Voice, Spatial and Beyond

The long arc of the interface

New interfaces rarely replace old ones. The keyboard survived the mouse, the mouse survived touch, and touch will almost certainly survive voice and spatial input. What changes is the default: the mode most people reach for first. History suggests that a new default takes hold only when three things line up at once: a capable device, an affordable price and at least one task that is clearly better done the new way.

The timeline below marks moments where those conditions began to align. Dates refer to public demonstrations or consumer releases.

What looks real, and what still looks like a demo

Conversation is finally conversational

Earlier voice assistants failed because they parsed commands rather than understanding requests. Large language models change that. Speaking to software now works for open-ended tasks such as drafting, planning or describing what a camera sees. The remaining barriers are social more than technical: many people still feel awkward talking to devices in public.

Spatial computing is waiting for its spreadsheet

Headsets are impressive and genuinely useful for some work, such as large virtual displays, design review and training simulations. But the category has not yet found the task that makes it indispensable for most people, the way the spreadsheet did for the personal computer. Weight, battery life and price are improving each generation; we expect lightweight glasses rather than full headsets to carry the next wave.

Subtle inputs are the sleeper

Electromyography wristbands, which read the electrical signals of muscle movement, may matter more than any display. A barely perceptible finger twitch as a click would make computing quieter and more accessible. Brain-computer interfaces, such as those trialled by Neuralink and Synchron, remain medical research for people with paralysis, and should be discussed as such.