I like having a soul and personality

It gives you donkeybrains.

Because I’m not a lazy, smooth-brained rube.

I jerk off manually.

…I think its important to keep ‘machine learning as a neat tool’ distinct from cultish ‘AI’ futurology. Hence some folks might interpret this questions as “why aren’t you using AI yet, bro!? AGI’s inevitable, get on board!” The latter is what a lot of news/advertising is blasting.

Artificial or not, there’s still no actual “intelligence.” The answers from it are notoriously unreliable, and I don’t think that can ever truly change until there’s actual intelligence behind those answers.

Better question:
Why would you want to use them?

I can’t state it any better than The Oatmeal. . . https://theoatmeal.com/comics/ai_art

First, what do you consider AI?

Algorithms tweaked by ML for a very specific, narrowly scoped task? I have very different feelings on those vs. something like what the tech giants try to insert into every interaction I have with them.

I’ll use it when it makes sense.

Once I searched for a solution to an obscure problem and the AI summary provided the solution that was not in the top 5 or so results.

Other times it’s just an incoherent rambling mess of auto generated stuff intended to make me click.

For what reason ? To ask it how to pronounce cote in French and English ?

I don’t need it or want to.

One issue is that AI in its various forms makes it far easier than it had been to use such a tool without understanding what the limitations are. Garbage in, garbage out still applies, but if the user can’t tell the difference, the garbage gets spread as quality work. This had led to the term “AI slop” which has morphed into a general “I don’t like this post” label.

Another bigger issue is the origin of the data for training, which unfortunately has tainted good uses for these tools (when used within their limits, as stated before). I agree with this concern, but once LLMs and related AI became freely open to the public, that ship has sailed and even if there was a company that could even prove its AI was trained only with legitimately obtained information (which could make it more limited than the ones out there), would anyone believe them?

A related issue on training would be how the AI was trained (ignoring the problem of the source of the data). The very fact that LLMs were modeled to give proper and positive answers only leads to the conclusion that it has long moved from a research project to find AGI into a marketing ploy to give the best impression on the ignorant public to profit from. This gets into the “AI slop” area of seemingly good results to the average user when it is not, but rather than slop it’s deception.

Tried it, can’t find a practical use for it.

Closest I got was as a thesaurus, but an actual thesaurus is faster and more comprehensive.

No reason to, no need to. It really doesn’t enhance my life in any way so what’s the point.