The company behind the ChatGPT app that churns out essays, poems or computing code on command on Tuesday released a long-awaited update of its artificial intelligence (AI) technology that it said would be safer and more accurate than its predecessor.
GPT-4 has been widely awaited ever since ChatGPT burst onto the scene in November last year, wowing users with its capabilities that were based on an older version of OpenAI’s technology, known as a large language model.
“We’ve created GPT-4, the latest milestone in OpenAI’s effort in scaling up deep learning,” a company blog said, adding that the AI technology “exhibits human-level performance” on some professional and academic tasks.
Photo: AFP
The company said the model is “more creative and collaborative than ever before” and would “solve difficult problems with greater accuracy” than its earlier versions.
With the update, text responses from GPT-4 would be more accurate, and — in future — would come from both image and text inputs in a major leap forward for the technology, although this aspect has not yet been released, it said.
For example, if a user sends a photograph of the inside of a refrigerator, GPT-4 would not only correctly identify what is there, but also concoct what could be prepared with those ingredients, it said.
OpenAI said it was working with a partner company, Be My Eyes, to prepare the next advance. Much of the new model’s firepower is now available to the general public via ChatGPT Plus, OpenAI’s paid subscription plan and on an AI-powered version of Microsoft Corp’s Bing search engine that is being tested.
OpenAI is backed by Microsoft, which earlier this year said it would finance the research company to the tune of billions of US dollars. Microsoft then swiftly integrated the tech into its Bing search engine, Edge browser and other products.
Earlier on Tuesday, Google-backed Anthropic, a start-up founded by former OpenAI executives, also announced the release of its Claude chatbot to business customers, while Amazon.com Inc, Baidu Inc (百度) and Facebook owner Meta Platforms Inc are also wading in, eager to avoid being left behind.
OpenAI said that the new version was far less likely to go off the rails than its earlier chatbot with widely reported interactions with ChatGPT or Bing’s chatbot in which users were presented with lies, insults or other so-called “hallucinations.”
“We spent six months making GPT-4 safer and more aligned. GPT-4 is 82 percent less likely to respond to requests for disallowed content and 40 percent more likely to produce factual responses,” OpenAI said.
Despite the anticipation, GPT-4 “is still flawed, still limited, and it still seems more impressive on first use than it does after you spend more time with it,” founder Sam Altman said.
“The power of the algorithm will increase, but it’s not a second revolution,” said Robert Vesoul, CEO of Illuin Technology, a French AI start-up.
Vesoul questioned the safety measures taken by OpenAI, which has already been hit by criticism from billionaire Elon Musk that the company is overly curbing speech on its AI to avoid more embarrassing responses.
“I am not sure if I want an AI to block responses on unknown topics... Should an AI decide if I smoke or not?” Vesoul said.
Taiwan Semiconductor Manufacturing Co (TSMC, 台積電), the world’s largest contract chipmaker, on Monday issued a statement about the balanced life environment it provides its employees, in response to a Fortune article at the weekend in which several former and current employees in the US were quoted complaining about the company’s “brutal” corporate culture. In the statement, TSMC said average work hours at the company have not exceeded 50 hours a week over the past two years with only a few exceptions, such as when the company introduces a new technology process or speeds up building a new plant. In such situations,
At a red-brick factory in the German port city of Hamburg, cocoa bean shells go in one end and out the other comes an amazing black powder with the potential to counter climate change. The substance, dubbed biochar, is produced by heating the cocoa husks in an oxygen-free room to 600°C. The process locks in greenhouse gases and the final product can be used as a fertilizer, or as an ingredient in the production of “green” concrete. While the biochar industry is still in its infancy, the technology offers a novel way to remove carbon from the Earth’s atmosphere, experts have said. Biochar could
Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) shares yesterday rallied 2 percent on the local stock market after Nvidia Corp said the contract chipmaker would be the sole supplier of its latest graphics processing chip, defusing speculation that Intel Corp would get a share of the orders. TSMC’s share price climbed to NT$562, snapping a three-day losing streak. It outperformed the benchmark index’s 1.18 percent gain. Net purchases by foreign institutional investors yesterday totaled 8.37 million shares, reversing net sales of 2.9 million shares on Thursday. The rebound follows Nvidia’s announcement that its latest artificial intelligence graphics processing unit (GPU), codenamed H100, would
WEAK PROSPECTS: The contract electronics manufacturer expects revenue to drop this quarter due to product transitions and a high comparison base a year earlier Hon Hai Precision Industry Co (鴻海精密), a major iPhone assembler, yesterday posted a second straight monthly growth in revenue to NT$45.07 billion (US$1.47 billion) last month, thanks to a pickup in smartphone demand. Last month’s revenue marked its best performance since January. On an annual basis, sales dipped 9.45 percent, the smallest annual decline since February, compared with NT$49.78 billion in May last year. Revenue from smart consumer electronics products, primarily iPhones and smartphones for other brands, delivered a “strong” double-digit, month-on-month growth in May because of customers’ pull-in, Hon Hai said in a company statement. That was the only business category