I wish there was a speech/language technology that could:

1. Interface with my brain, somehow.
2. Instantaneously (like milliseconds) create a rough model of the ideas I want to express verbally at any given moment.
3. Find words, phrases, and social scripts that are as close to a 1:1 correspondence with the ideas I want to express verbally as our language permits, and present these to my brain for me to choose on my own volition.

Could we hypothetically leverage LLM technology to do this? I feel like this would be super practical for me specifically because of my speech issues.

I feel like it would also enhance my language skills, because:

1. More information/options would be presented to my brain that would enable me to use my best human judgment in choosing appropriate words, phrases, and social scripts for a given social situation.
2. After some amount of repetition of these patterns, I would eventually learn better ways to express ideas for a given social situation, and gradually become less reliant on the technology.

Though, there would still be times when I just don't have the cognitive spoons to articulate things myself, and in these instances I wouldn't mind the technology doing it for me, and all I would have to do is the physical act of vocalizing it.