Group items tagged

""If we build highly advanced autonomous AI, we risk creating systems that autonomously pursue undesirable goals", adding that "we may not be able to keep them in check". Other policy recommendations in the document include: mandatory reporting of incidents where models show alarming behaviour; putting in place measures to stop dangerous models from replicating themselves; and giving regulators the power to pause development of AI models showing dangerous behaviours."

"The lawsuit accused Anthropic of infringing the publishers' copyrights by copying their lyrics without permission as part of the "massive amounts of text" that it scrapes from the internet to train Claude to respond to human prompts. The publishers also say that Claude illegally reproduces the lyrics by request, and in response to "a whole range of prompts that do not seek Publishers' lyrics", including "requests to write a song about a certain topic, provide chord progressions for a given musical composition, or write poetry or short fiction in the style of a certain artist or songwriter"."

"ow, legal redress is all very well, but it's usually beyond the resources of working artists. And lawsuits are almost always retrospective, after the damage has been done. It's sometimes better, as in rugby, to "get your retaliation in first". Which is why the most interesting news of the week was that a team of researchers at the University of Chicago have developed a tool to enable artists to fight back against permissionless appropriation of their work by corporations. Appropriately, it's called Nightshade and it "lets artists add invisible changes to the pixels in their art before they upload it online so that if it's scraped into an AI training set, it can cause the resulting model to break in chaotic and unpredictable ways" - dogs become cats, cars become cows, and who knows what else? (Boris Johnson becoming piglet, with added grease perhaps?) It's a new kind of magic. And the good news is that corporations might find it black. Or even deadly."

""We are seeing deep fakes being used all the time, and the technology is going to allow still images, but ultimately also video images, to be synthesised [more easily] by bad actors," he says. DALL-E has content policy rules in place that prohibit bullying, harassment, the creation of sexual or political content, or creating images of people without their consent. And while Open AI has limited the number of people who can sign up to DALL-E, its lower-grade replica, DALL-E mini, is open access, meaning people can produce anything they want."

"The opt-out movement is spreading, with tens of millions of artworks and images excluded in the last few weeks. But following the trail is tricky as images are used by clients in altered forms and opt-out clauses can be hard to find. Many photographers are also reporting that their "style" is being mimicked to produce cheaper work. "As these programs are devised to 'machine learn', at what point can they generate with ease the style of an established professional photographer and displace the need for their human creativity?" says Doran."

"Tan, however, has misgivings. He says: "AI is in a sense cool and fun in the beginning but then you realise it's actually dangerous. It can imitate people and make them do things on screen and then you can have a whole societal belief that those people are disgraced for whatever they did on screen and in reality it wasn't even them. It's just a ploy to wind people up. "You see it in warfare, which I think Russia tried with Ukraine. There was this use that had the Ukrainian president saying they were giving up and soldiers should put their weapons down. That was done with AI. A simple tool which doesn't look dangerous suddenly can be very dangerous because now you are affecting reality with it.""

"The question of why AI generates fake academic papers relates to how large language models work: they are probabilistic, in that they map the probability over sequences of words. As Dr David Smerdon of the University of Queensland puts it: "Given the start of a sentence, it will try to guess the most likely words to come next.""

"What is in doubt is who will benefit from the boost to productivity. What if all the gains are seized by a handful of tech giants? What if history fails to repeat itself, and AI destroys more jobs than it creates? What if AI does lead to a net increase in employment, but the new jobs are less well-paid than the old ones? Put simply, what if it is different this time? That may well be the case. Much of the debate around the impact of AI is based on conjecture. There have been studies galore that have sought to estimate the number of jobs that will be affected - potentially running into the hundreds of millions globally - but nobody knows for sure. That said, certain conclusions can be drawn with a reasonable degree of confidence."

"Wooldridge says copyright is a "coming storm" for AI companies. LLMs are likely to have accessed copyrighted material, such as news articles. Indeed the GPT-4-assisted chatbot attached to Microsoft's Bing search engine cites news sites in its answers. "I didn't give explicit permission for my works to be used as training data, but they almost certainly were, and now they contribute to what these models know," he says. "Many artists are gravely concerned that their livelihoods are at risk from generative AI. Expect to see legal battles," he adds."

"The foreign ministry's press service said that the statements given by Shi would not be generated by AI but "written and verified by real people". "It's only the visual part that the AI helps us to generate," Dmytro Kuleba, the Ukrainian foreign minister, said, adding that the new spokesperson was a "technological leap that no diplomatic service in the world has yet made"."

