, which is thought to have contributed to major investors in the platform pulling funding. Earlier this month, Mr Musk responded to a Twitter/X user who accused Jewish people of hating white people by saying it was “the actual truth.” His actions sparked global outrage, with several world leaders condemning his endorsement of the conspiracy theory. Speaking at the New York Times’ Dealbook Summit this week, Mr Musk discussed the post and said that the outrage caused was “not my intention”. “[It was] one of the most foolish — if not the most foolish — thing I’ve done on the platform,” he said. “I should in retrospect not have replied to that one person and should have written in greater length what I meant,” he said. “But those clarifications were ignored by the media and essentially I handed a loaded gun to those who hate me and arguably to those who are antisemitic. And for that I’m quite sorry, that was not my intention.” Earlier this week, Mr Musk visited Israel and toured a kibbutz attacked by Hamas, meeting with Israel’s prime minister Benjamin Netanyahu. The pair visited the Kfar Aza kibbutz, which Hamas attacked on 7 October, killing an estimated 52 residents. Another 20 people are reported as missing. Mr Musk said witnessing the scenes of the massacre was “jarring”, during a live event on Twitter/X with Mr Netanyahu after the tour. Speaking at the Dealbook Summit, he denied that the visit had been – as some had described it – an “apology tour”. “[The trip to Israel] wasn’t in response to that at all,” Mr Musk said, adding: “I have no problem being hated.” Mr Musk later went further, addressing companies that had pulled marketing from X over concerns related to far-right content and accusing them of “blackmail”. “Don’t advertise,” he said at the New York Times’ Dealbook Summit. “If someone is going to try to blackmail me with advertising, blackmail me with money? Go f*** yourself. Go f*** yourself. Is that clear? I hope it is.” Read More Elon Musk believes OpenAI may have made ‘dangerous’ discovery OpenAI may have made a ‘dangerous’ artificial intelligence discovery, Elon Musk says Elon Musk publicly tells advertisers to ‘go f*** yourselves’ Elon Musk mocked for trying to resurrect QAnon Pizzagate conspiracy Elon Musk set to meet Netanyahu and hostage families in Israel Elon Musk weighs in on Dublin riots claiming country’s PM ‘hates the Irish people’

More than 100,000 user accounts of the popular artificial intelligence chatbot platform ChatGPT have been compromised over the last year using information-stealing malware, a new report has revealed. The report, published by Singapore-based cybersecurity firm Group-IB, identified 101,134 compromised accounts, the credentials of many of which have been traded over the last year on illicit dark web marketplaces. At its peak in May, nearly 27,000 credentials of compromised ChatGPT accounts were traded on the dark web, the group noted, adding that the Asia-Pacific region experienced the highest concentration of ChatGPT credentials offered for sale. This region, according to the report, accounted for almost 40 per cent of compromised accounts between June 2022 and May 2023, followed by Europe. Since its widespread rollout in November last year, ChatGPT has seen growing use, with employees taking advantage of the chatbot to optimise their work across fields from software development to business communications. As the chatbot stores the history of user queries and the AI’s responses, experts have warned that unauthorised access to ChatGPT accounts could expose confidential or sensitive information. “Employees enter classified correspondences or use the bot to optimize proprietary code. Given that ChatGPT’s standard configuration retains all conversations, this could inadvertently offer a trove of sensitive intelligence to threat actors if they obtain account credentials,” said Dmitry Shestakov, the head of threat intelligence at Group-IB. Several businesses, institutions and universities across the world, including several in Japan, have either banned use of the chatbot, or have warned staff to not reveal sensitive information to the AI bot as such data can be exploited for targeted attacks against companies and their employees. The Singapore-based cybersecurity group warned in its latest report that ChatGPT accounts have already gained popularity within underground communities on the dark web that are accessible only via special software. Using malicious software known as info stealers, credentials saved in browsers, bank card details, crypto wallet information, cookies, browsing history and other information from browsers installed on infected computers are being stolen and sent to operators. Logs containing user information, including data on the IP addresses, are being actively traded on dark web marketplaces, according to Group-IB. A majority of logs containing ChatGPT accounts have been breached by the infamous Raccoon info stealer, the group noted. Experts urge users to update passwords regularly and implement two-factor authentication for accessing their ChatGPT accounts. Users are also advised to disable the chatbot’s chat saving feature from its settings menu or manually delete conversations immediately after use. Read More ChatGPT ‘grandma exploit’ gives users free keys for Windows 11 Protect personal data when introducing AI, privacy watchdog warns businesses How Europe is leading the world in the push to regulate AI Scientists warn of threat to internet from AI-trained AIs ChatGPT ‘grandma exploit’ helps people pirate software Hundreds attend ‘soulless’ AI-generated church service

In a recent exchange on Twitter, MrBeast and Jacksepticeye have resolved their differences

So you've signed up for Threads. Despite the millions who've supposedly also joined, it's kind

HOUSTON--(BUSINESS WIRE)--May 9, 2023--

JOHANNESBURG (Reuters) -U.S. ecommerce firm Amazon said on Tuesday it would launch its online shopping service in South Africa in