I find it hard to justify the value of investing so much of my time perfecting the art of asking a machine to write what I could do perfectly well in less time than it takes to hone the prompt.
And a professional guitarist can probably pull off a better solo than an audio model, in real time.
And a professional artist can certainly draw exactly what they want faster than talking the robot into rendering it.
Why do we keep comparing the robot to expert humans? You already learned how to do the thing the hard way. No shit the tech isn’t superhuman. There’s still obvious value in a tool that does things for people who know what they want but not how to do it.
I think, the problem is that management wants the expert humans to use the non-expert tools, because they’re non-experts and don’t recognize that it’s slower for experts. There’s also the idea that experts can be more efficient with these tools, because they can correct dumb shit the non-expert tool does.
But yeah, it just feels ridiculous. I need to think about the problem to apply my expertise. The thinking happens as I’m coding. If I’m supposed to not code and rather just have the coding be done by someone/-thing else, then the thinking does not occur and my expertise cannot guarantee for anything.
No, I cannot just do the thinking as I’m doing the review. That’s significantly more time-consuming than coding it myself.
Idk I’m in college and every time I try to use chatgpt to write something I don’t want to learn, I spend like 10 hours prompting and getting nowhere but when I try to read the docs it just works after 30 mins
With music “sounds good” is a sufficient judgment for completeness. With generated code, someone that is an expert has to review it to make sure it does what it’s supposed to, covers edge cases, doesn’t have any security flaws etc. Only an expert is capable, and it is generally faster and produces better quality for the expert to just write the code instead of fixing up what the slot machine dispensed. It’s a cute analogy but all it does is make it obvious that you have no idea what you’re talking about.
And a professional guitarist can probably pull off a better solo than an audio model, in real time.
And a professional artist can certainly draw exactly what they want faster than talking the robot into rendering it.
Why do we keep comparing the robot to expert humans? You already learned how to do the thing the hard way. No shit the tech isn’t superhuman. There’s still obvious value in a tool that does things for people who know what they want but not how to do it.
I think, the problem is that management wants the expert humans to use the non-expert tools, because they’re non-experts and don’t recognize that it’s slower for experts. There’s also the idea that experts can be more efficient with these tools, because they can correct dumb shit the non-expert tool does.
But yeah, it just feels ridiculous. I need to think about the problem to apply my expertise. The thinking happens as I’m coding. If I’m supposed to not code and rather just have the coding be done by someone/-thing else, then the thinking does not occur and my expertise cannot guarantee for anything.
No, I cannot just do the thinking as I’m doing the review. That’s significantly more time-consuming than coding it myself.
Idk I’m in college and every time I try to use chatgpt to write something I don’t want to learn, I spend like 10 hours prompting and getting nowhere but when I try to read the docs it just works after 30 mins
With music “sounds good” is a sufficient judgment for completeness. With generated code, someone that is an expert has to review it to make sure it does what it’s supposed to, covers edge cases, doesn’t have any security flaws etc. Only an expert is capable, and it is generally faster and produces better quality for the expert to just write the code instead of fixing up what the slot machine dispensed. It’s a cute analogy but all it does is make it obvious that you have no idea what you’re talking about.
As if amateurs never code stuff the normal way.