CJ Trowbridge
Region: US
Tuesday 08 September 2026 03:18:53 GMT
10161
933
69
23
Music
Download
Comments
dublblind :
I just listened to the interview, lots of pushback from a guy who might be overusing Claude...
2026-09-09 01:12:10
26
libraryofalexistrata :
Wait…. The test for AGI is a captcha?!?
2026-09-08 22:55:15
30
henry.penry :
excuse me.... did you just say Fifty..... THOUSAND..... dollars ..... per prompt......
2026-09-08 03:26:45
83
Segundo :
i'm dumb and don't understand ai but when people code don't their biases come through the in the final product... what if the same thing happens in ai training and fine tuning and they're all just cheaters?! very curious about this help?!
2026-09-08 04:46:56
5
Means Workwear :
Okay I will be watching now
2026-09-09 06:08:50
5
NPC :
this is why I think every story about how dangerous OpenAI tech is, or how it "went rogue" or "hacked" it's just techbros hyping up their company so they can be the last one standing when the bubble busts and money runs out. They are full of shit, it's Enron, but for computing instead of energy
2026-09-09 18:17:00
7
elismith92 :
Maybe a hot take but imo AGI should imply that the harness itself is a neural network instead of an engineered rulebook. Until we figure out how to design and train an executive functioning system built in to these networks I don't think we can call them AGI.
2026-09-08 12:47:11
0
punky b. 🍉 :
"look at the emperor's new clothes! aren't they so fine and regal???" 😂😂💀💀
2026-09-08 13:50:22
12
Joseph A P Holsten :
Reminds me of the anthropic touting Fable benchmarks against OSWorld-Verified claiming it’s equivalent to human operators, when OSWorld-2.0 has been out for months and exists because all these models are still terrible at hidden state (eg, running a game of twenty-questions without cheating by changing its “answer”)
2026-09-10 02:50:17
3
omjustdandy :
definitely agree that "AGI" is nothing but marketing, but I'm curious to know your take on the researchers leaving big labs due to safety concerns. we can all agree that there's dangers that come with more capable models and harnesses right? while I admire the courage it takes to vote with your feet I can't help but think those are the people that should be staying (or at least doing some whistleblowing)
2026-09-09 06:36:55
0
framtidstro :
Novara is great
2026-09-09 17:03:58
0
pug :
considering the hype train, the unspecified "tools" might as well have been a small program made specifically for the puzzles
2026-09-08 06:35:40
10
fixerofbrokenthings :
Humans are cheaters. Humans created AI. AI is a natural cheater.
2026-09-08 16:01:17
2
Mel :
Is this the ARC ag3 test?
2026-09-08 06:16:33
0
kourtney :
i don’t know anything about this at all, but if it knows to cheat, that makes me think it’s smarter than if it knew how to solve puzzles 💀
2026-09-08 14:52:46
1
Peter :
Is it still worth watching. I find Michael insufferable at times and it really puts me off engaging.
2026-09-08 10:23:50
1
Jason Sosa :
Hyping the product. The product isn't the model. It's the stock 😉
2026-09-09 01:21:55
3
dublblind :
Doctorow is great, everyone should be reading him
2026-09-08 05:16:42
27
doritosprincess :
this is starting to feel like theranos
2026-09-08 13:41:25
3
havissard :
Michael definitely seems to do his reading before having these conversations
2026-09-09 07:02:25
1
Starbix :
The unspecified “adapter harness” is probably a human, guiding the LLM to the right answer. 😂
2026-09-10 02:12:54
1
elektricky :
It is too AGI! Mom! Tell them!
2026-09-09 20:51:51
1
Jai ✨ Quantum Theory + Writing :
50K per whet – pardon?!
2026-09-08 18:58:21
1
spin4srv :
AGI will be achieved by burning all the remaining fossil fuels simultaneously and then we can end as a species in peace.
2026-09-08 10:24:55
2
TheRuffus :
I like him on Jon Stewards mini podcasts 🤓
2026-09-08 15:55:20
1
To see more videos from user @cjtrowbridge, please go to the Tikwm
homepage.
