A picture of actor Shawn Wallace portraying the character Vizzini; is is mid-exclamation. Across the bottom, black text on a yellow field reads: "Inevitable!"

…But will there be any R.O.U.S.? 

This morning, when I read what felt like the five hundredth post about how AI is “inevitable,” I found myself recalling a great piece of screen writing. 

In the movie The Princess Bride, the scheming Vizzini finds his plans being repeatedly thwarted by the Man in Black. Each time, Vizzini exclaims, “Inconceivable!” 

After the fourth or fifth time, Vizzini’s hired swordsman, Inigo Montoya, says, “You keep using that word. I don’t think it means what you think it means.”

I think Inigo has it exactly right. Because AI is anything but “inevitable.” 

Don’t Believe the Hype

I’m sure by now we have all come to realize that the term “Artificial Intelligence” is a marketing term. It’s certainly not a statement of fact or a promise of what that name will deliver. We’ve seen articles about how some “AI” has been people pretending to be machines. We’ve seen articles about how AI has repeatedly given wrong information and harmful advice. 

Of course, the companies pushing AI into every crevice of our lives are going to promise that it will revolutionize every aspect of our world. Their multi-billion dollar valuations depend on the hype to draw in more investors. 

And of course these same companies are going to confidently claim that AI is “inevitable.” If they could have their way, AI would absolutely be inevitable – they would happily force it on us. 

Fight the Power

But therein lies the rub. AI is only “inevitable” if we choose to buy into it. And the more I learn about the problems inherent in AI – from its creation to its execution to its effects on our minds – the faster I run screaming away from AI. 

I have done a 75-minute presentation on the many issues created by AI, and still had to cut myself off, because there’s that much evidence out in the world that this technology is bad for us. 

I will qualify that previous “us” as “most people.” I will readily admit that there are specific fields of research that might find AI beneficial. But even in those areas – like NASA or medical researchers’ examinations of massive amounts of data – there is a major caveat to any AI use. 

Because AI is not only not intelligent, it’s really quite stupid. It literally cannot think. It can only attempt to find patterns and draw attention to them. Whether those patterns are actual useful information or not still requires careful review and analysis by experts in that particular field. 

GOGI?

Computer programmers used to regularly throw around the acronym “GIGO” – Garbage In, Garbage Out. In other words, if you give a computer bad information or bad programming, then the computer’s outputs will be bad. 

It seems to me that AI has somewhat reversed that equation. It outputs information that is garbage almost as often as it is useful; and too many people are taking those outputs at face value. This is because AI companies have been doing a great job of marketing their product. 

Imagine for a second if a company existed that sold an automatic meal maker. It would take in ingredients, and the AMM would mix those ingredients and output a product. Now imagine if you didn’t know what ingredients were fed into the AMM. And you didn’t get any indication of how the AMM was combining those ingredients, or how they were being processed. 

But one of the steps built into the AMM was to output the final product such that it always looked like a well-cooked hamburger. This appearance was independent of what ingredients were actually used; the AMM was simply passing its output through a mold and colored with dyes as necessary to achieve its final appearance. 

Personally, I would be extremely reluctant to eat that hamburger until I had gone over each piece of it with a fine-toothed comb. And since I’m not a food scientist or nutritionist, I would probably miss some important elements. Sooner rather than later, I’d fall ill from something rotten inside an AMM “burger”. 

A More-Opaque Search Engine

A huge part of the problem we’re facing with the widespread use of AI is that we never had universal information literacy classes. If the general public could’ve had thoughtful school librarians teaching them information literacy as they grew up, I don’t think anyone would have given AI more than a passing glance before laughing it off the stage. 

AI is essentially all the worst parts of a search engine. First, no matter how specific your search string, AI will try to find synonyms. (If you haven’t tried searching your Gmail account lately, be ready for an exercise in frustration when you want results for a specific word!) 

On top of the dilution of your search terms, AI also adds a layer of opacity to how it’s generating the results it provides. Search engines would usually point users to the most popular web page that matched the query. This was not always the CORRECT result. But when you clicked through to the site, you could evaluate it for reliability. 

AI does the same thing, but less well. It looks at what pieces are statistically most likely to follow each other and spits out results that SOUND like they make sense. But even when AI cites sources (which is not always), it’s difficult to know how much AI has “summarized” or re-written the original. And it doesn’t take a lot of branches on the synonym tree to alter the meaning of the message. 

Garbage In…

Add to the issues with AI as a bad search tool the fact that more and more reputable online sources are installing code that blocks AI web scraping. That means AI is increasingly being fed a diet of mediocre-at-best sources. 

Then top THAT with the fact that an increasing percentage of online information is being created by AI. Think about that – AI is creating mediocre content; that content is posted to the web; other AI scrapes the content; that AI dices up the AI-created content and outputs more content; that content is posted to the web…

It’s truly becoming a case of the snake eating its own tail. Except this snake is being touted as the answer to all our problems, even as it becomes decreasingly effective. And this snake’s venom is still leaking all over. 

AI output is toxic. This is true literally, in that there are massive environmental harms caused by the machines that make AI work. But it’s also true figuratively. Its visual outputs diminish the work of human artists. Its textual output needs to be carefully checked by someone with knowledge of the topic. That latter point means that, to use AI “successfully,” one must spend time fact-checking outputs – which not only does not save time, but it causes damage to our cognitive functions. And that’s on top of how AI devalues human skills, knowledge, and experience. 

In short, AI is corrosive. 

You Don’t Have To Eat The Poison!

So despite what the multi-billionaires are trying to tell us, we don’t have to eat whatever slop their automated meal maker spits out. And we shouldn’t be teaching our students to eat the slop, either. 

We absolutely need to teach technology literacy so that students know how AI functions, and all the damage and costs associated with it. The only ROUSs should be our students Repudiating Outrageously Unsound Software

We should unquestionably be teaching information and media literacy skills that will allow our students to analyze facts rather than accepting marketing strategies..

And we really ought to teach our students what the word “inevitable” means.

Because the AI companies keep using that word; and I do not think it means what they think it means.