Consider the now-viral genre of questions people pose to artificial intelligence: “The car wash is 100 metres away. Should I walk or drive?”
The AI, applying basic logic, replies: “You should probably walk.”
The internet erupts. Screenshots are shared. Think pieces are written. AI, we are told, is the latest failure of modern civilisation.
But pause for a moment. Did you expect the AI to ask whether you intended to wash your car or merely conduct a survey of people who do?
No context was offered. No objective was stated. And so the AI did what any obedient entity – human or silicon – would do: It made a reasonable guess.
This is not an AI problem. This is a human hobby.
We all know, in theory, that clear instructions, or “prompts” in today’s jargon, reduce confusion and save time. Yet, this basic rule is ignored with remarkable consistency.
Instead, we prefer a more exciting workflow: vague instructions followed by precise disappointment.
And nowhere is this more evident than in the workplace.
Consider this scenario: A report is initially rejected due to grammar errors in the first few pages. The boss stops reading it and hands it back to the subordinate with a terse “correct it”.
This is said with the calm assurance of someone who believes his contribution to the process is now complete.
The subordinate corrects every grammatical flaw known to the English language in the entire document. When the boss reads the amended text in full the next day, the report is now found to be repetitive, its tone is off, and its font lacks authority.
What follows is a slow-burning saga. Change follows change, with the boss repeating: “This isn’t what I meant.”
No further clarification is provided because that would spoil the mystery.
Very soon, everyone is exhausted, and the subordinate is thoroughly demoralised.
Many bosses insist that they are too busy to give detailed instructions. “If I have to explain everything, I might as well do it myself,” they say.
And yet, in a triumph of irony, they end up doing exactly that – just not all at once. Instead, they do it in installments, stretched across multiple revisions, emails, and increasingly confusing comments in Track Changes.
Compared to humans, the main difference is that AI remains unfailingly polite. It does not sigh or accidentally utter a “tsk”. Nor does it roll its eyes and mutter: “You could have mentioned that earlier.”
It simply responds instantly, and after completing the task, it sometimes even asks: “Would you like it punchier?” as if it has not already done enough.
This creates a dangerous illusion. It makes us appear to be excellent communicators.
The truth is far less flattering. Both humans and AI thrive on the same things – clarity, context and specificity.
We must provide background. We must state expectations. We must ask yourselves whether our instructions have covered all the bases.
So, the next time an AI produces an absurd answer, or a subordinate delivers something wildly off target, consider this uncomfortable possibility.
The problem is not artificial intelligence. It is our natural vagueness.
And, until that improves, no amount of intelligence – artificial or otherwise – is going to save us from ourselves.
