The type of factual error that blighted the launch of Google's artificial intelligence-powered chatbot will carry on troubling companies using the technology, experts say, as the market value of its parent company continues to plunge. Investors in Alphabet marked down its shares by a further 4.4% to $95 on Thursday, representing a loss of market value of about $163bn (£140bn) since Wednesday when shareholders wiped around $106bn off the stock. Shareholders were rattled after it emerged that a video demo of Google's rival to the Microsoft-backed ChatGPT chatbot contained a flawed response to a question about Nasa's James Webb space telescope. The animation showed a response from the program, called Bard, stating that the JWST "took the very first pictures of a planet outside of our own solar system", prompting astronomers to point out this was untrue. Google said the error underlined the need for the "rigorous testing" that Bard is undergoing before a wider release to the public, which had been scheduled for the coming weeks.

ChatGPT is a version of GPT-3, a large language model also developed by OpenAI. Language models are a type of neural network that has been trained on lots and lots of text. Because text is made up of sequences of letters and words of varying lengths, language models require a type of neural network that can make sense of that kind of data. Recurrent neural networks, invented in the 1980s, can handle sequences of words, but they are slow to train and can forget previous words in a sequence. In 1997, computer scientists Sepp Hochreiter and Jürgen Schmidhuber fixed this by inventing LTSM (Long Short-Term Memory) networks, recurrent neural networks with special components that allowed past data in an input sequence to be retained for longer. LTSMs could handle strings of text several hundred words long, but their language skills were limited.

Shares of Google's parent company lost more than $100bn in market value on Wednesday after its Bard chatbot advertisement showed inaccurate information and analysts said its AI search event lacked details on how it will answer Microsoft's ChatGPT challenge. Reuters was the first to point out the error in Google's advertisement, which debuted Monday, about which satellite first took pictures of a planet outside the Earth's solar system. Shares of the company's parent Alphabet fell 8 percent or $8.59 a share to $99.05 and was one of the most actively traded on US exchanges. The tech giant posted a short GIF video of Bard in action via Twitter, describing the chatbot as a "launchpad for curiosity" that would help simplify complex topics, but it delivered an inaccurate answer that was spotted just hours before the launch event for Bard in Paris. "This is a hiccup here and they're severely punishing the stock for it, which is justified because obviously everybody is pretty excited to see what Google's going to counter with Microsoft coming out with a pretty decent product," said Dennis Dick, founder and market structure analyst at Triple D Trading.

Microsoft's search engine Bing is getting an AI refresh. At the company's campus in Redmond, Washington, today, executives unveiled a new version of Bing incorporating technology behind startup OpenAI's viral chatbot ChatGPT. The updates will see Bing results include smooth, written responses to queries that summarize information found on the web, and the addition of a new chatbot interface for complex queries. Satya Nadella, Microsoft's CEO, claimed the new features signal a paradigm for search. "In fact, a new race starts today," he said.

Similar to human co-workers, robots can make mistakes that violate a human's trust in them. When mistakes happen, humans often see robots as less trustworthy, which ultimately decreases their trust in them. The study examines four strategies that might repair and mitigate the negative impacts of these trust violations. These trust strategies were apologies, denials, explanations and promises on trustworthiness. An experiment was conducted where 240 participants worked with a robot co-worker to accomplish a task, which sometimes involved the robot making mistakes.

Q&A platform Quora has opened up public access to its new AI chatbot app, Poe, which lets users ask questions and get answers from a range of AI chatbots, including those from ChatGPT maker, OpenAI, and other companies like Anthropic. Beyond allowing users to experiment with new AI technologies, Poe's content will ultimately help to evolve Quora itself, the company says. Quora had first announced Poe's mobile app in December, but at the time, it required an invite to try it out. With the public launch on Friday, anyone can now use Poe's app. For now, it's available only to iOS users, but Quora says the service will arrive on other platforms in a few months.

AI chatbots are designed to answer questions and find information. They use what's on the internet as an enormous database of knowledge although there are concerns that this can also include offensive material and disinformation. ChatGPT is the best-known example.

Offering no criteria by which to define unacceptable risk for AI systems and no method to add new high-risk applications to the Act if such applications are discovered to pose a substantial danger of harm. This is particularly problematic because AI systems are becoming broader in their utility. Only requiring that companies take into account harm to individuals, excluding considerations of indirect and aggregate harms to society. An AI system that has a very small effect on, e.g., each person's voting patterns might in the aggregate have a huge social impact. Permitting virtually no public oversight over the assessment of whether AI meets the Act's requirements.

These image-generating AI models are trained on vast data sets consisting of images with text descriptions that have been scraped from the internet. The latest generation of the technology works by taking images in the data set and changing one pixel at a time until the original image is nothing but a collection of random pixels. The AI model then reverses the process to make the pixelated mess into a new image. The paper is the first time researchers have managed to prove that these AI models memorize images in their training sets, says Ryan Webster, a PhD student at the University of Caen Normandy in France, who has studied privacy in other image generation models but was not involved in the research. This could have implications for startups wanting to use generative AI models in health care, because it shows that these systems risk leaking sensitive private information.

Last week, the New York Attorney General's office sent Madison Square Garden Entertainment a letter demanding answers. The state's top law enforcement agency wants to know more about how the company operating Radio City Music Hall and the storied arena where the NBA's Knicks play uses a face recognition system to deny entry to certain people, and in particular lawyers representing clients in dispute with Madison Square Garden. The letter says that because the ban is thought to cover staff at 90 law firms, it may exclude thousands of people and deter them from taking on cases "including sexual harassment or employment discrimination claims." Since the face recognition system became widely known in recent weeks, MSG's management has stood squarely behind the idea of checking faces at the door with algorithms. In an unsigned statement, the company says its system is not an attack on lawyers, though some are "ambulance chasers and money grabbers."