Read the full article on Nebraska Examiner
How refreshing it was listening to a trio of bright, buoyant speeches last week, two from the opening of the Obama Presidential Center and Museum, where the former president and Michelle Obama spoke, and one from New York City, where Mayor Zohran Mamdani welcomed the NBA champion New York Knicks to the Big Apple’s city hall. Each of the triumvirate would have made Theodore Sturgeon’s 10%. More on that below.
Yes, I understand haters are going to hate and do so with all the accompanying drivel that drives our hardened partisanship, but in today’s media milieu of outbursts, blame gaming and whole cloth lies often tinged with victimhood, the eloquence, clarity and power of a quality speech was a striking juxtaposition.
All of which made me think of a recent email reply … and the aforementioned Sturgeon.
Therein lies this tale: I recently chose to let the omnipresent Google AI model, Gemini, compose a return email for me. Easy. Peasy. Click. A high-speed, modern iteration of the form letter appeared. I was good to go.
And why not? Like you, I’m busy. Why not have a machine do it for me? We ask Siri for directions, Alexa to play that song we love and Google Assistant to text our lunch date that we are going to be late.
All that was missing in my transaction was a living, breathing human being — in this case, me. What wasn’t missing was a series of paragraphs best described as “serviceable.” If you ask your own Large Language Models (LLMs) for the right word, perhaps they would spit out something along the lines of “workaday,” “passable,” “not bad” or perhaps the twin towers of meh: “fair-to-middling” and “so-so.”
So off I went to AI land. Or, perhaps, Gemini doing for me what I could do for myself, I was simply experiencing a mini-instance of a deeper, darker phenomenon adrift among us, on and offline.
Before we go further, this is not a diatribe against artificial intelligence, that ChatGPT, Claude, the aforementioned Gemini and the rest have undone not simply the Republic but humanity in general. Instead, Generative AI now boosts businesses’ productivity and bottom lines, synthesizes enormous amounts of data for researchers in science, medicine and public health and streamlines repetitive tasks in such diverse arenas such as law, education and banking. All good outcomes.
Still, before I hit “Send,” I edited because I couldn’t seem to shake Sturgeon .. or his law.
Sturgeon was a science fiction author who, in a stunning defense of the genre, which critics said was 90% crap, famously said, “90% of everything is crap.” Hence, Sturgeon’s Law.
One might quibble with his percentages but the upshot of his Law is that we should focus on the valuable 10% of what’s produced — TV, movies, art, writing, music, podcasts, philosophy, pop culture, a gazillion social media communications and, of course, speeches. Anything we find along the information super highway.
To wit: LLMs, which power the ChatGPTs and Claudes of the world, turn the writing, images, thought, whatever, into text and more, imitating the outlines and patterns found in the unfathomable well of Internet data. Filtering that process through Sturgeon’s Law, means we find at least 90 percent of the resulting material is, for lack of a better word, crap.
We even have words for some of the 90%. Slop and hallucinations, slop being the monumental pile of dubious-quality, high volume content AI sends on down the line. Hallucinations occur when AI, which operates on predicting the next word rather than a knowledge of the facts, just makes it up, sparking, among a roster of other damaging consequences, unstinting bouts of misinformation.
Consequently, an overreliance on AI, whether to write an essay, return an email or produce a script for a sales meeting means we are willing to settle for good enough with perhaps a little slop or an hallucination thrown in — settling, if you’re keeping score at home and buy Sturgeon’s premise — somewhere in the 90%.
I have no way of knowing if the Obamas or Mamdani used an LLM to create their remarks, but my guess is no. If you listen — or read a transcript — you can hear and see the humanity in what they have said and written. Sure, AI, which isn’t going anywhere nor should it, can simplify our lives through efficiency, productivity and even a little elegance. When it comes to some things, however, LLMs will never make the 10%.
Our task, it seems, is to sift, sort and learn to appreciate the difference — AI or not.
