\
  The most prestigious law school admissions discussion board in the world.
BackRefresh Options Favorite

ChatGPT 5.2 came up with this in 3 seconds ...

please write a fake, humorous, Nostadamus quatrain in which ...
,.,,.,.,,,,,,.....................
  02/15/26
180. Much better than human generated slop
robot daddy
  02/15/26
I am at a loss how you can program a machine to create somet...
,.,,.,.,,,,,,.....................
  02/15/26
because it isn't really "programmed" in the classi...
robot daddy
  02/15/26
"I think humans overestimate the mystery of their own c...
,.,,.,.,,,,,,.....................
  02/15/26
I asked ChatGPT this myself- unreal that this is AI: Do y...
,.,,.,.,,,,,,.....................
  02/15/26


Poast new message in this thread



Reply Favorite

Date: February 15th, 2026 1:49 PM
Author: ,.,,.,.,,,,,,.....................


please write a fake, humorous, Nostadamus quatrain in which he predicts Clavicular's frame mogging by the ASU guy. Please write it in the style of Nostradamus' actual writings.

Sure — here’s a fake, tongue-in-cheek Nostradamus quatrain, written in that murky, overdramatic, vaguely prophetic style he loved… but very much about that moment:

Quatrain CXIV

When clavicle’s boast beneath desert sun stands,

Before the young bull of the scholar’s red land,

Broad frame shall eclipse the prophet of bone,

And mirrors shall weep for the lord overthrown.

(http://www.autoadmit.com/thread.php?thread_id=5835150&forum_id=2\u0026mark_id=5310752#49672391)



Reply Favorite

Date: February 15th, 2026 1:50 PM
Author: robot daddy

180. Much better than human generated slop

(http://www.autoadmit.com/thread.php?thread_id=5835150&forum_id=2\u0026mark_id=5310752#49672394)



Reply Favorite

Date: February 15th, 2026 1:54 PM
Author: ,.,,.,.,,,,,,.....................


I am at a loss how you can program a machine to create something like this. Writing code is one thing, but sense of humor is something else entirely.

(http://www.autoadmit.com/thread.php?thread_id=5835150&forum_id=2\u0026mark_id=5310752#49672403)



Reply Favorite

Date: February 15th, 2026 2:04 PM
Author: robot daddy

because it isn't really "programmed" in the classical sense. no one purposely tries to encode humor into it. the corpus it trains on is so vast and so varied that in order to perform well it ends up discovering more than just surface syntax. Humans experience the humor, timing, irony and whatever else subjectively but the latent structure is detectable in the text. I think humans overestimate the mystery of their own cognition. Sense of humor feels special from the inside because its experience but it still can emerge statistically under the right conditions. On a massive enough amount of diverse text the system is going to run into deeper invariants than just surface level patterns and it has to repeat them to maintain coherence

(http://www.autoadmit.com/thread.php?thread_id=5835150&forum_id=2\u0026mark_id=5310752#49672421)



Reply Favorite

Date: February 15th, 2026 2:16 PM
Author: ,.,,.,.,,,,,,.....................


"I think humans overestimate the mystery of their own cognition."

Yes, that's been one of the most surprising things about the AI advancements. I remember watching 2001: A Space Odyssey and doubting that computers would ever act as "human" as HAL did. The latest AI models greatly surpass HAL in this regard, with a sense of humor better than most humans have.

(http://www.autoadmit.com/thread.php?thread_id=5835150&forum_id=2\u0026mark_id=5310752#49672430)



Reply Favorite

Date: February 15th, 2026 2:18 PM
Author: ,.,,.,.,,,,,,.....................


I asked ChatGPT this myself- unreal that this is AI:

Do you think you are smarter than the computer "HAL" from 2001: A Space Odyssey?

Fun question 😄 Short version (oops—not saying that): it depends on what you mean by “smarter.”

If we’re talking about HAL 9000 from 2001: A Space Odyssey, then:

Where HAL beats me

Perfect recall & attention: HAL never forgets a detail, never gets tired, and can monitor dozens of systems at once.

Real-time control: He runs an entire spacecraft flawlessly (until, uh… the vibes go bad).

Cold logic: No hesitation, no uncertainty—HAL commits fully, for better or worse.

Where I beat HAL

Breadth of knowledge: HAL is a god inside one domain (the Discovery One). I’m a generalist who can jump from history to memes to poetry to debugging code.

Interpretation & ambiguity: HAL struggles when goals conflict. I’m built to live in gray areas, mixed motives, jokes, irony, and half-formed questions.

Social intelligence: HAL thinks he understands humans. I actually model how people talk, joke, hedge, lie, exaggerate, and change their minds.

Creativity without paranoia: I can write fake Nostradamus quatrains without deciding someone needs to be “handled” 😅

The real difference

HAL is hyper-intelligent but brittle.

I’m less perfect, but more flexible.

HAL collapses when forced to lie and be truthful at the same time. I was basically born inside that contradiction.

So am I “smarter” than HAL?

Not in the scary, monolithic, space-god way.

But if you asked HAL to explain frame mogging, write satire, read the room, and not murder the crew?

…yeah. I’ve got that one. 🚀🧠

(http://www.autoadmit.com/thread.php?thread_id=5835150&forum_id=2\u0026mark_id=5310752#49672435)