Shock: Nuance in AI Essay
From listening to my songs (The AI Song and John Henry) and looking at my AI Page you might get the impression that I think NotAI/LLM is dangerous and useless. That’s true for many of the things it is being asked to do/will soon be asked to do.1
But there are tasks for which I think it will be a godsend.
And, as a friend recently said, “I wouldn’t villainize the tool necessarily —it is up to the one who uses it to learn how to wield it properly.” She’s right, but this begs the question of whether we need an elephant gun or a chainsaw (Elon Musk to the contrary notwithstanding) to deal properly with competent artists and writers.
I believe NotAI/LLM is perfectly suited for trivial work. For creative work or thought: never. Which doesn’t mean it won’t be an extinction-level event. And not the first one. Not even the most recent one.
I try to refer to it every time as NotAI/LLM because it is not Artificial Intelligence, it is simply a fast, stupid cut and paste engine: a vastly larger and speeded up version of iPhone predictive text, but no more.
NO
NotAI/LLM’s music, art and writing will always be mindless, lacking a creative soul, literally incapable of anything more than pastiche.
It shouldn’t be your friend or your therapist. No therapist would advise you to commit suicide, or tell you how to do it. A professional therapist guides you in deciding for yourself. An AI aims only to please.
Not AI/LLM shouldn’t be used to translate fiction from English to any other language. It will never write War And Peace or paint the Mona Lisa. It can only produce cover versions.
It shouldn’t be used to analyze job applications. That shouldn’t be done with an algorithm based on keywords either, unless you are looking for someone with terrific skills in gaming the system. Either use a human being or just toss all the résumés up the stairs and hire the person who comes out on top.
NotAI/LLM shouldn’t be used as a customer service bot, unless the human being option is easy to access. My calls are weird: when I don’t fit the tree I am doomed to an endless loop).
It shouldn’t be used for corporate strategy advice, unless you want a strategy that is identical to all the others, regardless of your specific needs.
NotAI/LLM shouldn’t be used as a life coach, unless you want pre-chewed Tony Robbins, useless below 50,000 feet and gibberish on the ground where you live. If you want that kind of cheap coaching, buy the books. Instead of pre-chewed advice, chew it yourself and apply it to yourself, picking out the parts that apply to you (something an AI could never do). Or hire a human life coach with intelligence, experience, compassion and creativity—things NotAI/LLM will never have.
It is totally incapable of Literary translation, which emphasizes appropriate cultural nuances, humor translation, feelings, emotions and the author’s unique literary style. What with it lacking feelings and emotions, or understanding what those are, or being able to develop them based on experience, this is another no-go area.2
Steven Vaughan-Nichols suggests that maybe it should not write code. “<Yawn> I can copy and paste open-source code without understanding what I’m doing and do a bad job of it as well, but no one’s praising me to high heaven.”
As For Art and Music
I have a song called How Many Times. I saw this illustration and asked for permission to use it as the cover. A detailed discussion of AI’s role. Before your click there, here are the two illustrations involved:
I found this one on the Internet and asked for permission to use it.
I ended up using this one, created with the exact same prompt.
I defy any NotAI/LLM to do better than Sweet Mia’s cover for Bask In My Love. It’s creative. You could only write an AI prompt if you knew exactly what you wanted in advance. I have a motto: don’t tell a professional how to do their job, just tell them what result you want. That is impossible with an AI, if you want creativity and have no specific end result in mind.
I even challenge NotAI/LLM to beat me, a rank amateur. AI could never understand this clip art.
If asked using a reasonable prompt, it could never guess what is going on three feet above. So, it could not deduce this illustration implies a kiss.
Instead, it would find something 100% on the nose—two people kissing in some way. There would be no mystery, no implication. This art would only show up in a NotAI/LLM search if someone entitled the stock art “illustration of a kiss.” Save that assistance, NotAI/LLM would never find this in a million years. Yes, you can reverse engineer a prompt to produce (or imitate) this artwork. That’s not possible if you don’t know what you want, but will know it when you see it. AI can’t wander around looking; I can.
As for music, I feel certain it could never write Amazing Eyes, based on my everyday experience, or My First Real Kiss, based on the journal of a friend. NotAI/LLM has no friends and no experiences.
YES
NotAI/LLM is perfectly suited to searches and simple mechanical tasks requiring no intelligence or understanding.
It could competently write a Reader’s Digest Condensed Book3. Not AI/LLM can create a countdown calendar to August 28, 2026, in a month per page format, with the number of days remaining underneath the date in each box. It took three tries. The only way you get a perfect AI product is with a perfect prompt. A human being could do it faster in one try if they were the least bit competent in Excel.
NotAI/LLM can easily produce a literal illustration, exactly as specified. I can only imagine what clever touches the late Mad Magazine artist Mort Drucker would have added to this illustration. The one I got is competent, literal to the prompt, but on the nose with nothing significant added. In short, a pastiche, not a creative product. Because, not being artist, I don’t know how to frame a query that would produce something I can’t imagine, less on the nose but equally illustrative of the concept. I made do with what I got because of time and money constraints, but I would hate to live in a world where, despite having time and money, NotAI/LLM was my only choice.
Concluding Thoughts
There are historical precedents for the effects of AI. Shortly after the invention of the phonograph, history indicates, there was a great reduction in the ranks of performing musicians. Since there was a standard for comparison, only the competent remained.
Perhaps AI will have the same effect as recorded music. When the Comet AI strikes the professional world, it may decimate the market for writers, musicians and artists who are average or below.
Darwinisn says those who remain will be the most talented. It will leave the incompetent and derivative behind. Will they cost more than AI? Surely4, but remember what John Ruskin said: “There is hardly anything in the world that someone cannot make a little worse and sell a little cheaper, and the people who consider price alone are that person’s lawful prey.”
Or the Comet AI may throw every creative person out of work, regardless of their skill or talent. It has happened before. My friend Reid Ashe notes, “We always tend to overestimate the short-term impact of new technology and underestimate the long-term.” And, he adds, “Because this is fundamentally a matter of curve-fitting, expect the machines to be better at interpolation than at extrapolation. They can excel at pattern recognition but come up short in creativity. Expect them to be bad at formulating hypotheses, but good at testing them. Like any tool, we have to learn how to use them.”
In the short term will we be buried in AI slop? Without question. After the Industrial Revolution, the world was buried in lousy machine-made cloth and clothes until they got better, under human guidance. Will AI slop improve? We don’t know how fast, we don’t know how. We just know that the only constant is change, that things will be different. Will they be better? Even I don’t know. Will they be different? That I know. Yes.
FOOTNOTES
Paul E. Schindler Jr.
