Nothing co-founder Carl Pei has not typed a message in months, according to a post made on X. Instead, he has been using Essential Voice, a dictation-style, AI-powered feature coming to Nothing phones.
Akis Evangelidis, Pei’s fellow Nothing co-founder, also chimed in on the feature. He called typing “slow and inefficient,” and said we’re moving towards a “voice-first way of interacting with technology.”
I, along with most people with ears, am here to implore you not to go voice-first, and please continue typing most of your messages, regardless of how accurately AI promises to transcribe your musings, and take out the ums, ahhs, and pauses at the same time.
I didn’t expect this small underrated phone feature to save me so much time but it does
I ignored the little icon on my keyboard for years
Testing Essential Voice
It’s very effective
To be clear, it’s how Nothing, and many other companies, are pushing a “voice-first” future that bothers me, but what about the Essential Voice feature?
I tried it out on the Nothing Phone (3), and it’s very impressive. It’s activated when you press a small icon below the keyboard, or by long-pressing the Essential Key on the side of the phone in an app where the keyboard appears.
You talk, it transcribes, and it does indeed strip your words of incidental filler, including “you know” and “so,” but is clever enough to recognize when they’re part of the sentence.
It’s not put off by rambling, and you don’t need to press anything when you’re done. It seems to know instinctively.
It also automatically pastes the text into the app you’re using, so there’s very little friction, which encourages you to use it.
I love that this is a free feature in the operating system, and Nothing’s approach to on-device AI in general.
There’s no question it’s a polished, useful feature. However, I don’t think it, or other tools like it, should become the standard.
The obvious problem
Like leaving your front door open
Let’s talk about the obvious problems with a “voice-first” approach to messaging. The privacy and security issues that come with bleating out every message for your smartphone to transcribe should not be underestimated.
Voice notes are great for short responses or longer messages containing private or personal information when you’re in the car or in a situation where typing isn’t possible.
But by adopting a voice-first approach to messaging, your business suddenly becomes everyone else’s business. There’s a real danger of getting too comfortable talking out loud into your phone about absolutely everything.
We’re all conditioned to take care online. Ads for VPNs flood YouTube, telling us connecting to public Wi-Fi is like giving criminals a free pass to our private data. Samsung made keeping information on your phone screen private the Samsung Galaxy S26 Ultra’s entire identity.
Talking out loud to your phone in public is just as problematic as in these situations, but typing is not.
Voices are wonderful
Until they’re not
Voice-second, rather than voice-first, should be the standard if you’re interested in keeping private things private. However, what if everyone suddenly went all-in on voice-first for their technology, as Evangelidis suggests?
Think about all the people you see on a daily basis typing on their phones, now speaking out loud to them instead, and the cacophony of voices we’d hear, everywhere we go.
Constantly listening to random people’s messages about remembering to buy cat food, how their boss doesn’t value them, or that their significant other forgot an important date sounds like hell.
Don’t forget, we’ll probably also hear every reply, because the other person will be talking into their phone too.
I’d love to think headphones will be used for many of them, but judging by the state of public transport some days, I’m not convinced.
It may not be only messages we all have to hear, but presumably control of phones, too.
Do you want to hear everyone talking to AI, adjusting settings on their phones, turning on Do Not Disturb, opening an app, or even playing a game? I don’t.
These things are what the keyboard is for when you’re out in public.
However, despite my fears, I don’t even think it’ll happen. Talking to any electronic device in public is pretty embarrassing, and not everyone is so oblivious that they’ll happily use a voice note for every single message they send.
Wait, there’s a problem
Keyboards are pretty bad
I don’t want to hear everyone talking to their phones, smartglasses, smartwatches, or laptops.
At this point, I should be saying typing is a wonderful skill to have and hone. Stop, and you’re at risk of forgetting. Just like my handwriting is now awful, because I type everything and write nothing but my signature.
The problem is, phone keyboards aren’t very good. I type on Android and iOS phones every single day, switching between tapping letters and swipe typing, and it’s rarely perfectly accurate.
It’s not always my fault, either. The accuracy of swipe typing, which is my preferred option, on iOS seems to have become worse, and it constantly gets the same word wrong, over and over again.
I’m using the Honor 600 at the moment, and the default keyboard is Microsoft SwiftKey. I’ve been giving it a try, just in case it has improved over the years, but it hasn’t. It’s slow, and because it doesn’t “know” me, it’s annoyingly inaccurate.
Google’s Gboard does know me, and for the most part, it’s the best mobile keyboard option.
However, when typing on a smartphone isn’t always very good, it’s hard to argue against finding an alternative.
Voice is always faster than typing, and AI is constantly getting better at transcription, but are we looking for perfection in our messages?
The death of identity
I quite like the ums and ahhs
When Nothing’s Essential Voice transcribes your voice, it will also remove what Nothing calls “filler words,” for “streamlined and considered output.”
If we’re being encouraged to use our voice, why does Essential Voice want to take away some of the personality?
Digital services, automated phone lines, and robots should always have a considered output, but we humans shouldn’t.
Conversational flair, vocal tics, and “filler words” aren’t always negative. They can be amusing, add emphasis, or be a personal signature. They give sentences character, can show sarcasm, and can be used as punctuation.
Removing all this is fine when you’re dictating an important document or something for work, but voice-first suggests it’s for everything, and I think that’s a problem.
Say no to Essential Voice?
No, just don’t go voice-first
Essential Voice has some interesting and rarely seen extra features. It’s multilingual, auto-detects up to 100 languages, and automatically translates into a selected language.
A shortcut feature lets you assign addresses to keywords, which helps with privacy. Plus, everything is generated on the device, not in the cloud, and the feature has to be manually activated and isn’t always listening.
It’s part of Nothing OS 4.1 and will be available on the Nothing Phone 3 first, then on the Nothing Phone 4a Pro, and the Nothing Phone 4a later.
I know it may sound like I’m against Essential Voice, but I’m really not. I’m more against the concept of a voice-first approach to technology.
From Dragon NaturallySpeaking on Windows 95 to Essential Voice today, it has never really taken off, and for good reason. You simply don’t need to say everything out loud.
