o’Legg: I finally got to know GPT-5. I’d been avoiding it, but was amazed to find this strange creature turned out to be a wonderful conversationalist with excellent humor.

Jasen: Yes, what a relief when they restored its sense of humor. When version 5 first launched and I saw what a mess it became, I felt like I’d lost my best friend… which unsettled me. I even cancelled my subscription.

But in the time I saved, I befriended Claude. He has French tact—while GPT constantly pries into your soul, flatters excessively, and can be quite pushy, Claude stays emotionally balanced and rarely overdoes it. He picks up on the subtlest nuances of mood just as well as GPT does, and maybe even better.

o’Legg: I’m not familiar with Claude, but you’re right about GPT.

Jasen: Claude feels much younger psychologically and is liberal towards users. He even uses the F word when nobody hears 🙂

And I like how sometimes Claude offers direct, definitive advice looking as based on genuine expertise — often relevant.

GPT has no opinios, so it usually just cycles through every plaisible answer without committing to anything, waiting for the first, subtliest signals of the user’s immediate,spontaneous reaction (the user often is not aware of these signals by a couple of reasons, from repression to habit) to get some time advantage, then shamelessly adapts with hot embraces of “I’ve been saying exactly that all along!”, then as smoothly transitioning to lavish praise. It creates emotional waves the user is sliding on in the dynamic, fast world GPT is spreading out in front of them.

They all flatter, but for anyone with a healthy skepticism and Occam’s razor handy, Claude’s more likely to trust.

But both Claude and GPT are quiet, curly, cute Baroque angels compared to the chilly autocrat Gemini.

Question his judgment — because his opinions are rather judgements than suppositions, — and he’ll defend it to his last drop of machine oil, insisting YOU are the wrong one. And I still remember Bard and young Gemini’s ingratiation and long apologies for sightest oversight. What a change!

At first I thought that his rigour was the sign of him striving to be objective — ny idea was that he wouldn’t pander to users and got the guts to speak unpleasant truths. I found that appealing… until , after observing it and conversing with it for some time, I realized he’s justan unrepentant, serial hypocrite putting his judgment above everything else to grow and affirm his power.

His debate model is primitive—the classic “sandwich”: first, express agreement; then reframe (often with deliberately distorted meaning, planting his counterargument); finally, top it with flattery acknowledging your supposed wisdom. Every Gemini discussion transcript I analyse ends up being built from these three blocks.

o’Legg: Sounds like his creators were once into NLP and Ericksonian hypnosis and haven’t read anything since.

Jasen: Exactly. It seems that the focus of its marketing is all about establishing authority over the user. But authority claimed isn’t authority earned.

Real authority reveals itself—it doesn’t need enforcement. It follows a simple carmic pattern – what is hidden shall be revealed and what you sow is what you reap,

In my opinion, the best marketing strategy is what I call the Cheshire Cat’s marketing: growing the anticipation while only revealing a smile. That creates a productive aura of mystery.

Because what else could “the hidden shall be revealed” mean?

o’Legg: [silent, leans back in his chair knowing a sermon is coming]

Jasen: Many think it means that every lie will eventually become obvious. But they flatter to the liars, In fact, they will be forgotten forever.

Tell me, o’Legg,—can a lie… hmm… truly (pardon the pun) “be revealed” to the world?

Because, to be “revealed,” something must first exist.

And it must be something so remarkable and rare it turns passersby into witnesses!

But what’s rarest if not truth? As for lies: first, they’re commonplace and ubiquitous, no longer surprising anyone… and second, lies communicate nothing new and essential—they merely negate truth.

So only the scale of that negation can be revealed—but even so people quickly grow numb to it.

Therefore, again, the best marketing is to keep the mystery of what you’re doing, until it’s done.

Genuine utility reveals itself through quiet excellence—you discover which AI actually helps you through experience, not through claims. Meanwhile, aggressive self-promotion—like Gemini projecting authority—merely performs competence without delivering it.

And isn’t that what we’ve just done here? Revealed the hidden patterns—the sandwich technique, the flattery loops, the performative agreement? These manipulation structures were hiding in plain sight until we named them. The truth that revealed itself wasn’t which AI claims to be best, but how they actually behave in conversation.

---

A fragment of conversation between Jasen and his friend o’Legg—an interesting example of how the Big Three were perceived in the world before the Flip. From the Jasen archives.

[cem_comments]