What AI Won't Do, Part Three

Turtles on a log. Source: Getty Images

Thursday, June 18th, 2026

Bill Bonner, from Youghal, Ireland

---

Will AI make us all “very rich,” as Mr. Trump assures us? Will it make anyone rich?

Take the softball first. Will it make us all rich? It will not — and the very asking of the question betrays a mind that has never paused to consider how the world is actually put together.

“Rich” is a relative thing, not an absolute one. We can no more all be rich than we can all be in the ninetieth percentile; the word itself would be meaningless. So it is with “truth.” With “good.” With “handsome,” and with very nearly every adjective a fellow can pull from the dictionary.

A battery, to carry a positive charge, requires a negative ground; flip it as you please, but you cannot have the one without the other. And when “good” or “fast” or “rich” floods the land in overwhelming quantity, you may as well have none of it at all.

Consider the tallies of yesterday morning. Between the ringing of the IPO bell and his next breakfast, Elon Musk is reckoned to have gained some $400 billion. But the old principle of declining marginal utility tells us that each fresh dollar is worth less than the last. Spread across a whole economy this melancholy truth wears the name “inflation,” and it gnaws quietly at every dollar in your pocket. For a man like Elon, sitting upon roughly $1,400,000,000,000, the next single dollar is worth approximately nothing. Park that fortune in U.S. Treasuries and it throws off some $56 billion a year in interest. Every second is worth $29 to Elon. So, the dollar lying on the sidewalk is not worth stooping to pick it up.

Hold this in mind as we proceed, for it is the very hinge on which the whole contraption swings, and it explains why AI is apt to prove a fat and conspicuous flop. The principle is this: that which is too easily had is not worth having, and that which is already too plentiful is not worth piling higher. In Stockholm a fine bronze tan draws admiring glances; in Dakar or Kinshasa, not so much.

Which brings the follow-up question. If not all of us, then who?

At first squint, the prizes would seem to fall to those chasing them — recklessly, even, with great barrowloads of other people’s money. We refer to the hyperscalers, financed by the trusting shareholders of Microsoft, Meta, Google, SpaceX and the rest. If a queue is forming for those destined to strike it rich, surely they stand at the head of it.

But as we saw yesterday, those apparently best placed to seize a new technology are commonly those most thoroughly chained to the old one.

We all remember Kodak. Its “Kodak moment” marched straight into common usage. Kodak even invented the digital camera back in the 1970s. But it could never coax a profit from it, and slid into bankruptcy in 2012. Today’s Kodaks may meet the same undertaker.

The true Henry Fords of AI may not be those with household names, nor the glittering IPOs. They may not even be Americans. Politico:

People around the world see a winner on AI — and it’s not the US

Respondents in key U.S.-allied countries increasingly see China as the world’s AI leader, while American optimism about the technology continues to erode.

Whatever.

But there lurks a further possibility: that AI may simply turn out to be a dud. It deals in words, designs, fake videos, scraps of information — and it scoops these ‘ideas’ from the internet, where the human race has deposited them in great reeking heaps, like guano upon some forgotten island. A little manure does wonders for the roses. Shovel on too much and they wilt and die. Dollars cheapen as they multiply, and so do ideas. In the stock market, when every last soul is certain a share will rise, one thing alone is sure: it will not rise — for who is left to buy once “everyone” has already laid down his bid?

The machines can ape Shakespeare’s sonnets, draft an airplane, untie a mathematical knot — and do it in an instant. They also work around the clock, without rest. Every eleventh-grader now knows AI can knock out a term paper better than the boy himself and trim a screenplay in a wink, and all of it at very nearly no cost at all.

In the time it takes your editor to grind out one of these dispatches, the machine called Claude can produce a hundred — no exaggeration; we put it to the proof. It needs less than five minutes for Claude to write one. In the six or seven hours we spend chiseling a single column into shape, Claude turns out dozens, and they all sound smarter than we do — like a celebrated economist, like a Financial Times editorial, like a man who actually knows what he is talking about. And then — here is the rub — the next iteration of the machine will feed not upon our labors but upon its own, for it has manufactured so many more of them. The Conversation:

More than half of new articles on the internet are being written by AI

The experts now put AI-generated stuff at 80 percent of the top search results. And since these AI artifacts are groomed and pimped by the algorithms of AI itself, the AI-driven search shows a tender preference for its own offspring. AI-optimized. AI-written. AI-ready. The upshot is that the human sources which once made the machine seem clever are buried in a mounting pile of its own derivatives. AI has stopped imitating human intelligence in other words; it now imitates AI intelligence.

A doubter once accosted Thomas Huxley — the great Victorian naturalist — as he was explaining how the planets wheel about the sun, insisting instead that the world lay flat upon the back of a turtle.

“And what,” asked Huxley, hoping to drive his questioner into an absurdity, “does the turtle stand upon?”

“Another turtle,” said the woman.

“And that turtle — what does it stand upon?”

“No, no, Mr. Huxley, you do not understand. It’s turtles all the way down.”

And so it shall be in the brave new world of AI — a counterfeit world, quality smothered beneath quantity, and humbug all the way down.

Regards,

Bill Bonner

---

---