ZAUEY (Claire Zau)
Region: US
Wednesday 29 April 2026 03:19:19 GMT
39101
3231
66
250
Music
Download
Comments
Thabz 🇧🇼 :
The logic behind his premise is idiotic and unrealistic in nature. Human knowledge created the very chips and software he’s building his platform on. Human knowledge is baked into the algorithms used to build AlphaZero. By eliminating human intelligence you essentially eliminate everything that built what it is. It’s a fallacy to the highest degree. LLMs aren’t the way too btw because they fundamentally misunderstand human intelligence as language (words, numbers and algorithms), but human intelligence and communication is mostly nonverbal. Language is the dressing used and misused to better facilitate translation of the nonverbal languages across distances.
2026-04-29 23:47:31
8
Empire :
We have no choice but to move past just predicting text. We need AI that can actually reason through hard problems
2026-04-29 03:44:57
77
okkmaer :
This shit is gonna be ultron
2026-04-29 07:05:58
31
Optimus Prime Rib :
First Principles …. Aka logic, is still inputted by humans. To make it generalized you would need to input sets of logic for a variety of scenarios or “games” like Alpha had the first principles inputs for chess.
2026-04-30 01:55:49
16
molemohi tsebela :
sadly for him first principles are established by humans
2026-05-01 19:55:34
1
TikTokTouristTraveller :
huge hands
2026-04-30 11:05:30
0
hermosa_jason :
I would like to know what is novel about his approach. On the surface, it sounds just like ehat has been tried for decades.
2026-04-29 21:03:38
1
nic.wins :
How can you train it on anything though? Everything is human experience or knowledge
2026-04-30 10:07:26
0
jerm :
Ask the AI to learn how to be human by simulating humans from first principals. You combine the approaches: use all of human knowledge to teach what the 'first principles are', then ask it to simulate a billion humans all tasked with winning at life.
2026-04-30 02:57:52
0
TrailGuyC :
This is so dumb. How do you do reforce learning without a fixed non inmutable "win" scenario. Just for example how you train an AI to write a good email if there is no "ideal/pefect" email. They learned how to play chess, go, LOL by their own becuause there is a clear fixed immutable objective...
2026-04-30 06:04:55
2
Tommy Chan :
That’s one big af gym env LOL
2026-04-29 07:54:00
2
jcraig949 :
How do I apply for a job?
2026-05-01 12:47:36
1
Tina Duong, ImpactPro Tech :
“Human data ceiling” is a phrase that seems like only AI would write. Since the dawn of time, human ingenuity has always created new things never before imagined. Ok, not every human, but still.
2026-04-29 14:02:39
2
bravozulu00786 :
lol Gemini keeps calling me a principles first thinker
2026-04-29 09:45:08
2
Tyler Andrew :
chess is definitely, go is predictable from human logic
2026-04-29 04:22:45
0
Arthur :
i wonder if in a simulation the a. i. could find penicilin also by chance as we did
2026-05-06 02:19:59
1
Sean Everett :
Sounds exactly like PROME.ai, only it was a decade ago
2026-04-30 17:31:28
1
serpentimiel :
I’ve thinking long and hard about this problem
2026-04-29 08:24:53
1
Tyler Andrew :
Ai only beat the # human living being by learning his thinking patterns
2026-04-29 04:24:35
1
dawufi :
honestly, as a human who knows very little and understands even less, i'm here for it
2026-04-29 04:15:32
1
Bn1m :
It's definitely the wrong thing - LLMs are statistical models they are a party trick they are not the future of AI
2026-06-23 02:13:15
0
Mark :
Interesting and scary. Not sure our species has much time left
2026-06-16 09:26:05
0
itsmayabest :
this is the thesis worth betting on. that is a fully autonomous ai - the future baby.
2026-05-30 21:42:55
0
rjuniyer.com :
good luck! the "general" part is the hardest
2026-06-05 18:20:20
0
Bob Mulligan :
Cold Fusion!!!!!
2026-06-02 20:28:10
0
To see more videos from user @zauey, please go to the Tikwm
homepage.
