I have no need to use LLMs or other generative AI, and I have no desire to use them just because.

For creative outlets, it isn’t satisfying to me to use a tool that will instantly complete the task. If I’m prepping for a TTRPG session I would rather come up with the content myself, or use a random generator to give several ideas to build off of. I don’t have artistic skills, but I have more fulfilment from the basic drawings I can do, and for anything more complex I’ll just find something that comes closest to what I imagined as a visual aid.

For non-creative work, I can’t trust the results of an LLM to be factual. If I need to check the sources and confirm the output anyway, I might as well skip that step and just read the original sources myself. Or, use Wikipedia and other wiki sites as a quick reference for basic information and links to more detailed sources.

If I was working in a field that had to sit through large data sets or complex equations, I probably would look at machine learning models. But I don’t, so I have no need for it.

Genuine question, for people that regularly use AI (LLMs), why? What do you get out of it that makes you return to it again? Is it just convenience?

Is there insufficient BS on the internet?

First, there’s no such thing as actual Artificial Intelligence. In it’s current usage, AI is simply a Large Language Model that takes the enormous amount of data it’s been fed and tries to generate a response that seems like it may be an answer to your question. It has no understanding of the question or the answer, it’s just an estimation of what might be an answer. The fact that there is no guarantee whatsoever that the answer you get is accurate is simply a modern example of the old adage, “Garbage in; garbage out.”

Secondly, there isn’t a single LLM made by a company that I would trust to guess my weight let alone the answer to a question I thought was important.

I use it once in a blue moon and every time think it wasn’t really worth my time.

For creative pursuits, it defeats the purpose of why I’m doing the activity.

For factual pursuits at work, I need to 100% back my answer. Like put my signature to it, testify on a witness stand confident. When I make assumptions in my logic, I need to be able to identify and clearly articulate them. And LLMs are only as good as “highly probable” by their nature.

AI so far has not produced a single useful answer for me. That’s why I ignore it, except to verify it was wrong in the first place.

What people mistakingly call “AI” are just LLMs. Think about what this means: They are parrots with a dictionary, not more. A stolen dictionary, too.

I used it way back when image generation was pretty new… I used it to generate some muse-type stuff for a project I have in mind. See what it was capable of and what the hype was about.

It’s really really bad. I had to save dozens of things that I’ll need to piece together into the cohesive whole I’m looking to paint. But that’s ok because it’s going to be multiple iterations so I need to make it myself anyway.

I could get much better examples now I’m sure, but it’s only for rough models, especially mythical being body proportions (I’m not super artistic, I need examples) and it just doesn’t need to be good in all ways to be good for specific parts of the whole. Plus I really don’t want to pollute my own vision with the average of everyone else’s. That defeats the purpose of making art.

I will probably continue to not use it; image searches that omit AI would do just as well for what I need. Lots and lots of real humans have made art of these things in the past. It’s just unnecessary.

Environmental issues aside, it’s just kinda useless for anything I’d want to use it for. It’s unreliable for even basic research so I’d have to double check anything it’d spit out, so I might as well just skip it and do the research myself to begin with. I don’t care to chat with it or generate images/videos. As a new hobby programmer, I’ve been warned against using it for that from people I know who do it regularly. According to them, it’s not very good at debugging, and the code it writes tends to be bad even when it works.

So, it just seems like a waste of time.

funny calling people lemmings for nothing more than being reluctant to embrace what’s being touted as some scientific miracle to the greedy and gullible who would seemingly all follow it off a cliff - much like a lemming.