The AI Experts Are Missing the Point

I met a friend earlier to plan our upcoming book. Halfway through, I noticed something: neither of us had a laptop. Just pens, paper, and two hours of actual conversation.

No notifications. No "let me just check something." No AI assistants offering to summarise our discussion. Just two humans, thinking together.

It was the most productive planning session I've had in months.

Meanwhile, my LinkedIn feed is drowning in AI experts.

They've got frameworks. They've got prompt libraries. They've got "10 ways to use ChatGPT that will blow your mind." They've got courses, newsletters, and a suspicious number of rocket emojis.

And look, some of it is genuinely useful. I'm not here to gatekeep. Learn your tools. Get good at prompting. Automate the boring stuff.

But sitting there with a pen in my hand and an actual human across the table, something clicked: almost all of it focuses on the wrong bottleneck.

The Bottleneck Has Moved

For decades, the constraint was production. Creating content, analysis, code, research... it was slow and expensive. Skilled humans were the bottleneck.

That's no longer true.

Content is cheap now. Drafts are instant. Analysis that took days takes minutes. The production bottleneck has been obliterated.

So what's the new bottleneck?

Taste. Judgement. Connection. Trust.

The things that are irreducibly human.

Denmark Just Proved It

While everyone else doubles down on edtech, Denmark ran the opposite experiment. At the start of the 2025/2026 school year, they pulled tablets, laptops, and computers out of classrooms. Back to physical textbooks. Handwriting. Workbooks.

The results? Undeniable.

Here's the uncomfortable context: Gen Z is the first generation in recorded history to be less cognitively capable than their parents. Since we started measuring cognitive development in the late 1800s, every generation outperformed the previous one. Until now.

Dr. Jared Cooney Horvath, a cognitive neuroscientist who testified before the US Congress, pinpointed the cause: screens in classrooms, starting around 2010. Across 80 countries, once digital technology was widely adopted in schools, performance dropped. Kids using computers for about five hours a day in school score over two-thirds of a standard deviation lower than kids who rarely touch tech at school.

Denmark's response wasn't to lower the bar. It was to remove the distraction.

Even the students admit it's working. One Copenhagen teacher noted the core problem: "It's pretty easy to swipe with three fingers over to a different screen and have a video game going in class."

The tools weren't the answer. The humans were.

What AI Is Actually Good For

This isn't an anti-technology rant. AI has genuine, powerful applications. But they're not what the LinkedIn gurus are selling.

Last week, the European Space Agency used a custom AI tool called AnomalyMatch to scan 35 years of Hubble telescope archives. Nearly 100 million image snippets. In less than three days, the neural network identified over 1,300 cosmic anomalies, more than 800 of which had never been documented in scientific literature.

Jellyfish galaxies. Gravitational lenses. Galactic mergers. Objects that defied classification entirely. Hidden in plain sight for decades.

That's AI doing what AI is good at: pattern recognition at inhuman scale. Finding needles in haystacks the size of galaxies.

But here's what the AI couldn't do:

• Decide those anomalies were interesting
• Ask what a jellyfish galaxy tells us about galactic evolution
• Feel wonder at the scale of the universe
• Write the paper that makes sense of the findings
• Determine what questions to ask next

The AI found the needles. Humans still have to thread them.

The Irony Isn't Lost on Me

Yes, I see the irony of writing a blog post about how content is cheap. But that's precisely the point.

This post isn't valuable because I wrote words. Words are free. This post is valuable (if it is) because of the perspective behind it. The years of watching technology hype cycles. The experience of leading product teams through genuine transformation versus theatre. The judgement to know what matters.

You can't prompt your way to that.

What the AI Experts Should Be Teaching

Instead of "How to write better prompts," how about:

• How to develop taste that distinguishes good from generated
• How to build trust when everyone's sceptical of authenticity
• How to create genuine connection in a world of infinite content
• How to exercise judgement when AI gives you ten plausible options
• How to stay curious when answers are instant
• How to focus deeply when everything is designed to distract

These are human capabilities. They're also the capabilities that will be in desperately short supply.

Denmark figured this out for their kids. The rest of us need to figure it out for ourselves.

The Real Competitive Advantage

Here's my prediction: In three years, everyone will be competent with AI tools. It'll be like knowing how to use a spreadsheet. Table stakes.

The people who thrive won't be the ones with the best prompts. They'll be the ones who:

1. Know what questions to ask (not just how to ask them)
2. Can read a room (not just a dataset)
3. Build relationships that survive satisfying algorithm changes
4. Exercise judgement under uncertainty
5. Create meaning beyond information

None of these are technical skills. All of them are deeply human.

The Uncomfortable Truth

Most "AI transformation" right now is theatre. Companies buying tools, running workshops, creating centres of excellence... while fundamentally changing nothing about how they create value.

The real transformation isn't about AI at all. It's about humans.

It's about asking: now that the production bottleneck is gone, what are we actually here for? What do we uniquely contribute? What connections matter?

Those are harder questions than "what's your favourite prompt?"

But they're the right ones.