this post was submitted on 05 Feb 2024
188 points (83.1% liked)
Asklemmy
43965 readers
1875 users here now
A loosely moderated place to ask open-ended questions
Search asklemmy ๐
If your post meets the following criteria, it's welcome here!
- Open-ended question
- Not offensive: at this point, we do not have the bandwidth to moderate overtly political discussions. Assume best intent and be excellent to each other.
- Not regarding using or support for Lemmy: context, see the list of support communities and tools for finding communities below
- Not ad nauseam inducing: please make sure it is a question that would be new to most members
- An actual topic of discussion
Looking for support?
Looking for a community?
- Lemmyverse: community search
- sub.rehab: maps old subreddits to fediverse options, marks official as such
- [email protected]: a community for finding communities
~Icon~ ~by~ ~@Double_[email protected]~
founded 5 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
If I am going to poke small holes in the argument, the exact same thing happens every day when coders google a problem and find a solution on Stack Exchange or the like and copy/paste it into the code without understanding what it does. Yes, it was written initially by someone who understood it, but the end result is the exact same. Code that was implemented without understanding the inner workings.
The difference being that googling the problem and visiting a page on stackoverflow costs 50-500 times less energy than using ChatGPT.
Really? I haven't done the ChatGPT thing, but I know I have spent days searching for solutions to some of the more esoteric problems I run into. I can't imagine that asking an AI then debugging the return would be any more intensive as long as the AI solution functioned enough to be a starting point.
That's the thing, how do you determine whether or not the "AI solution functions enough" without having a human review it?
The economics aren't there because LLM outputs aren't trustworthy, and the kind of expertise you'd need to validate them is functionally equivalent to that which could be employed to write the code in the first place.
"Generative AI" is an inefficient solution to a problem that's already been solved by the existence of coding support forums like StackOverflow. Sure, it can be neat to ask it for example code or a bedtime story, but once the novelty wears off all you're left with is an expensive plagirism machine that won't even notice when it confidently lies to you.
I have a strong opinion that the problem is more one of people attempting to solve every problem with their shiny new hammer. AI, in the current incarnations, is very good at many things. When implemented properly, LLMs are great at filtering huge amounts of text data or performing semantic analysis. SD does produce images and can be directed.
LLMs are not a replacement for thought. SD is not a replacement for an artist. They are all tools for helping people do things.
I am designing a hypothetical LLM architecture for analyzing the relational structure of a story and mapping it out. I am hoping that it will be capable of generating a meaningful relationship network at the end. It is a very specific goal and a very specific structure. It won't write a story; it won't produce dialog; it won't build a plot. What it will do is build a network of places and characters that can be used to make decisions for all of those things. I want something that helps with internal consistency of models doing other things. So if a GPT model were to write something, it could be fact-checked against the world network to see if what it is saying is reasonable.