“When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things.
-
RE: https://infosec.exchange/@darkuncle/116778528340746246
“When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things. How many draft texts to colleagues or family members have we all stared at in frustration, wondering why they don’t feel quite right—until we finally realize that they need to be rethought completely, or not sent at all? When a book I was writing became an almost hopeless grind, I tore up 90 percent of the manuscript; it became a far more honest work for having been halted at a conceptual dead end, forcing me to turn back.
AI can’t make that kind of judgment.”
Holds equally for code.
@aral I've started writing some solarpunk-inspired stories and I can't imagine using an AI at all. The whole writing process is frustrating, but rewarding
-
RE: https://infosec.exchange/@darkuncle/116778528340746246
“When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things. How many draft texts to colleagues or family members have we all stared at in frustration, wondering why they don’t feel quite right—until we finally realize that they need to be rethought completely, or not sent at all? When a book I was writing became an almost hopeless grind, I tore up 90 percent of the manuscript; it became a far more honest work for having been halted at a conceptual dead end, forcing me to turn back.
AI can’t make that kind of judgment.”
Holds equally for code.
@aral
Not sure I agree re code.Sure, AI won't make the same judgement as us, whatever it is. But, AI can be directed to produce what we want to get out of it.
So, don't leave it without direction... and then, it's up to each to decide what that means for them, and for their users.
-
RE: https://infosec.exchange/@darkuncle/116778528340746246
“When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things. How many draft texts to colleagues or family members have we all stared at in frustration, wondering why they don’t feel quite right—until we finally realize that they need to be rethought completely, or not sent at all? When a book I was writing became an almost hopeless grind, I tore up 90 percent of the manuscript; it became a far more honest work for having been halted at a conceptual dead end, forcing me to turn back.
AI can’t make that kind of judgment.”
Holds equally for code.
@aral @darkuncle
« Ce que l’on conçoit bien s’énonce clairement, Et les mots pour le dire arrivent aisément ». - Nicolas Boileau, 1674 -
RE: https://infosec.exchange/@darkuncle/116778528340746246
“When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things. How many draft texts to colleagues or family members have we all stared at in frustration, wondering why they don’t feel quite right—until we finally realize that they need to be rethought completely, or not sent at all? When a book I was writing became an almost hopeless grind, I tore up 90 percent of the manuscript; it became a far more honest work for having been halted at a conceptual dead end, forcing me to turn back.
AI can’t make that kind of judgment.”
Holds equally for code.
My longer perspective might be useful here - having completed my schooling and early university studies before personal computers. Imagine having to write/type a thesis without being able to copy and paste, move text around, search, etc... It was a very different process, especially with regard to the extent that you thought about issues, and what to say, before producing any text, and in marshalling different perspectives and sources that to a large extent you held in your head, or brief note form, rather than linked documents that magically reappear on demand.
An elderly lawyer friend of mine laments the ever-increasing length and complexity of legal documents, which he believes also goes back to the copy and paste culture - you see a clause somewhere that might conceivably come into play in some remote eventuality - why not include it ?
Maybe AI is not different in kind - just another aspect of our ongoing double-edged interaction with technology ?
-
RE: https://infosec.exchange/@darkuncle/116778528340746246
“When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things. How many draft texts to colleagues or family members have we all stared at in frustration, wondering why they don’t feel quite right—until we finally realize that they need to be rethought completely, or not sent at all? When a book I was writing became an almost hopeless grind, I tore up 90 percent of the manuscript; it became a far more honest work for having been halted at a conceptual dead end, forcing me to turn back.
AI can’t make that kind of judgment.”
Holds equally for code.
@aral As much as I hate AI being shoved into everything, and despise when it's used to replace human creativity, I understand that to fight something, you have to know what you're up against. And I find the technology itself quite interesting, even if everything else surrounding it has become rather awful.
So, I've been experimenting. Since Google insists on shoving their new search AI in my face at every opportunity, I figured that I might as well bounce some ideas off it; see if it can teach me how to do something, rather than do the work for me. And maybe every token used will hasten the AI bubble bursting by a few milliseconds.
What I found is that AI is a bit like a puppy with ADHD. When presented with any problem that isn't particularly straightforward, it will happily dive down a rabbit hole of increasingly esoteric "solutions", each one increasing in complexity and failing to solve the issue while making more and more of a mess. If it gets stuck, it'll just start going in circles. What it will NEVER do is suggest going back to an earlier point and trying something else. It will never say “Maybe this…”, everything is always confidently stated: “This is absolutely because…” even when it's not.
I don't know if this is endemic to LLMs, or if this is just a result of its system prompt and/or model and/or training. I wonder if this is an artefact of capitalism more so than an artefact of the technology.
And I kinda hate that I was convinced to try it, because it means what they're doing is working. Regular search is now such a poor experience that you're made to rely on AI to collate the results into something usable for you… if you trust that it'll always do so correctly. Which I don't.
-
RE: https://infosec.exchange/@darkuncle/116778528340746246
“When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things. How many draft texts to colleagues or family members have we all stared at in frustration, wondering why they don’t feel quite right—until we finally realize that they need to be rethought completely, or not sent at all? When a book I was writing became an almost hopeless grind, I tore up 90 percent of the manuscript; it became a far more honest work for having been halted at a conceptual dead end, forcing me to turn back.
AI can’t make that kind of judgment.”
Holds equally for code.
@aral One of the things I told junior developers at my last job was how many lines of code I ended up throwing away. Because sometimes you realize that you’re writing too much code, or it’s too annoying or tricky to write, and that’s a signal. It’s telling you that you should stop, step back, and rethink the problem, likely from the start.
LLMs can’t do that, because they have no actual understanding and no introspection. They’ll happily churn out huge amounts of ugly code that solves the problem the wrong way. And the people who love AI will say “Look! See how much boilerplate it wrote! See how much time it saved me!”
-
@aral One of the things I told junior developers at my last job was how many lines of code I ended up throwing away. Because sometimes you realize that you’re writing too much code, or it’s too annoying or tricky to write, and that’s a signal. It’s telling you that you should stop, step back, and rethink the problem, likely from the start.
LLMs can’t do that, because they have no actual understanding and no introspection. They’ll happily churn out huge amounts of ugly code that solves the problem the wrong way. And the people who love AI will say “Look! See how much boilerplate it wrote! See how much time it saved me!”
-
-
RE: https://infosec.exchange/@darkuncle/116778528340746246
“When writing is hard, it’s often not just because we are tired, underfed, or inefficient but because our mind is trying to tell us crucial things. How many draft texts to colleagues or family members have we all stared at in frustration, wondering why they don’t feel quite right—until we finally realize that they need to be rethought completely, or not sent at all? When a book I was writing became an almost hopeless grind, I tore up 90 percent of the manuscript; it became a far more honest work for having been halted at a conceptual dead end, forcing me to turn back.
AI can’t make that kind of judgment.”
Holds equally for code.
-
@davidgerard @aral notoriously great at dealing with numbers
-
S sigmundur@radikal.social shared this topic
