varkk.bsky.social
@varkk.bsky.social
Reposted by varkk.bsky.social
Unfortunately this is *absolutely* the wrong lesson.
The real lesson is:
🔹️most work of exec class is substance-less illusions.
If AI is useful in a given use case its because you really only care about form, not substance [as an aside, this why such a great plagiarism machine].
The real lesson is:
🔹️most work of exec class is substance-less illusions.
If AI is useful in a given use case its because you really only care about form, not substance [as an aside, this why such a great plagiarism machine].
November 30, 2025 at 12:06 AM
Unfortunately this is *absolutely* the wrong lesson.
The real lesson is:
🔹️most work of exec class is substance-less illusions.
If AI is useful in a given use case its because you really only care about form, not substance [as an aside, this why such a great plagiarism machine].
The real lesson is:
🔹️most work of exec class is substance-less illusions.
If AI is useful in a given use case its because you really only care about form, not substance [as an aside, this why such a great plagiarism machine].
Reposted by varkk.bsky.social
(2) AI is mostly useless for workers, because they have to produce things with substantive content. People care that they get it right - AI just can't do this consistently.
A useful rule of thumb: genAI gets it 80% right about 70% of the time. If you like those tolerances, it's fab - if not 😬
A useful rule of thumb: genAI gets it 80% right about 70% of the time. If you like those tolerances, it's fab - if not 😬
November 30, 2025 at 12:06 AM
(2) AI is mostly useless for workers, because they have to produce things with substantive content. People care that they get it right - AI just can't do this consistently.
A useful rule of thumb: genAI gets it 80% right about 70% of the time. If you like those tolerances, it's fab - if not 😬
A useful rule of thumb: genAI gets it 80% right about 70% of the time. If you like those tolerances, it's fab - if not 😬
Reposted by varkk.bsky.social
This has three implications.
(1) Most exec will grossly overestimate the broad utility and efficacy of AI because of confirmation bias.
Its good for them, so must be good for everyone. They will only listen to evidence that validates THEIR experiences
(1) Most exec will grossly overestimate the broad utility and efficacy of AI because of confirmation bias.
Its good for them, so must be good for everyone. They will only listen to evidence that validates THEIR experiences
November 30, 2025 at 12:06 AM
This has three implications.
(1) Most exec will grossly overestimate the broad utility and efficacy of AI because of confirmation bias.
Its good for them, so must be good for everyone. They will only listen to evidence that validates THEIR experiences
(1) Most exec will grossly overestimate the broad utility and efficacy of AI because of confirmation bias.
Its good for them, so must be good for everyone. They will only listen to evidence that validates THEIR experiences
Reposted by varkk.bsky.social
So:
🔹️Execs ❤️ AI.
🔹️Workers (who care about getting it right) dont.
🔹️But because of confirmation bias, execs tend to take lesson that it's great for everyone.
🔹️So execs want everyone to use it, and are convinced it will increase efficiency
🔹️Execs ❤️ AI.
🔹️Workers (who care about getting it right) dont.
🔹️But because of confirmation bias, execs tend to take lesson that it's great for everyone.
🔹️So execs want everyone to use it, and are convinced it will increase efficiency
November 30, 2025 at 12:06 AM
So:
🔹️Execs ❤️ AI.
🔹️Workers (who care about getting it right) dont.
🔹️But because of confirmation bias, execs tend to take lesson that it's great for everyone.
🔹️So execs want everyone to use it, and are convinced it will increase efficiency
🔹️Execs ❤️ AI.
🔹️Workers (who care about getting it right) dont.
🔹️But because of confirmation bias, execs tend to take lesson that it's great for everyone.
🔹️So execs want everyone to use it, and are convinced it will increase efficiency