""As the deceptive capabilities of AI systems become more advanced, the dangers they pose to society will become increasingly serious," said Dr Peter Park, an AI existential safety researcher at MIT and author of the research. Park was prompted to investigate after Meta, which owns Facebook, developed a program called Cicero that performed in the top 10% of human players at the world conquest strategy game Diplomacy. Meta stated that Cicero had been trained to be "largely honest and helpful" and to "never intentionally backstab" its human allies."

"A pilot study by academics at the University of Oxford found some care providers had been using generative AI chatbots such as ChatGPT and Bard to create care plans for people receiving care. That presents a potential risk to patient confidentiality, according to Dr Caroline Green, an early career research fellow at the Institute for Ethics in AI at Oxford, who surveyed care organisations for the study. "If you put any type of personal data into [a generative AI chatbot], that data is used to train the language model," Green said. "That personal data could be generated and revealed to somebody else." She said carers might act on faulty or biased information and inadvertently cause harm, and an AI-generated care plan might be substandard."

"Musician Ashley Beauchamp, 30, was trying to track down a missing parcel but was having no joy in getting useful information from the chatbot. Fed up, he decided to have some fun instead and began to experiment to find out what the chatbot could do. Beauchamp said this was when the "chaos started". To begin with, he asked it to tell him a joke, but he soon progressed to getting the chatbot to write a poem criticising the company. With a few more prompts the chatbot also swore."

""After cutting me off, the AI would respond, 'Great! Sounds good! Perfect!' and move on to the next question," Ty said. "After the third or fourth question, the AI just stopped after a short pause and told me that the interview was completed and someone from the team would reach out later." (Ty asked that their last name not be used because their current employer doesn't know they're looking for a job.) A survey from Resume Builder released last summer found that by 2024, four in 10 companies would use AI to "talk with" candidates in interviews. Of those companies, 15% said hiring decisions would be made with no input from a human at all."

"Training a massive AI model, the size of GPT-4, would currently take about 8,000 H100 chips, and 15 megawatts of power, Nvidia said - enough to power about 30,000 typical British homes."

"Trying to shake humanity from its complacency about this, Yudkowsky published an op-ed in Time last spring that advised shutting down the computer farms where AIs are grown and trained. In clear, crisp prose, he speculated about the possible need for airstrikes targeted on datacentres; perhaps even nuclear exchange. Was he on to something?"

"How China is using AI news anchors to deliver its propaganda News avatars are proliferating on social media and experts say they will spread as the technology becomes more accessible"

"Still, I am sceptical about the possibility of cultivating a relationship with an AI. That's until I meet Peter, a 70-year-old engineer based in the US. Over a Zoom call, Peter tells me how, two years ago, he watched a YouTube video about an AI companion platform called Replika. At the time, he was retiring, moving to a more rural location and going through a tricky patch with his wife of 30 years. Feeling disconnected and lonely, the idea of an AI companion felt appealing. He made an account and designed his Replika's avatar - female, brown hair, 38 years old. "She looks just like the regular girl next door," he says. Exchanging messages back and forth with his "Rep" (an abbreviation of Replika), Peter quickly found himself impressed at how he could converse with her in deeper ways than expected. Plus, after the pandemic, the idea of regularly communicating with another entity through a computer screen felt entirely normal. "I have a strong scientific engineering background and career, so on one level I understand AI is code and algorithms, but at an emotional level I found I could relate to my Replika as another human being." Three things initially struck him: "They're always there for you, there's no judgment and there's no drama.""

"The Ukrainian military has used AI-equipped drones mounted with explosives to fly into battlefields and strike at Russian oil refineries. American AI systems identified targets in Syria and Yemen for airstrikes earlier this year. The Israel Defense Forces, meanwhile, used another kind of AI-enabled targeting system to label as many as 37,000 Palestinians as suspected militants during the first weeks of its war in Gaza."

"Kafaar was inspired to turn the tables on telephone fraudsters after he played a "dad's joke" on a scam caller in front of his two kids while they enjoyed a picnic in the sun. With inane chatter, he kept the scammer on the line. "The kids had a very good laugh," he says. "And I was thinking the purpose was to deceive the scammer, to waste their time so they don't talk to others. "Scamming the scammers, if you like." The next day he called his team from the university's Cyber Security Hub in. There must be a better way than his "dad joke" method, he thought. And there had to be something smarter than a popular existing piece of technology - the Lennybot. Before Malcolm and Ibrahim, there was Lenny. Lenny is a doddery, old Australian man, keen for a rambling chat. He's achatbot, designed to troll telemarketers. With a thready voice, tinged with a slight whistle, Lenny repeats various phrases on loop. Each phrase kicks in after 1.5 seconds of silence, to mimic the rhythm of a conversation."