OpenAI Envisions AI Voice as the Future of Computing

OpenAI believes the future of human-computer interaction may be driven more by conversation than keyboards. According to OpenAI President Greg Brockman, advanced voice interfaces such as the revamped ChatGPT Voice could eventually become the primary way people interact with computers for most everyday tasks.

From Typing to Talking

Speaking during a podcast with technology journalist Joanna Stern, Brockman said he expects users will eventually prefer talking to their computers instead of typing for “the vast majority” of tasks.

While he acknowledged keyboards will still have a place—particularly for short messages—he believes AI voice assistants will increasingly handle everything from brainstorming to software development.

Testing a Keyboard-Free Workflow

The idea was recently put to the test during a hands-free coding session using ChatGPT Voice alongside ChatGPT Work and Codex on the desktop application.

The experiment began awkwardly. Simple development tasks such as confirming the current working directory or navigating between project folders proved much slower through voice than with a keyboard.

Commands like:

• “Are we in the project?”
• “What files are you seeing now?”
• “Move up one directory.”

required repeated clarification, highlighting that precise filesystem navigation remains better suited to traditional input methods.

AI Agents Become the Real Advantage

As the session progressed, however, another capability emerged as the standout feature: AI agents working in parallel.

While the user continued a voice conversation—and even folded laundry—ChatGPT was instructed to:

• Investigate a bug in a Jellyfin media server.
• Check the health of a Raspberry Pi 3.
• Report progress as each task completed.

Rather than interrupting the conversation, ChatGPT dispatched separate agents to perform these jobs simultaneously, returning periodic status updates while the discussion continued.

Voice Works Well for Brainstorming

The experiment also demonstrated voice’s strengths during creative work.

Using only spoken conversation, ChatGPT reviewed an older software project, analyzed its files, discussed possible improvements, and helped generate ideas for an entirely new project focused on how local AI models handle uncertainty when answering questions.

After agreeing on the concept, ChatGPT created the initial project scaffolding and documentation in a new directory, allowing development to begin immediately.

Not Ready to Replace the Keyboard

Despite the productivity gains, the experience reinforced that voice is unlikely to replace keyboards anytime soon.

Tasks requiring precision—including directory navigation, code editing, and long-form writing—remain significantly easier with traditional input devices. The experiment concluded that while brainstorming and multitasking benefit from natural conversation, writing articles and performing detailed technical work still rely heavily on keyboard accuracy.

Voice as a Companion, Not a Replacement

OpenAI’s long-term vision suggests voice will become a central interface for AI-powered computing, particularly as autonomous agents become more capable of completing background tasks. However, current real-world experience indicates that voice is best viewed as a complement to the keyboard rather than its successor.

For now, conversational AI excels at ideation, delegation, and multitasking, while keyboards continue to offer the precision needed for programming, writing, and other detail-oriented workflows. Together, the two interaction methods may ultimately provide a more flexible way to work than either could alone.