So. The internet is dead. There are no humans here anymore, only bots. That was the theory that crawled out of 4chan around 2021, sharing a midwife with Pizzagate and QAnon. 

And, like everything from 4chan (which I hope you don't know firsthand), this theory was quickly disregarded by anyone with a stable job and a stable psyche. I met at least one of those requirements.

Well. I have an update. The internet indeed is kinda... dead.

In June, Cloudflare, which routes a fifth of the web, reported that 57.5% of webpage requests now come from bots rather than people.

Thales, the cyber guys, count 53%.

Another cyber shop, HUMAN, found that traffic from AI agents surged nearly 8,000% in a year.

Meanwhile, 74% of new webpages contain AI-generated content, and roughly half of new articles are AI-generated.

The internet most of us grew up with is not quite dead, then, but it is on its deathbed, and we, its family, should probably start having the conversation about non-revival.

Now, the human brain is a fantastic mechanism. It adjusts to everything. It has simply stopped believing anything written. We are all growing a little Turing-test muscle.

There are some casualties, I must admit. God forbid you were taught to use an em-dash. Or if you are one of the few people alive who deploys "load-bearing" in a sentence, good luck proving you have a pulse.

Yours truly has stopped correcting his own spelling mistakes, in the hope that my earlier sin of letting AI edit with em-dashes will be forgiven.

Clients and stakeholders have stopped reading emails, memos and reports at face value, on the suspicion that nobody wrote them. Universities are retreating to oral exams and handwritten blue books.

Recruiters, at least those who haven't become bots themselves, discard cover letters.

But Sultanbek, you ask, what does all of this mean? Are we, as my children say... "cooked"? Shouldn't we be "tokenmaxxing"? (OK. Enough cringe from me today.)

We are not cooked... yet.

In a fantastic book by Nicholas Carr, The Shallows, which I can't recommend enough, there is a passage about Socrates and his views on the disruptive technology of his day: writing.

Socrates believed writing was a poor way to demonstrate intellect. Anyone can sound smart on paper, given enough time.

Real intellect, he argued, shows only in live speech, where you can be questioned. The irony, of course, is that his argument survives exclusively as... text.

In Plato's dialogue Phaedrus, Socrates makes the case through an Egyptian myth: the god Theuth, inventor of numbers, geometry, astronomy and letters, presents writing to King Thamus, pitching it as "an elixir of memory and wisdom."

The king's review of the Egyptian god's elevator pitch:

"An elixir not of memory, but of reminding" is the most precise description of an LLM ever written.

And "the appearance of wisdom, not true wisdom" describes every AI deck, every LinkedIn thought-leadership post, every memo every steering committee has ever nodded at: "they will hear many things without being taught them, and will appear to know much."

Two things. Train lateral reading, and start trustmaxxing your content. (I promised to stop the cringe... I am a weak man).

So, on lateral reading. It's actually a technique from fact-checkers.

There is an interesting research by Sam Wineburg and Sarah McGrew at the Stanford History Education Group (2017, published in Teachers College Record in 2019). They put three groups in front of unfamiliar websites on contested topics: ten PhD historians, twenty-five Stanford undergraduates, and ten professional fact-checkers.

The historians and students read vertically. Normal stuff. They stayed on the site, scrolled through it, and judged credibility from what the site said about itself: the professional design, the official-looking logo, the ".org" domain, the About page. And they got fooled, repeatedly.

The fact-checkers behaved differently. Before even reading the webpage, they opened new browser tabs and searched the internet about the source. Who owns the website? Who funds it? What do independent outlets say about it?

The result: the fact-checkers reached correct credibility judgments in a fraction of the time, and on the key tasks got it right essentially every time. Wineburg's summary phrase became the technique's motto: "reading less and learning more."

So basically, we need to make sure that when we write materials, they are indistinguishably ours. And that what we write meets our plausible experience.

In other words, if you are an IT specialist who suddenly started writing about Islamic finance, and the timing of your publications corresponds to the releases of new models... that's bad luck, I guess.

Anyhow, I think I'll write a large article about this someday.

Talk soon,
Sultanbek