OpenAI will not train on customer data, says Sam Altman
OpenAI's CEO, Sam Altman, has confirmed that the company has not been training its large language models, such as GPT, on paying customer data "for a while." The company has changed its plans after customers made it clear that they did not want OpenAI to train on their data. OpenAI's terms of services were quietly updated on March 1. Altman revealed the news in an interview with CNBC. The company's portfolio of customers includes Microsoft, Salesforce, and Snapchat. Altman said, "We don't train on any API data at all."
Remote work experiment over, says Altman
During an event hosted by fintech firm Stripe, Altman also revealed that he believes the era of remote work is over. To create new products, staff need to work from the office, he said. Remote work only adds to confusion. Altman considered remote work an experiment and believes it was one of the tech industry's biggest mistakes to assume that everyone could go fully remote forever. Altman said, "There was going to be no loss of creativity. I would say that the experiment on that is over, and the technology is not yet good enough that people can be fully remote forever, particularly on startups."
OpenAI doubles losses to $540m in 2022
According to The Information, OpenAI doubled its losses to $540m in 2022 due to developing ChatGPT and hiring some executives from Google. Three people with the knowledge of the company's financials shared the information. The steep costs of training large language models were highlighted much before ChatGPT was made available to the public. Recently, OpenAI launched a paid version of the chatbot. However, as the number of customers using its AI technology increases, the costs of training future versions of its software are likely to rise.
