AI voice prompting improves how we communicate
With other humans AND ourselves
In a survey I did recently on LinkedIn, 65% of pollsters said they’re not using the voice features in their AI tools. (Yes, it was a small poll. 11 people. But it got me thinking.) So I want to share what I think is going on and I want to know what you think about this.
And just for fun, let’s do the poll I did on LinkedIn here also.
If you like experiments like this:
And if you know someone who does
Possible reasons why voice prompting isn’t as popular as text prompting?
1. The options genuinely aren’t good enough yet
ChatGPT has the best voice functionality available right now. You can watch me use it here: → How I use ChatGPT’s voice feature
But a lot of people are walking away from ChatGPT at the moment, for reasons I understand. Claude’s voice interface feels clunky with multiple clicks, unnatural pacing, very “early software.” I haven’t found another large language model with voice functionality worth recommending. Have you? Drop it in the comments.
2. Accent bias is a real barrier and an underreported one
I taught language for many years and have a flat Western North American accent, which means I’ve essentially trained my speaking quirks out of my voice. I can revert to near-newscaster pronunciation when I need to be clearly understood. That probably skews my sense of how usable these tools are for most people.
Over years of podcasting with international guests, I’ve seen how badly AI transcription handles vocabulary, rhythm, and pronunciation outside a narrow “neutral” accent band. I’d expect the same problems to show up in LLM voice functions and maybe worse, since the stakes of being misunderstood are higher in a conversation than in a transcript.
Have you hit walls because of this? Are you actively working on making it better? I’d genuinely like to know.
3. Privacy has two separate problems
Most people lump voice privacy into one concern, but there are actually two distinct issues:
Data privacy — does your spoken conversation get used to train the model? Many people don’t know the answer to this, and the default settings aren’t always in your favor.
Ambient privacy — do you want the people around you hearing your entire conversation with an AI? This one is underrated. Talking out loud to an AI in an open office, a coffee shop, or even at home changes the dynamic entirely. I wrote about how I’m struggling with this at my current coworking space here: Using ChatGPT as a thinking partner in public spaces
Why voice prompting might make you better at communicating (not just talking)
This might be the most underrated use case for AI voice tools. Not productivity, not convenience, but communication. And that means writing too, not just conversations.
It helps you think through a problem before you bring it to someone else
How many meetings have you sat in where someone is clearly working out their thinking in real time, out loud, at everyone else’s expense? How many emails have you received that were obviously written before the sender knew what they wanted to say? Voice prompting gives you a place to do that first. You can talk through the problem, pressure-test your framing, and figure out what you actually want to say before the conversation or the email that matters.
Talking bypasses your internal editor
Writing is slow enough that your logical brain catches up and starts shaping, softening, and second-guessing what you’re trying to say. Speech is faster than that filter. When you talk, you often get closer to what you actually think. The unpolished version that’s frequently more honest and more useful than the carefully worded one. Voice prompting gives that rawness somewhere to go, and what comes out can become the skeleton of a message, an email, or a difficult conversation.
It externalizes the thinking that’s already happening in your head
A lot of people are highly verbal thinkers. Their inner world is basically a running monologue. But translating that into writing loses something. Voice prompting short-circuits that translation step. You’re not converting thought into text, you’re just thinking out loud, which is probably how that idea wanted to exist in the first place. The transcript becomes your first draft.
Sometimes you need to say it before you know how you feel about it
This one is more emotional than tactical. There are decisions and conversations, and plenty of hard emails, where you genuinely don’t know your own position until you hear yourself articulate it. Voice prompting gives that process a container. A place to be uncertain, to contradict yourself, to figure it out, without the pressure of an audience.
It builds fluency for the communications that make people nervous
Presentations. Difficult feedback. Advocating for yourself in a high-stakes meeting. A message you’ve been putting off sending. For people who freeze up or over-rehearse in their head, talking out loud regularly, even to an AI with no judgment and infinite patience, builds a kind of muscle memory. You get used to the sound of your own voice making an argument. That comfort is transferable, whether you’re walking into a room or staring at a blank page.
What’s actually stopping you?
These are my three guesses, but I suspect I’m missing things. Maybe it feels strange to talk to a machine. Maybe the latency breaks your thinking. Maybe it works fine for you and I’ve got this completely wrong.
Tell me in the comments: what’s your real reason for not using voice or if you are using it, what made it click for you? I’m building a picture of this, and your answer matters.
Really looking forward to your ideas on this.
Steph