361–370 of 786 posts
I do it, on purpose: it teaches people to do basic research by themselves, and not hit me with questions they could have gotten with a simple chat with their llm. It's a kind of RTFM for current days.
no reviewing 'a.i' code. talking to actual humans not 'a.i' jargons
Earlier quoted context omitted.
Asking the LLM a question _is_ a skill, just like googling. The problem is not asking a LLM, but pasting the 300-chars response without any context.
"I've been asked to confirm that our random number generator satisfies section 14 of the Foo-Bar Accord of 2021. I'm fairly certain it does, and that foorandom.c is a good choice. Assuming I'm broadly correct, write one short paragraph that would confidently assuage anyone of concern without raising any new concerns. Cite the specification and link to it. Afterwards, for my benefit, write a short bullet list with any…
Earlier quoted context omitted.
You cannot walk 100 miles in a day, but anyone with a bit of training can cycle that in an afternoon. It's not a matter of effort or laziness, it's simply not possible. The stitching comment is just pedantic. People don't hand stitch like a machine not because they are lazy, it's because it's ineffective and impractical. It's not impossible to drive screws directly into softwood by hand, but you'd never get anything…
Can you write 10k LOC a day without a coding agent? It's the same issue. You can walk 100 miles. Just not in a day. And you can create 10k LOC without a coding agent. Just not in a day. I don't think that coding agents enable technical possibilities. They just allow things that wouldn't be economically possible without them.
Easily, by writing a code generator. Or disassemble a 1k C program.
> They just allow things that wouldn't be economically possible without them.
It’s not economical to have more code. It’s economical to have correct code and the way to do that is with the power of abstraction. Instead of dealings with opcodes, you have assemblers and macros. Then it’s C, next it’s python, and what would have been a 10k program is now just a few lines.
I agree but to be fair, often people could have asked an AI to get their answer. Especially seniors are prone to this (thinking the AI could not possibly know this or other biases). The responder wrote they asked Claude so as to not take credit for the answer and also to indicate they did not vet it. Then the person asking can decide if they want to read the answer or not (often a 1 sec glean can reveal if it is prom…
Earlier quoted context omitted.
I think a lot of that is people who were always very insecure because they’re mediocre engineers. Previously asking questions or not understanding something was a bit painful but now you can hear “you’re absolutely right!” and make progress all day every day. Until you need to interact with actual humans and that’s why you try to minimise it, hiding behind ai generated content.
It's also very easy to dismiss everyone falling into the "AI trap" as being mediocre in the first place, though. But one just can't know this without having seen their work pre-AI.
Maybe this «spearheading AI person» just sucks at AI related stuff, as clearly that approach is bananas, but they could still be a OK developer.
I do it, on purpose: it teaches people to do basic research by themselves, and not hit me with questions they could have gotten with a simple chat with their llm. It's a kind of RTFM for current days.
That isn't the takeaway. The takeaway is that when someone asks for your input you give them Claude's. Why do we have you?
In my experience this sort of thing happens all the time when something has to cross between different knowledge domains and the sender doesn't apply their mind in order to co-operate on the issue at hand.
In the pre-LLM past, the sender would add no value, acting as a simple email forwarder, and they would blame you for the delays caused by your inevitable clarificatory questioning. But I always had a defence, which is that your email had no inputs and I had to ask questions to clarify.
Now, though? With LLMs, they just run whatever it is (contract, memo, policy) through whatever LLM they have available and paste the output in an email, giving them the appearance of having done work and added value to the project.
But their LLM outputs don't make sense, or don't apply to our organisation, or is a fluffy and abstract "right answer" with no connection to the specific concerns of the business. Parsing it is a chore, and takes time.
And since now I am the only one actually taking that time, I become the visible cause of the delay.
It's infuriating. Any tips on how to deal with this would be greatly appreciated.
If you create a machine for laziness you're going to get lazy people. It's only going to get worse I'm afraid. Do you guys think we're going to see a de-evolution of human beings due to technology?
This is somewhat good news because it means the end (to some extent) of the Peter Principle. My hope, though, is that this doesn't turn into an actual caste system, but I'm not holding my breath. I think that's where this all goes: we're in the process of the world's intelligence being strip-mined and resold back to the world at a premium. The only ones who "survive" long-term are the ones who refuse to submit all of their mental faculties to AI.
That inevitably leads to a rapid degeneration of the species. Only way to avoid that outcome entirely is to remind people that their brain is valuable and they need to exercise it just like they would the rest of their body.
Woodworkers often will have a preference for their favorite tool. Once they become comfortable with that tool they will sometimes prefer to use it for things it wasn't designed. For me it was the band saw. When I needed to split a board, bandsaw. When I needed to cut a curve, bandsaw. Cut a horse figure? bandsaw! I once saw a woodworking television show that seemed to do everything on the table saw. Rip, make frames…
Except that it’s difficult to imagine an even more general tool that would replace AI in this role.