I've changed my mind. I will not make any more tests. This topic bores me. Everyone has their mind already set and I'm not a proponent of any sh*t in particular. My experience has been explained; the scripts have been shared. Period.
AI has advantages.
AI has disadvantages. And flaws.
Above all AI increases the production output by many magnitudes, while at the same time quality is not raised.
And it does nothing really creative new.
All it does is, it finds patterns in already produced stuff, and reorganize those patterns into other patterns by rules of also known patterns. For many routine things that's completely sufficient; to assign boring routine work to a machine. That's what it's orginally meant for, like any other kind of automation.
But this is very boring for real humans interested in anything actually new, particulary creative humans interested in producing creatively new stuff.
Also AI's product quality needs to be controlled very carefully. The two largest flaws of AI are inherent to its very system:
1. While it avoids already known errors, which makes it look so cool, it produces complete new ones instead; sometimes even really tricky, and very hard to find ones. As an experienced programmer you know, it's harder to find errors in something you think, it must be flawless, than when you already know it's buggy. While program's source code is something that can be checked by distinct rules, alas many other things AI produces are neither produced nor tested with expertise, but being released into the wild anyway without being quality approved at all.
2. The speed of production output overtook quality control way faster than anybody can keep up, to pick the useful stuff, while gigantic heaps of useless garbage can be produced in no time.
So, to use AI right also debugging, and quality control, which still need to be done by conventional methods, so cost almost the same time and effort as before like without AI, need to be either adjusted, too, first, to keep up with the speed of output, or new output has to wait, until the former one is tested, and quality approved.
Otherwise places trash up.
Anybody ever lived in a shared accommodation knows: Everybody likes to be the kitchen's chef, whirling with flaming frying pans, impressing others with his/her cooking. But afterwards nobody is there to clean up the kitchen.
The people producing are in the majority, now having a tool in their paws to produce even larger amounts of stuff in even way shorter times, while the people who clean up the mess afterwards neither became more, nor got better tools.
Plus the cost for quality control are not a neglectable, tiny part compared to the benefits AI may bring. Not seldom the costs for to correct some AI output are even larger than if one did it all by her-/himself without AI in the first place. So, it also always needs to be weighed up. Which is another task needed realistically being added to the bill deducted from AI's benefits. Which makes AI to anybody, who worked with and tested it a while, not so very dazzling shiny anymore, as it seems at the first glance, or being hyped to sell a new technology having billions invested.
Not a quarter goes by, when the topic AI is discussed here. Most here don't "just refuse themselves to the topic by prejudices", but (also) already have actual experience with it. Just because they don't share the same enthusiasm doesn't mean they are all by principle completely against this technology. Mostly it's just because they already have enough experience to see like any other new thing it not only has benefits, but also downsides - see things more realistic. They already know how they use it for what, or not. They don't need to be told. And above all, most here are simply tired of this topic AI - not tired of AI per se, but of discussing it again, and again, and again.
However,
when I see "made by AI" written over it, or even smell, a text was written by (or with the help of) AI, I immediately stop reading. I lose interest instantly.
Why shall I read a text by somebody obviously too lazy to write it him-/herself?
Yes. Writing a text needs ten times the time to read it - at least when you try to write it readable, interesting -
for the reader. AI does not change that. It produces texts way faster than anybody can read. But you still need the same time afterwards to make it readable for your readers, anyway. I don't waste my life time for reading some garbage presented by somebody who does not care about the readers.
Particulary I don't waste my time to do any quality control on other's AI output.
Anybody may, or may not use AI. But if, you do it for yourself, only. It's very personal.
But you don't present its output to others. Especially not, as if it was something they've never seen, or being incapable to produce it themselves. That's another downside of AI: The creative value of its products is, as I already said above, none. Worthless. It's the tool, that delivers the product. Everybody can have access to this tool and its sources, so everybody can do it. And what everbody can do is nothing special, not worth to appreciate, not even worth to talk about.
Presenting AI output is like another kind of "Here, let me google that for you!"
To me it's like a little child crying from the toilet:
"Mommy, Mommy! Look, what I've made!!"
"That's nice, pumpkin. Please, just flush now."
www.forbes.com
www.theregister.com
github.com
