What's unique to AI is that CEOs now have a robot that supports that disconnect. Our CEO recently announced that he has now started doing frontend programming, by which he meant that he had told ChatGPT to output some HTML. No doubt it also told him how smart and clever his ideas were and what a great engineer he was.
This kind of thing only increases the disconnect between what CEOs think employees do and what they actually do.
Partially yes, but I think there's a substantial difference: LLMs are like always-on Yes Men that also roleplay as very competent engineers. So if your assistant Joe Sycophant always tells you how great you are, you might become inured to it; deep down you know Joe is just a shmuck. Also, Joe needs to sleep and you might drive him into burning out.
LLMs, however, roleplay as alway available, very capable, very detail-oriented engineers who also happen to be sycophants. And, unlike Joe above, they have no self-preservation and their jobs are not on the line if they royally fuck up. Much more dangerous, in my opinion.
but a c-suite picking up some tool, making a toy example, then declaring “well doing X is super easy, roll it out” (or “change the kpi”, etc.) is something i have seen dozens of times.
HTML is not relevant here. I think the problem is that LLMs are qualitatively different.
"No code" tools are usually just tools. They have the pitfalls you describe, but they aren't ego stroking machines like LLMs.
LLMs not only share the same pitfalls, they also encourage you to make the dumbest things. They will make this CEO believe they are the smartest engineer in the world, "you're building exactly the right product", "you're asking precisely the right questions", etc. Ego stroking when leading you to the abyss is very dangerous.
LLMs roleplay as smart human engineers who constantly tell you you're the smartest being in the multiverse.
This kind of thing only increases the disconnect between what CEOs think employees do and what they actually do.