Ok, first: forgive this title. Yes, it’s here to grab your attention, but has less vitriol behind it than perhaps it sounds. Truth is, whether I like it or not, AI is around in my field. So, spiritually, I’m working on acceptance, and materially, working on artisan web work without it.
My friend from England just asked me what I thought about “all this A.I. stuff… since it’s being pushed on everybody.” Since our mutual countries’ leaders recently tossed out some words about AI, this seemed like a good time to clarify my thinking. My reply seemed written well enough (he said it sounded like AI wrote it 🙄) to share here.
I have many thoughts on it. Let me see if I can get a few down…
1) AI is derivative, trained on what has come before, so unlike a REAL artist or poet, not truly innovative, soulful or “of the times.”
2) AI lies (“hallucinates”), and defends it’s lies in decidedly “charming” and persuasive language, which I see as problematic. People already maybe having a hard time discerning facts, may get tired and just believe some crap it spews out. But the lying ALONE is the worst part = no morality = gets readers used to a plethora of lies that may “sound” true.
3) AI glorifies efficiency, basically is efficiency’s bastard child. Efficiency has a lower strata value than truth, facts, scientific data, or Art.
4) AI was built to respond to prompts, fast. If it doesn’t answer or produce, it is failing its primary purpose.
However, sometimes, silence is the answer. AI can’t do that, or say “I don’t know,” which is also an answer. Sometimes gibberish is a real human answer, but AI is not programed for that- (this may be a good thing in spotting it down the road)
4) AI was trained hugely on thieved materials: copyrighted books, art, systems, etc.. Also it was created because it “could be” but not necessarily because it “should be.”
5) AI is a tool that is erratic: output may be correct, or “overly helpful” and spew extra stuff that complicates a simple task (I have examples for these assertions, but too much to write here).
AGI is different, and I can’t speak to that, BUT if it is trained on what AI is trained on, all of the above *stinky issues* will be at it’s core, so 💩.
We then went on to discuss preventing the scraping of our data. I think I may do that soon. But probably too late. AI has read much of my blog already.
My friend also said my writing reads like AI, a problem that happens to my kid too, so that’s another level of concern. What if a person sounds like AI at times? How do they navigate this time and “prove” their authenticity?
