Perfection Is the Enemy

Tools and machines are designed for perfection (at least until they break down). I naturally carried that expectation to AI tools when they first arrived and shook the internet.

At first, it was nothing short of magic. It was hard to think of something ChatGPT couldn’t do.

But after the initial fascination wore off, its flaws became painfully obvious in all their glory.

For AI to be most useful to me, I needed it to be near-perfect in all the ways it was not. I learned that it hallucinates. Gives bad advice with the confidence of a scam artist. So using it for reliable research is often out of the question.

And its ability to count is worse than a dumb JavaScript code bookmarklet I use to count the length of a piece of text when analyzing pages for SEO.

Yet, there’s an interesting contradiction with AI. It's perfect in all the ways I don’t need it to be.

AI excels at structure and conformity to its underlying rules.

This means when generating content, it will never make grammatical mistakes or use the wrong verb tenses. Or use inconsistent parallelism. Or make spelling errors. And it loves punctuation.

God it loves punctuation...

(As much as I appreciate a well-punctuated piece of text, it’s hard to match the pedantry with which AI often applies punctuation to text).

It’s this invariable adherence to writing conventions and grammatical correctness that makes AI devoid of character and personality that the best writers breathe into their writing.

Real writers use different sentence structures throughout a piece.

They vary their sentence lengths, letting ideas dictate the complexity of their language rather than following set instructions without fail.

They relate personal anecdotes and find layers of meaning in the ordinary, told compellingly through their unique lens. They're, first and foremost, thinkers.

Great writers break rules.

Human or AI: Does It Even Matter?

We’re naturally developing an intuition for detecting AI-written vs human content.

I strongly suspect that the quirks, imperfections, and the looser degree of uniformity that I alluded to earlier is how most people determine if the content has a thinking human mind behind it. 

But the real question is, does it even matter if something is AI-written or not? 

Is quality of content the only deciding factor or does origin also play a subtle role?

Consumer perceptions about AI are mixed and nuanced at this stage. In the paper Human Favoritism, Not AI Aversion, researchers evaluated how people perceive human-written ad content vs AI-generated content.

The result? People actually preferred AI-generated ad content.

Well, there goes my entire argument. All the human-ness of content amounts to nothing if people aren’t even liking it.

The consumer is the ultimate judge. And the verdict isn’t looking good for human writers.

There is, however, a very important catch that the researchers also discovered.

While people preferred AI content when they weren’t told if AI was involved, people rated the quality of human-written content higher when they knew its source.

In other words, consumers showed a bias toward content produced by human experts. At the same time, they had no aversion to AI content either.

A Content Reset Is Inevitable

Based on results of most studies done so far on the subject, it doesn’t look like people have a strong preference for human made content. Yet.

It’s hard to forget that AI is still fairly new. ChatGPT only became publicly accessible at the end of 2022. Any conclusions we may draw from surveys today are premature at best.

In fact, I think there's a solid chance the trends observed in these surveys might completely reverse in a few years.

Here's why:

At this point, apart from regular AI users, the world isn’t as familiar with AI and its patterns as more experienced practitioners are. 

Once this exposure increases, I have a strong feeling that the average person’s reactions to AI content will become much less enthusiastic.

The monotony of AI content is already becoming hard to ignore.

Secondly, the internet is witnessing an enormous content explosion. We’re producing content much faster than humanly possible, and it’s only going to keep accelerating.

With this oversupply of AI slop, the web is well on its way to get saturated to grotesque proportions. 

Recently, I found it difficult to tell apart two separate newsletter emails from completely different brands. The pattern and overpolished perfections were unmistakable, drowning any individual voice they had of their own.

As my friend Claire noted in her insightful piece, Patterns Will Save Us In The End, we’re heading into a snake eating its own tail situation, with slop snowballing into a Big AI Slop Catastrophe. 

Continued overreliance on AI beyond that point is where originality will die out. Unless the human touch makes a comeback. 

So hold onto your writing quirks and imperfections; there’s no greater impetus for originality.

The propensity for error and human weaknesses might become the most sought-after skills in marketing sooner than we realize.