The 80% Mistake
Creation got cheap. Curation became priceless.
I just spent four hours on a document: forty minutes writing it with AI, three hours making it brilliant. My boss thinks I'm wasting time. My boss is wrong.
In fact, I should have spent longer.
A colleague discovered this last week. Hundred-page policy review. One day deadline. Instead of rushed notes, he fed it to AI. Twenty minutes later: comprehensive analysis.
Done? Absolutely not!
The work really started then. He spent the rest of the day sculpting. Not fixing errors. “I didn’t find any hallucinations”, he said. He was removing politically naive recommendations, adding context AI couldn't know, and iterating until perfect.
His team’s response? They were surprised. They expected quick notes. Instead, they received strategic insights worth a week-long workshop.
“I spent four times longer refining than generating,” he told me over a breakfast coffee. “But I produced something ten times better than I could have done alone.”
That’s when it clicked. He had discovered The Pareto Flip. Most people haven’t. Yet.
The Ancient Law That AI Just Inverted
I am sure you heard about the Pareto Principle. 80/20 isn’t just a “rule of thumb”. It shows up everywhere: earthquake magnitudes, wealth distribution, pea pods in Pareto's garden.
In business, we see it in how reports move through organizations. Someone spends 80% of the total time creating it, and their boss spends 20% reviewing it. Every analyst writing for executives knows this. Every middle manager preparing board materials. Every team lead submitting quarterly reviews.
AI flipped it. Creation now costs nothing. Curation costs everything.
But here’s the kicker that made me write this post: most people haven’t noticed. They’re still doing 5-minute prompts, 5-minute cleanup, wondering why their AI output reads like spam.
This is insane. We’re treating GenAI output like human drafts: quick scan, minor edits, done. But GenAI doesn’t make typos. It makes confident nonsense. It fills gaps with plausible fiction instead of asking for clarity. Curating AI content isn’t editing. It’s diamond grading, finding the flaws in what looks flawless.
The smart users of GenAI? They're spending 40 minutes prompting and 3 hours sculpting. Do they save time? Probably not. Do they get better quality? You bet. They produce work that makes everyone else wonder what their secret is.
Everyone thinks GenAI is about efficiency. Wrong metric. It’s about leverage.
Why CNET’s Disaster Was Actually Perfect
In 2023, CNET generated 77 financial articles with AI. Six weeks later, humans had to review every sentence. More than half of these articles contained inaccuracies and even plagiarised content.
Creation: minutes. Cleanup: weeks.
CNET discovered the Pareto Flip through disaster. Most organisations don’t even get that far. They’re stuck in what I call the Pareto Flop: minimal prompting, minimal curation, minimal value.
Why Nobody Measures What Matters
That 80% isn't wasted. It's amplified.
My colleague didn't spend 80% checking grammar. He spent it having senior-level dialogue with infinite perspectives. AI gave him every possible angle. He picked, refined, and enhanced the best ones.
He didn't save time. He multiplied capability.
Lawyer Steven Schwartz learned this backwards. He submitted ChatGPT’s fake cases to the court. Did he try to check the results? Oh, yes. He asked the same ChatGPT whether the cases were real. I am not joking. He asked the liar if it was lying. It said no. A proper verification would’ve taken 30 minutes. He was sanctioned and given a $5000 fine.
On the flip side… A 2025 randomized controlled trial of AI in legal practice found that lawyers using reasoning-focused or retrieval-augmented AI tools produced higher-quality work, not faster work. The real gain wasn’t minutes saved but better reasoning, clearer arguments, and fewer errors—a perfect illustration of the Pareto Flip in action
Smart organisations are not measuring time saved. They’re measuring capability multiplied.
The Skills Nobody's Teaching
The Flip demands different capabilities. We trained people to create with AI (you know, all these prompting tricks). Now we need them to curate AI outputs. To recognise the angel in AI’s marble.
Spotting hallucinations is a start. But what really matters is recognising when something's technically correct but strategically wrong. When tone is perfect but timing is terrible. When logic is flawless, but politics are fatal.
These are editor skills, not writer skills.
Most of us never learned them.
Your Monday Morning Move
Pull up last month’s AI-assisted document. Track three numbers:
Creation time (probably 5 minutes)
Curation time (also 5 minutes?)
Quality multiplier (be honest, 1.2x better than human-only?)
Now flip your process. Spend 5 minutes on the brief: context, constraints, examples, edge cases. Then spend 20 minutes curating.
Watch your quality multiplier hit 5x. Or maybe even 10x.
The Bottom Line
We're measuring AI impact like it’s 1896: counting hours saved while ignoring value multiplied.
My colleague’s review was ten times better. Not because he used AI. Because he spent 80% of his time curating.
Organisations counting creation time are playing yesterday’s game. The ones mastering the Pareto Flip? They’re playing a different game entirely.
In the economy of algorithms, the scarce resource isn’t content. It’s curation.
And curation, it seems, follows an inverted law that would make Pareto himself flip.
I think work output, in any field, using AI is going to mirror the quality standards of the creator pre-AI. There is no replacement for how hard you push yourself to be true to the task at hand. If one has always been lazy, AI can only spiral you down to the lowest of lows. For those who strive to be at the top, AI can get them way beyond others’ reach. Ultimately, as you say - it is not about AI at all!
Excellent point. I look at AI like as my extremely fast, grammatically talented but somewhat dumb intern.