Given we have court stenographers who need to use a very specific keyboard, months of training and can STILL be overwhelmed by too rapid of a speech pattern - our world?
The hurdle there is comprehending multiple speakers with different backgrounds fast enough. Typing that fast isn’t unrealistic if it’s your own thoughts you’re putting out.
Anything structured and thought out, sure. But vibecoding, tokenmaxxing or just waffling out some unfinished thought is faster and more comfortable when spoken. See the rise of voice notes over texts
Still uncomvinced even if you’re vomiting just random vibe code thoughts you’d still be slower via voice than via text. I dont see how this device would ever be successful for this niche.
Anyway, not condoning or supporting any of this, but those muzzles would be nice on the train for those people recording their voice notes (and are listening to them on speaker)
You might be underestimating how fast people working in tech type. Actual issue studied today in comprehension is the fact that keyboard typing (or even phone) is faster than our thought thus hindering comprehension and memorization. This is primary factor that is making pen and paper trend in tech and education circles right now.
Honestly speech as input is still extremely niche because it’s slow and even with massive accuracy improvements still inaccurate.
2 reasons : first there are multiple speakers, second speech contains a lot of words that could be optimized away in a prompt but matter for courts. My voice prompts are so dirty and unclear compared to well thought, text ones
Vibe coding means he’s just whispering ‘claude, generate a function to take input from grok api and pipeline it to user input’ instead of typing. Or something. Idk, I don’t code.
But how is whispering the prompt better than typing the prompt?
Cause LLM handles the part of actually ‘writing’ the code with all the nitty-gritties. You just prompt it what you want in a couple of lines and chill waiting for the output. 🤷♂️
In what world it’s faster to talk than to type?
Given we have court stenographers who need to use a very specific keyboard, months of training and can STILL be overwhelmed by too rapid of a speech pattern - our world?
Try to outrun a court stenographer using a text-to-speech program.
A trained court stenographer can 225 to 360 words per minute. This guy, optimistically, claims he can do 150.
The hurdle there is comprehending multiple speakers with different backgrounds fast enough. Typing that fast isn’t unrealistic if it’s your own thoughts you’re putting out.
Nah thats nothing alike. Have you tried dictating things as notes or instructions? It’s impossible to beat even average typing speed.
Anything structured and thought out, sure. But vibecoding, tokenmaxxing or just waffling out some unfinished thought is faster and more comfortable when spoken. See the rise of voice notes over texts
Still uncomvinced even if you’re vomiting just random vibe code thoughts you’d still be slower via voice than via text. I dont see how this device would ever be successful for this niche.
You must be typing faster than me.
Anyway, not condoning or supporting any of this, but those muzzles would be nice on the train for those people recording their voice notes (and are listening to them on speaker)
You might be underestimating how fast people working in tech type. Actual issue studied today in comprehension is the fact that keyboard typing (or even phone) is faster than our thought thus hindering comprehension and memorization. This is primary factor that is making pen and paper trend in tech and education circles right now.
Honestly speech as input is still extremely niche because it’s slow and even with massive accuracy improvements still inaccurate.
2 reasons : first there are multiple speakers, second speech contains a lot of words that could be optimized away in a prompt but matter for courts. My voice prompts are so dirty and unclear compared to well thought, text ones
I thought we were talking about vibe coding
I bet this guy henpecks
Vibe coding means he’s just whispering ‘claude, generate a function to take input from grok api and pipeline it to user input’ instead of typing. Or something. Idk, I don’t code.
But how is whispering the prompt better than typing the prompt?
Unless all their employees are functionally toddlers in regard to typing
Cause LLM handles the part of actually ‘writing’ the code with all the nitty-gritties. You just prompt it what you want in a couple of lines and chill waiting for the output. 🤷♂️
There are horror stories about it now.
https://uxdesign.cc/its-all-fun-and-games-until-your-boss-starts-vibe-coding-106283e98347
I wasn’t talking about writing code. Take the verbal prompt they’re whispering and type that verbatim. Faster than whispering.