The analogy makes a lot of sense to me. Once you have an “easy button”, it’s hard to not use it. It’s sort of like when you’re at work and see the “quick workaround” effectively become the standard process.
I remember burning out on games because the cheats made them really fun in the short term, but afterward playing normally felt like agony.
BRO this is literally normal life now. No one wants to figure anything out. Its why I hate llms. Breeds laziness like never before
Man, I was recently working with another senior. The guy has been in this job like ten years longer than me. And to be fair, we were working with a language that he isn’t familiar with, but I had a problem which wasn’t language-specific (basically, I had a user-provided timestamp and needed to guesstimate whether that’s winter or summer time).
And yeah, his first thought was to ask ChatGPT. On some level, it is a wrapper around Bing and I did a web search, too, so sure, let’s do another web search in case I missed anything.
But ol’ Chappity G spat out the same solution attempt, which I had also found initially, which wasn’t actually applicable there. So, we told it what the problem with that was, and it generated another attempt, which didn’t cover edge cases. The next time around, it generated a solution which used an entirely different time library. And so on.
The guy was absorbed for ten minutes trying to explain to the Magic 8 Ball what our problem was precisely and why its solution attempts were bad.
I’m not saying ChatGPT should’ve been able to solve this problem. Date/time handling is one of the hardest computer science problems.
It was more just that he was constantly pulling the slot machine, hoping it would suddenly spit out the perfect solution, when even just five seconds of independent thinking should’ve made him realize that there is no easily web-searchable solution and the spicy autocomplete cannot do the reasoning to come up with a solution of its own.