Shifting from typing to speaking
Typing gives its own high, its own flow. This is how we were connected to the machine so far - via the fingers. Until three months ago, my brain’s thoughts were channeled so they could sequentially flow through my fingertips and give the computer the input I wanted, and my eyes and ears received the output. The output side is still the same. The input has drastically changed.
My fingertips stay with me - they are no longer on the desk, and my wrists are no longer sore after hours of holding the mouse in a weird position. The newest part of my setup is a headphone and a microphone that stays near my neck. It is the latest input device, and I speak into it confidently for two reasons.
One, speech to text has improved radically. Even the free options are very, very accurate. Two, even for day-to-day work that needs special characters - brackets, underscores, dashes, numbers, camel casing - I do not need to press combinations of buttons anymore. I just need to know what I want to get done, and I can talk to an AI in plain English to get it done. As long as I can speak in a manner that makes sense, I can get what I need.
It is a weird transition, because now I have to make sure I am speaking way more coherently. It is similar to how my touch typing was: mostly there, with imperfections. My pinky would go to the wrong letters; it was not touch typing per se, but it was close enough and it got my work done for all the years I have worked as a developer. Voice is in a very similar state. My statements are not always coherent or perfect, because they will need a round of review later on. There is no backspace like on a keyboard. Most of the time I cannot even see what the output looks like, because of how fast I speak. It is an unfiltered, ungated, uncensored path from my brain to my mouth.
Earlier, I would think fast and then structure the thoughts so they flowed through my fingertips into the computer. Now I have to think even faster, because I can speak way faster than I can type. And if I want the input to be coherent, structured, useful, information-dense, non-repetitive - because tokens cost money - then my thoughts themselves have to be more coherent.
So the physical process has changed, but the input quality has not decreased - I would argue both input and output have really increased. I can be redundant and imprecise with my words and sentences, and the AI still gets it. The voice technology is good enough that the mistakes are minimal. And if I pay for Whisperflow, I get the best voice transcription we have had in history - one that makes no noticeable mistakes.
The technology is already there, and it keeps you in a flow state. I am dictating this post - like every post on this blog - and I am really in the flow: thinking things through, not getting distracted, not having to focus hard. It is way easier as an input mechanism. The shift has been totally positive.
Rather than me saying it, a photograph of my work desk will do better justice.
TODO: add a photograph of my work desk - at both home and office, the headphones now hold the prominent position in the center of the table
On both my desks the pattern is the same: a headphone as the primary tool. In fact, I switched to two headphones - one just for the microphone and one for output. My Sony XM5 gives bad audio output when I also use it as an input device, but as an output-only device the sound is great. So you will see me wearing a Sony wireless headphone and, around my neck, a wired earphone with its microphone very close to my mouth.
TODO: add a photo of me wearing the two headphones
I cannot wait to try thought-based typing and thought-based communication with computers. Pretty sure we will get there soon too.