LLMs Will Never Work, You Need to Type What You Know
The LLM delusion isn't the first of its kind. Voice recognition is too imprecise to be practical or really save any time if you can type fast.
Musical instruments are sometimes keyboards*. And using keyboards can be like playing a piano provided you've had sufficient practice. Then it's the most pragmatic input method. We have 10 fingers and our brain coordinates those rather well. Of course keyboards had centuries to evolve. With some exceptions, they adapted to human capacities and helped enhance those capacities.
A lot of gullible people still wrongly assume that getting LLMs to spew out words is the same as a person writing those words "properly"; when GPT-5 turned out to be a total disaster people quit assuming that LLMs will magically improve over time; throwing more data and energy at the task can yield even worse LLMs. Economically speaking, that might end up costing an order of magnitude (x10) more to train and query while actually worsening things, exacerbating the rate of falsehoods (euphemised as "hallucinations" as if LLMs can think, understand, know and so on).
Managers (in the traditional sense, commanding people) that I've experienced in my career were never fast typists. Never. They used their mouths to spite (or mask) their ineptitude. "Faster, guys, come on, you can do it faster!"
Project managers or team leaders are another matter. Many of them don't even act or look (or behave) like managers. Some are like friends. They tend to have real skills, unlike "the suits" (like the cocaine users at the EPO).
Managers with suits and ties wrongly assume that if they gave LLMs to coders and let workers type E-mails with slop, then they'll become more productive. Sorry, sir, ain't gonna work! The overhead/toll isn't in the typing but the thinking (e.g. testing code, articulating relevant facts, considering context). Focus on quantity or speed is known as "smearing" when done with a pen or pencil (just filling up pages irrespective of importance - a practice that is disrespectful to the reader or coder who inherits the whole "shebang").
Cory Doctorow, in one of his many extensive writings about slop, explained that those "bosses" are basically "horny" (his word) for this kind of thing, even if it's merely a dumb fantasy, based upon misguided assumptions (false marketing). That's why the media claims "AI adoption"; never mind if those ambitions get abandoned some time later (by then this media already vanishes, looking for the next "AI adoption" example/s - it gets paid for hype, not projecting the hard reality).
Slop isn't going to replace human intelligence; similarly, musicians were never replaced by "CG" - or computer-generated - so-called 'music', predating the slop frenzy by several decades (the same goes for synthetic images). There's nothing new here except a hype pandemic, in which media - what's left of it anyway - seems complicit**. █
_________
* I've long been fascinated by accordions. They're amazing instruments. As a kid I learned to play the piano (my parents enrolled my sister and I for classes). I then fell in love with computer keyboards, instead. My sister carried on with piano lessons and nearly mastered the art by age 15.
** See for example: LLM Slop Could Not Rise to Prominence Without Media Complicity and Artificial Hype