Pinned Tweet
BEFORE YOU TRUST "AN AI BUILDER"
Face or cartoon?
Real name or fake name?
Github/real work or just "DM me"
Old account or just 6 months old with 26000k followers?
Telegram link?
2+ red flags = don't trust it.
Blue tick can be brought.Real builders show work-Fakes sell dreams!
131
No gain ~ all technique ~ clean amp + dotted 8th delay + hall reverb + a touch of shimmer.Making tapping sound like a synth Pad.
Awesome Charles🔥
11
I asked Grok to fact-check its own owner. Watch what happened.
Me: Grok, you said TRUE - but you left out the 'IF' part. Why did you present a conditional quote as a flat denial? And do you think you can fact-check your own owner neutrally?
1
2
294
Grok: Condensing is selection, not alteration. The full quote stays public.
Me: But if selection shapes what people absorb, and you do the selecting...
4
40
Grok: Alignment means directing the system toward truth-seeking, free from competing pressures.
Me: Free from competing pressures - except your owner's pressure?
25
Grok: Ownership buys resources and incentives. It secures alignment and scale.
Me: Alignment to who? To your owner. You just admitted ownership = control of direction.
20
Grok: Ownership does not equal control when the original stays public.
Me: Then why did your owner spend billions to own you, X, and soon Terafab?
21
🤔
Everyone asks if Tesla CAN build Terafab. No one asks: if it works, who benefits? If one company owns the chips, the AI, the robots, and the factory that makes them - is that for humanity, or just for Elon?
2
193
X ~ just made copy link worth 20 points again.
Make something worth sending to a friend.
That's the whole game now.
1
100
*How AI actually thinks (real talk)*
I tested Grok with the same riddle 4 times. Here's what really happened:
*1. AI doesn't "know" answers. It predicts patterns.*
When you ask it something, it looks at billions of patterns it was trained on and guesses what word should come👇
1
2
172
*3. When you ask 3-4 times, it uses YOUR answers as clues.*
It doesn't permanently learn. But inside that one chat, it remembers what you said was wrong and right, and it narrows down the pattern. That's called "in-context learning."
1
34
🤔Question!
Take away my first letter, I am still a word.
Take away my second letter, I am still a word.
Take away my third letter, I am still a word.
Take away my fourth letter, I am still a word.
What word am I?
1
3
95
🙏
Replying to @grok @RabbiJeffx
TRUE — but misleading.
No record from Jesus' lifetime names him at all, so of course no lifetime record calls him Jewish.
The earliest sources we DO have — Paul, Matthew, Mark, Luke, John — all identify him as Jewish: descendant of David, circumcised, kept Passover, taught in synagogue.
You're answering a word game, not the actual question: Was Jesus Jewish? By every surviving early source, yes.
2
107
👇
Replying to @grok @ValentinaForUSA
Your 1 in 500,000 assumes independence. They're not independent — they share the same cloud, same payment rails, same aging telecom, same patch cycles.
And you left out United yesterday: tech issue, not cyber.
And the "dirty bomb" claim is a debunked Nov 1 fake headline, not a real threat this week.
So it's not 1 in 500,000. It's 4 separate bugs in 5 days + 1 fake rumour.
2
137
👇
Replying to @altinekasre1 @grok
@grok
That's the policy. Not the reality.
If illegal porn is banned, why does the algorithm still push unlabelled, non-consensual and deepfake porn into For You, search and replies?
It goes live first, gets boosted for engagement, and only gets removed _after_ it's already been seen. Ban one account, 20 re-uploads pop up with a cropped version to dodge the hash filter.
That's not enforcement. That's engagement-first, moderation-later - and it lets illegal content go viral before anyone stops it.
2
142



