OpenAI’s CEO says a dearth of computing capacity is hindering his company’s product rollouts.
“All of these models have gotten quite complex,” Sam Altman wrote in a Reddit Ask Me Anything (AMA) session when asked why OpenAI’s next artificial intelligence (AI) models were taking so long. “We also face a lot of limitations and hard decisions about [how] we allocated our compute towards many great ideas.”
His comments were reported last week by TechCrunch, which said a number of recent media accounts indicate that OpenAI is struggling to find the computer infrastructure it needs to run and train its models.
Altman said that this strained capacity will delay the launch of Advanced Voice Mode, the realistic-sounding conversational feature for ChatGPT.
He also said during the AMA that there is no release plan yet for the latest version of DALL-E, the company’s image generator.
In addition, Sora, OpenAI’s video-generating tool, has been held up by the “need to perfect the model, get safety/impersonation/other things right, and scale compute,” wrote Kevin Weil, OpenAI’s chief product officer, who also took part in the AMA.
Also last week, OpenAI debuted a significant upgrade to its ChatGPT chatbot, integrating web search capabilities that could challenge traditional search engines by providing users with direct answers with source citations. It’s a move, as noted here, that “brings artificial intelligence deeper into everyday internet use.”
The new feature, being offered initially to ChatGPT Plus and Team subscribers, lets users receive current information about everything from weather forecasts to stock prices, marking a shift from ChatGPT’s offering, which was limited to historical data.
“The development represents a strategic push into territory long dominated by Google, offering what OpenAI positions as a more natural way to find information online,” PYMNTS wrote. “Rather than requiring users to sift through multiple search results, ChatGPT now synthesizes information while providing links to original sources.”
And in other recent AI news, PYMNTS examined the way the technology is impacting Thanksgiving this year, with guides circulating to help shoppers use generative AI to prepare for the annual food-related festivities.
“From budgeting to retooling recipes based on the number of guests who are coming to accommodating dietary restrictions, these guides suggest to consumers that AI can take some of the headache out of holiday planning,” that report said.
And research by PYMNTS Intelligence shows that many shoppers are interested in using AI to guide their buying journeys, with 44% of consumers saying they are least somewhat interested in AI being involved in how they shop. That figure climbs to 60% among Generation Z and 54% among millennials.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
The post OpenAI CEO: Computing Capacity Is Slowing Product Rollouts appeared first on PYMNTS.com.