A Chinese-American cop sat down for a Bilibili interview. One question made 40,000 people in the comments go dead silent.
When asked about the dangerous calls, he paused. Looked at his left wrist.
Plain analog watch. No pings. No Claude. No agents. Just time.
A few videos over — a software engineer glued to his smartwatch with Claude running non-stop, fighting to stay relevant in a brutal 996 for 20k yuan a month. Every prompt, every automation — he's sharpening the knife that will eventually cut him out.
The cop's watch tells time. The engineer's watch is counting down to his own obsolescence.
One job pays serious money. Real time off. And situations where hesitation means game over.
The other? You optimize yourself into optional.
Here's the part that hit the hardest:
This is one of the last jobs AI genuinely cannot replace. Not for lack of intelligence — but because it demands a real human body and nervous system that has to act when there's zero time to think, let alone type a prompt.
The engineer uses AI daily to prove how valuable he is. His company uses that same AI to learn how little they actually need him specifically.
The cop just needs to come home alive.
Both work their asses off. Only one of them remains irreplaceable.
So tell me — would you rather be in the job AI physically cannot touch… or in the one you're actively helping make replaceable every single day?
23-year-old from Ohio pulled $67,000 in 28 days on OnlyFans without ever leaving his dorm room
Pause at 0:01
The girl doesn't exist and 2,140 paying subscribers have no clue
His roommate heard female moaning through the wall every night for three weeks and filed a housing transfer the next Monday, then walked in to grab his charger and found an empty room
Top fan is a software engineer in Zurich, married seven years, sent Sophie $2,300 last month convinced she's a college student in Miami, and yesterday he texted her a photo of his newborn signed "love you"
Wrong on every single count
Sophie is three markdown files totaling 9 KB, running on a $250 used ThinkPad he bought off Facebook Marketplace
Claude writes every reply she sends, Midjourney renders every selfie she posts, ElevenLabs cloned her voice from a TikTok creator who still has no idea her vocals are turning married men in six countries into paying subscribers
Compute runs him $380 a month, net last month was $51,200, starting capital was $250 flat
OnlyFans paid out $5.8 billion to creators last year and anyone with a folder takes a slice
So who's building your double right now?
@eng_khairallah1 We're no longer just promoters - we're now cycle architects
Whoever first learns to build systems that operate 24/7 without their involvement will win the next decade
Yes, but in the future, they'll settle down so that people won't even be able to afford computers and will just buy them for their arsenal
Because if you look at it that way, you have to pay for minimalism and power, especially since computer prices will increase exponentially in the future
@FurkanGozukara It's both funny and sad
Literally, to show yourself like that, you have to try hard, especially for a person who rules such a strong country
Things aren't really that great, because as we know, neural networks are good on one level
but Meta AI recently simply distributed celebrity logins and passwords, and we literally had to disconnect it from Instagram servers
And now the question is, what's the probability that it won't be used for malicious purposes?
Jailbreaks are still working, and billion-dollar companies can't fix this. And with neural networks like these coming out, it's terrifying
@SawyerMerritt@Tesla THIS IS WHAT FULL SELF DRIVING MEANS WHEN IT ACTUALLY WORKS
The key shift is not the distance but the consistency across highways, traffic, and edge cases without disengagement
If this level of supervised autonomy scales, it changes how people define ownership of driving itself
@chamath Luxury was never about products
It was about signaling
The problem is that logos became accessible
Now the highest-status signal is owning things that don't need to be recognized
@w1nklerr People love the $194K target, but the $54K call next week is what actually makes this prediction testable
Most forecasts get the destination right and the timing wrong
@Polymarket A 68% chance is high enough that insiders are probably positioning already
The probability isn't the story anymore - the market reaction is