The Question

Person dictating a message to a smartphone while a holographic interface displays the transcribed text

Look down at your keyboard. The letters are arranged in an order called QWERTY, developed in the 1870s by the newspaper editor and inventor Christopher Latham Sholes for early typewriters. Historians still argue about exactly why he chose it; the popular story that it was designed to slow typists down and stop the mechanical arms jamming is disputed. Either way, a layout built for a nineteenth-century machine outlived the typewriter, colonised the computer, and then shrank onto a glass screen where your thumbs now peck at keys designed for ten fingers.

We kept it because everyone already knew it. But habits built on old constraints do not last forever. A 2016 study by Stanford, the University of Washington, and Baidu found that speaking to a phone produced text about three times faster than typing with thumbs. Speech recognition has improved enormously since. And a new generation of gesture devices promises typing with no keyboard at all. The question is not whether the keyboard can be beaten on speed. It already has been. The question is whether people will stop reaching for it.

What the Evidence Shows

Start with speed. In the 2016 study, 32 participants entered text either by typing on an iPhone keyboard or by speaking to Baidu's Deep Speech 2 software. Speech was about three times faster in English, with a 20% lower error rate, and 2.8 times faster in Mandarin, with a 63% lower error rate. That result came from 2016-era technology; in 2016 and 2017, Microsoft researchers reported that their systems matched professional human transcribers on a standard benchmark of conversational speech. Today's recognition, built on the same deep-learning methods behind chatbots, is better still.

Behaviour is shifting, but slowly. WhatsApp users send around 7 billion voice messages a day, according to Meta. Among many younger users, voice notes are a normal way to communicate, and phone keyboards put the microphone button front and centre. But voice messages are sent as audio, not converted into text, and for most written work — emails, documents, search, messaging in public — people still type.

"The surprise for me was that it was that much better: three times faster! You would think everyone would be flocking to use it if they knew how much better it actually was."

— James Landay, Stanford University, to NPR, August 2016

Then there is gesture. Apple Vision Pro users can type in mid-air by pinching their fingers while looking at letters. The more radical device is Meta's Neural Band, developed from its CTRL-labs acquisition, which reads the electrical signals your brain sends down the motor nerves to your hand. In a 2025 Nature paper, Meta's researchers reported a generic wristband model that let new users write letters with their finger at a median of 20.9 words per minute — compared with roughly 36 for typing on a phone keyboard. In August 2026, Meta began rolling out this "neural handwriting" to early-access users of its Ray-Ban Display glasses.

Your keyboard layout was built for a nineteenth-century typewriter. You still use it for every message you send.

Why This Is Happening

Speech recognition finally works. For decades, dictation lost to typing because fixing recognition errors cost more time than typing saved. That equation has largely flipped. Modern systems understand context, punctuate automatically, and let you edit by voice. Humans speak far faster than they thumb-type, so once the error-correction tax disappears, voice's speed advantage is large.

A generation is growing up talking to machines. Children who learned to talk to smart speakers before they could spell treat voice as a natural way to command a device, and voice notes are a habit for many young people. But that habit is mostly about sending audio, not turning speech into written text.

The screens are shrinking and the keyboards can't follow. Smart glasses, earbuds, and watches are becoming the fastest-growing computing devices, and none of them has room for a keyboard. If you want to answer a message from your glasses, you will speak it or gesture it. Every device sold without keys is a vote for the post-keyboard world — and the hardware giants know it, which is why Meta is betting on nerve-reading wristbands and Apple on eye-and-pinch input.


What Could Happen

Voice and gesture become the majority input by 2032 Possible · 20%

Casual text — messages, searches, notes, emails — flips to voice-first, with AI cleaning up the words as you speak, while gesture wristbands handle quiet places. By 2032, most text created by ordinary users never touches a key.

Voice plateaus at the privacy wall and gesture arrives too slowly Most likely · 60%

Voice dominates cars, homes, and AI assistants, but stalls in offices and public spaces, where nobody wants colleagues or strangers hearing their drafts. Neural wristbands remain slower than keyboards and spread slowly. Typing remains the main way most people write, with voice a growing but minority partner.

The keyboard proves stubbornly permanent Possible · 20%

Editing, precision, and privacy keep typing dominant for anything longer than a text message. Voice remains a dictation accessory, gesture remains a gaming gimmick, and QWERTY celebrates its 175th birthday in 2048 still on every desk — the cockroach of interface design, surviving every extinction event.

Our Assessment
We assign 20% probability — unlikely that by 2032 the majority of text created by ordinary users will be spoken, gestured, or dictated rather than typed. We have lowered this forecast from 64%. Our earlier version included an MIT Technology Review quote we could not verify and repeated a disputed story about QWERTY. Speech has been about three times faster than thumb-typing for a decade, yet people have not switched — because of privacy, open offices, editing, and habit. Meta's nerve-reading wristband is real, but its handwriting is still slower than a phone keyboard. Voice will keep growing, especially with AI assistants; the keyboard is likely to remain the main way we write.

What Can We Do

Wrist wearable detecting subtle finger movements as a person types on an empty desk surface

You do not have to wait for 2032. The transition is available now, and the people who adapt early will feel the benefit in hours reclaimed every week.

Try dictation for one week on messages and emails. Your phone's microphone key already uses the new generation of recognition. Most people who commit to seven days report they never fully go back — the speed difference is that stark. Start with low-stakes texts and let the habit climb.

Learn to edit by voice, not just compose. The skill that separates frustrated dictators from fluent ones is speaking commands: new paragraph, delete that, replace X with Y. Ten minutes of practice removes most of the friction that made dictation feel clumsy.

If your work is keyboard-heavy, watch the wristband space. Meta's neural band is expected to ship alongside its smart glasses. If it works as demonstrated, silent gesture typing solves the open-office problem that voice cannot — and early fluency will be a genuine workplace advantage.

Teach children both. Typing will remain a required skill for at least another decade, especially in exams and professional software. But a child fluent at composing, structuring, and editing text by voice is learning the native interface of the 2030s. Treat it like a language: earlier is easier.