Has anyone been SLOWED DOWN by AI like ChatGPT?
Dexter Awoyemi
9 replies
When I rely too much on ChatGPT, it makes writing code harder, especially when developing complex features.
It seems hard for the LLMs to understand both the big picture AND finer details at the same time.
Unfortunately, it's faster not to use AI sometimes ๐ฎโ๐จ
I wonder if other people face this issue when using AI to write code or doing anything else?
Replies
Nick Naumov@nikita_naumov
Paige
Completely agree, when it comes to writing code, it might get slow, especially if you're senior guy. Sometimes it's faster to write a small JS for my website rather than trying to explain ChatGPT what exactly I need. Moreover, you can spend a lot of time asking AI for improvements
Share
Nack AI
@nikita_naumov I hear you. What often strikes me is how much we're all still learning to optimise development with AI
Yes - I've found for niche writing ChatGPT is too generic. Even the outline builders don't help that much.
Nack AI
@maxwellcdavis I like it for ideation, but I agree the default responses are very generic. It can often also be steered to work with your writing style, e.g. if you give it enough examples.
Nack AI
It's an amazing asset for a lot of tasks.
But definitely agree on the over-confident responses. I also find it super interesting that GPT can criticise its own responses pretty well when you feed them back to it.
What a time we live in.
As an AI enthusiast, I find the impact of ChatGPT and other AI technologies fascinating. While AI can sometimes present challenges, such as response delays or misinterpretations, it's important to appreciate its potential and continually improve its capabilities. Let's discuss how we can harness the power of AI while addressing any limitations it may have.
Nack AI
@priyanka_pahari1 For sure! We need to be cognizant of the limitations of our current tech, but it's clear that AI, and eventually AGSI, could transform every corner of our society for the better.
Schovis
Well, I only use ChatGPT to boilerplate the app and it has been a lifesaver there. I tried to use for debugging and fixing errors, but that made me realize how bad it can get in offering "over-confident" false solutions :)