OpenAI Has Declared ‘Code Red’ Multiple Times, Executive Says

(Bloomberg) — Sam Altman’s decision to declare a “code red” on OpenAI earlier this month may have caught the industry’s attention, but it wasn’t a first for the artificial intelligence company.

OpenAI Chief Research Officer Mark Chen said the San Francisco-based startup’s leadership had previously made the same statement, explicitly instructing employees to abandon low-priority tasks and focus on a single goal. “We do this when we want to make an effort to focus on a particular topic,” Chen said in an interview.

The latest code red comes two weeks after Alphabet Inc.’s Google released a widely acclaimed new AI model that outperformed OpenAI’s flagship software in a number of benchmarks. OpenAI’s Altman called on staff to redirect internal resources to accelerate improvements to ChatGPT and delay progress on other efforts such as autonomous AI agents and advertising.

“For me, what this means is conversation, judgment, and an effort to focus on the core ChatGPT product to make sure we get the fundamentals right,” Chen said. This includes ensuring that the chatbot operates quickly and reliably.

Since releasing code red, OpenAI has released several updates to support its flagship chatbot, including the release of a more advanced AI model designed to make ChatGPT better at coding, science, and a wide range of business tasks. The company also introduced a new image-generating artificial intelligence model to make images faster and better.

Heading into next year, Chen said his team is focused on finding the algorithms, infrastructure and research needed for OpenAI to train increasingly capable AI models. The company has committed to spending $1.4 trillion on infrastructure over the next eight years to support this effort.

“Everything else actually feels like a bit of an inconvenience or distraction,” he said.

More stories like this available Bloomberg.com