Cognitive Dissonance.

We seem to be living in something of a play by Henrik Ibsen. In his 1882 play "An Enemy of the People," the hero/villain is Dr. Thomas Stockmann. 

Stockmann's town is growing richer and richer because of its profitable mineral baths. Stockmann examines the baths themselves and realizes they're contaminated with toxic waste. He blows the whistle and his entire town turns against him. He essentially gets ridden out of town on a rail. That's worser than a ride on the Lexington line at rush hour.

I have two sets of experiences with AI. 

One comes second-hand. From friends and advertising that tells me, repeatedly as in one-thousand times a day, how magnificent the technology is. AI will make everything faster, easier, better, work-free and cheaper. It will fundamentally improve our lives and our world and our livelihood. (This is similar to the blather we heard about how the internet was going to give consumers power, democratize the world, and remove friction in all sorts of areas.)

The second experience with AI is one I've gained not from using it myself, but from having it used upon me. For instance, seeing AI-generated closed captions, or being told by AI how long a particular article will take me to read. (This is also similar to the internet. It seems everyone and his cousin are now UI people or CX people or something, but it's nearly impossible to buy something and even more impossible to return something.)

In each of those instances, and more, AI is wrong just about all the time. It has an accuracy rate well-under fifty-percent. And the scrupulousness of an alcoholic fact-checker for russia's version of wikipedia.

Here are three examples.

ONE:

eBay has just released it's updated Terms & Conditions. Though it is almost 12,000 words--AI asserts (without blushing) that they can be read in ten minutes.

In other words, AI is asserting humans can read 1,200 words per minute. The average reader can read about 250 words per minute. Mistakes like this just sit there. While they accumulate, they never add up the the point where people question the very efficacy of the technology.

TWO:

The other day I listened to an obscure song on YouTube. YouTube's AI gave the writing credits of the song, both music and lyrics to the performer. I questioned it and found the AI was wrong. Otherwise, the error would go unnoticed.

THIRD:

AI-generated captions.

AI-generated closed captions.
Is this a technology you're willing to trust?
https://www.youtube.com/watch?v=AyBH76_kDYo

Please look at this clip. And turn on closed-captions. Listen to the dialogue and read along. 

I'll admit from an enunciation point of view, Brando is no John Gielgud, but the errors in the AI captioning are vast.

If this the closed-captioning in this clip were the answers of a test of the writing ability of someone learning English, that student would fail abjectly. Yet here again, we see things like this every day--evidence that AI is a "low-C-performer," yet people and corporations are spending trillions on it and buying it hook line and sinker.

AI will run our cars. Our airplanes. Our healthcare, education, restaurants, power grid, our wars, our labs, our stock markets, our bank accounts, our old-age homes and more. Based on your own real lived experiences, is this a good idea?

On top of what AI does wrong, there's what AI can't and will never be able to do. That is create something that makes you laugh, think, remember.

On Mondays Frank Bruni of the New York Times compiles a short piece called "For the Love of Sentences." They're words he's read that he likes for their power and distinctiveness. Here are three examples from his June 1 column.

For the Love of Sentences

Will Warasila for The New York Times

On Ken Paxton's primary defeat of John Coryn: “Texas goes through these phases from time to time. Imagine a rich, middle-aged car salesman who ditches his wife and starts dating a 21-year-old stripper with a meth problem, and then imagine that guy is a state — that’s Texas.” 

On Trump's National Mall event: “This isn’t a concert lineup. It’s a clearance rack. A musical yard sale assembled by people who think patriotism means wrapping mediocrity in a flag and turning up the smoke machine.” The headline of the piece from which this was stolen: "The lounge act at the end of the Empire.”

On the current mania for "looksmaxxing":  “A man pumping himself full of testosterone so he can feel masculine again is practically a patriotic act now. Entire podcasts are built around men optimizing themselves into chemically enhanced action figures because God forbid anybody simply age like a mammal.” 

This is writing.
This is what we as communicators should be striving for.
When writing is unique, the brands behind that writing show up as unique.
And our job is to uniquefy brands and therefore make those brands stand out.

AI can't ever do "P as in pneumonia," or "N as in newel post," as below.

Can't.

P as in pneumonia has to be important.
Or We as in you and I are no longer important.
We as in you and I are no longer humans.
People who laugh at incongruity as humans have laughed at incongruity for 200,000 years.

Turn on closed captions here, too. 
For more AI acuity.

My point is not anti-AI.
My point is to urge you question what you're being told--what you're being force-fed. Especially when it doesn't align with the reality you're living in.

Questioning power is what makes a human human.
Power now is spending trillions annually to get us not to question.
If you accept, they win.