.svg)
No more carefully typed prompts. Since I switched to voice dictation mode, the way I use AI has radically changed. Faster, smoother, more natural. Feedback on this method that is transforming our relationship with ChatGPT and that may make you whisper on your screen in the open space.
The discovery: speaking as you think
Not conversational vocal-to-vocal
Let's be clear right away: I'm not talking about the voice chat mode with ChatGPT (voice question → voice answer). Personally, I am really not a fan of this format.
I am talking about the Dictatation/transcription mode : I dictate my prompt vocally, ChatGPT transcribes it into text, then respond in writing. This nuance changes everything.
The advantage: zero cognitive friction
The main interest? Speak as I think, without a structural filter. No need for:
- Building perfect sentences
- Structuring before writing
- Think “like a robot”
- Correct typos
- Losing the thread looking for words
I babble, I hesitate, I pick myself up... and ChatGPT understands perfectly.
It was my partner Yoni who pushed me to join it. He has been swearing by voice for a long time and was right: once tested, there's no going back.
What has changed in concrete terms:
- Considerable time savings on long prompts
- ChatGPT better understands my real intentions
- Fluidity in the expression of complex ideas
- Spontaneity that unleashes creativity
The concrete benefits of vocal
Speed: save a lot of time
The speed of speech (120-150 words/minute) greatly exceeds typing on the keyboard (40-60 words/minute for a good typist). The gain is mechanical.
But the real gain is not there. It's the reduction of mental friction : formulate your thoughts directly without going through the “written form” stage.
Understanding: ChatGPT captures context better
Paradoxically, my imperfect vocal prompts often work better than my slick written prompts.
Why? Because by speaking spontaneously, I deliver:
- No more implicit context
- Shades of intention
- Details that come naturally to me orally
- A more authentic thought structure
ChatGPT excels at extracting the very intent of a “poorly crafted” prompt.
Creativity: unleashing spontaneous thought
Writing imposes a certain rigor. Orally, it is easier to:
- Explore ideas out loud
- Testing formulations
- Bouncing on your own words
- Let intuition guide
This spontaneity often generates unexpected creative paths.
The limits you need to know
A necessary adjustment time
The transition is not instantaneous. The first few days, I had a tendency to freewheel, to babble without structure.
The key tip: Take 10 seconds to mentally formulate your idea before speaking. This micro-pause makes all the difference between an effective vocal prompt and a confusing monologue.
Not all prompts are vocal
I am not saying that voice replaces everything. Some use cases remain more suitable for the keyboard:
Stay on the keyboard to:
- Very structured technical prompts (JSON, code, formulas)
- Precise instructions that require copy and paste
- Fast iterations on an existing prompt
- Noisy or shared environments
Switch to voice to:
- Brainstorming and exploring ideas
- Writing long content
- Complex requests with lots of context
- Creative tasks that require spontaneity
The eyes of others in open space
Yes, you'll probably sound a bit weird whispering on your screen. That is the price to pay.
But as Nicolas Guyon points out in its Comptoir IA newsletter : you get used to it quickly. Whisper his prompts to Wispr Flow to him “like Tony Stark talks to JARVIS”. And frankly, that's exactly it.
“The future of work won't be typing faster”
Nicolas Guyon says it better than me: “The future of work won't be typing faster. It's going to be thinking out loud with AI.”
This sentence perfectly summarizes the paradigm shift that is taking place.
From keyboard to voice: a cultural change
For decades, professional competence has been measured in part by the speed of typing, by the mastery of the keyboard, by the shortcuts that are remembered.
Conversational AI turns this logic on its head: The value is no longer in technical execution but in the clarity of thought.
Knowing how to formulate an idea well orally becomes more valuable than knowing how to type well.
Tools that transform use
Beyond native ChatGPT, tools are emerging to optimize the voice experience:
Wispr Flow (quoted by Nicolas Guyon): improved AI dictation that transcribes and cleans automatically.
ChatGPT native functionality : built-in dictation mode available on mobile and desktop.
Browser extensions : allow dictation in any text field.
These tools reduce friction and gradually standardize voice use.
Psychological resistances to overcome
“I don't want to talk to my screen”
It is the most common resistance. It is cultural: we have integrated that talking alone = weird.
But remember: 20 years ago, wearing headphones on the street to make phone calls seemed ridiculous. Today, it is banal.
“My mind is not clear enough orally”
It is precisely the opposite. You will probably discover that your oral thinking is richer and more nuanced than your written thinking.
Writing forces us to simplify. Oral speech frees up the natural complexity of thinking.
“ChatGPT won't understand my hesitation”
Tested and validated: ChatGPT handles stuttering, repetitions, and on-the-fly corrections remarkably well. It extracts intent beyond form.
It is often even in these “imperfections” that the richest context lies.
Conclusion: definitely test
I am not saying that voice will completely replace the keyboard. Each method has its optimal use cases.
But if you want to gain speed, spontaneity and creativity in your use of ChatGPT, the dictation mode is definitely worth testing.
The concrete benefits:
- Mechanical time savings on long prompts
- Better understanding of your intentions through AI
- Freedom from creative spontaneity
- Reducing cognitive friction
As with any new practice, you have to overcome the initial discomfort. But once you get the hang of it, it's hard to go back.
In fact, the future of work won't be typing faster. It will be to think more clearly, and to do it out loud with AI as the preferred interlocutor.
And you, where are you with the vocal? Already tested or still skeptical?
.webp)
