Megan Sauer

You might think OpenAI CEO Sam Altman, head of the company that makes ChatGPT, uses artificial intelligence chatbots in highly complex ways.

Not necessarily, according to Altman himself. “I use [AI] in the boring ways,” he told Wharton organizational psychologist Adam Grant’s “ReThinking” podcast, in an episode that published last month. “I use it for like, ‘Help me process all of this email’ or ’Help me summarize this document.”

Altman is in good company: Nvidia CEO Jensen Huang, whose company makes many of the high-powered computer chips that power AI large language models, primarily uses chatbots to help him write first drafts, he said at a Wired event in December. Microsoft CEO Satya Nadella, unsurprisingly, uses his Outlook’s AI features to organize and prioritize his inbox, he said at the Fast Company Innovation Festival 2024.

All three CEOs are market leaders in the AI industry — but their personal, day-to-day uses of AI seem pretty straightforward, much like other chatbot users across all job levels. The most common chatbot uses are coming up with ideas, consolidating information and automating basic tasks, according to a Gallup survey published last year.

It’ll take time for the technology to become more useful across a broad swath of industries, Altman said. The next step, he wrote in a January blog post: AI “agents,” built to automatically complete multi-step tasks off a single user prompt.

OpenAI publicly launched an agentic AI model, a ChatGPT feature called Operator, last month. The company says Operator can automate taskssuch as planning vacations, filling out forms, making restaurant reservations and ordering groceries.

Google released its own agentic AI product last week, and other technology companies in the AI arms race — like Meta, Microsoft, Amazon and Anthropic — are reportedly developing their own versions.

“Imagine that this agent will eventually be capable of doing most things a software engineer at a top company with a few years of experience could do,” wrote Altman. “It will not have the biggest new ideas, it will require lots of human supervision and direction, and it will be great at some things but surprisingly bad at others.”

It’s unclear how long it could take those agents to become commonplace at most people’s offices. Only about 13% of U.S. employees currently use AI at work, according to a January McKinsey and Company report.

CNBC