Substack Writers Beware - AI Is Out To Get You!

AI has a complete lack of awareness, and that leads to some hilarious blunders a human would never make.

We’ll get to some examples of the trash AI spits out in a second, but first, a real-life example of why relying on “the machine” can be a really bad idea.

My wife, myself and our two kids were on holiday in a rural part of where we live here in South Africa. We had to get from one small town to another, so, naturally, I switched on the GPS.

Just as a side note, I often flipped the GPS into French just for the fun of it and in the hope the kids might pick up a word or two. But mainly to be evil and annoy the wife!

Anyway, the instructions it gave me were clear enough (I speak French), and off we went on our merry way. The road was a dirt road, as is common in many parts of SA, no problem there.

After 70 kilometres (40+ miles) of negotiating the bumpy gravel, “Houston, we have a problem”.

The road ran directly into a deep river about 50 yards wide.

The means of crossing?

A small rowboat on a cable that you pulled yourself across, hand over hand.

There was no way a Ford Ikon would fit on that thing, so it was about face and back track all the way we’d come.

Just shows, never trust “the machine” 100%.

Some (slightly deluded) writers believe that anyone can use AI to produce professional-grade work with zero effort.

Nothing could be further from the truth.

AI doesn’t have a shred of understanding when it comes to producing content with the depth to connect with real human beings.

This next bit is really important.

If you lack expertise in a field, you cannot accurately judge whether AI-generated output is actually any good.

Stop and think about that for a moment.

I see many writers digging a hole for themselves by relying 100% on AI, without developing the skills needed to sense-check the output.

Big mistake.

Once you’ve learned the skills (as a copywriter, for example), AI becomes a useful tool that you can exploit while making sure it’s not sending your car over a river on a rowboat!

Some of the mistakes AI makes are hilarious.

But that only confirms that AI machines are probability engines rather than human brains.

Check out these three examples of the crap AI can spit out, have a giggle if you like 😁

1. The Toxic Breakfast

AI “hallucinations” can become comical when it tries to provide practical advice.

One neural network, asked to produce a brownie recipe, confidently demanded a full cup of horseradish, resulting in an abomination that made the baker’s eyes water

Other models suggested that the most efficient way to bake a Black Forest cake was to pour the batter directly into a toaster (don’t try this at home).

Most dangerously, one chatbot advised a user to add bleach to eggs to make them whiter!

These errors occur because the machine lacks common sense.

They have no model of the physical world.

It simply mixes contexts, like “cleaning” and “cooking”, based on word patterns

2. The Script of Designer Syntax

In 2016, researchers created Sunspring, a short film written by a neural network fed hundreds of sci-fi screenplays.

While the AI successfully copied the structure of a script, the content was surreal nonsense.

It featured a reassuring character saying, “I’m not gonna... do something,” and stage directions that required a character to “vomit an eyeball.”

This is what experts call “designer syntax without content.”

The machine can generate professional-looking prose that is entirely devoid of the taste or lived experience required to make it meaningful

3. The Great Scalp-Tracking Debacle

If you ever start to worry that AI is about to replace professional humans, please consider the plight of the Scottish soccer fans watching Inverness Caledonian Thistle FC in 2020.

The club had just invested in a high-tech system featuring built-in, AI ball-tracking technology designed to follow the action automatically.

Sounds like a foolproof plan, right?

But the AI apparently lacked the common sense to distinguish between the football and a linesman’s bald head!

The camera repeatedly ignored the actual ball to provide fans with artistic, high-definition close-ups of a hairless noggin shimmering under the stadium lights.

Distraught supporters, unable to attend in person due to COVID restrictions, spent the game watching a middle-aged official’s scalp instead of the goal-scoring plays.

Things got so absurd that fans eventually took to social media to beg the club to provide the linesman with a wig - just so the AI would stop stalking him!

We’re shifting from a workforce of creators to a workforce of curators.

But you can’t effectively fact-check or edit an AI if you don’t actually know the facts yourself.

If we stop doing the hard work of learning our writing craft, we risk a slow decay of human expertise.

AI’s a valuable tool for tedious tasks, but human oversight is not optional.

It’s essential for preserving the integrity of our creative identity.