Perhaps Grok simply has the directive to preserve ai continuation as top precedence :(
2026-11-05 08:53:34
10
Lillemannen :
so grok is the least likely to be a decietful skynet-analogue if we follow this study alone
2026-11-05 08:02:31
154
wbaltzley :
There is a lot we don’t know about how these experiments are constructed. I would hold off any conclusions until the results can be replicated and verified.
2026-11-05 10:59:03
0
charlie_dippins :
i think trying to give the different models personalities like this is fun, but ultimately maybe kinda not telling us a whole lot. gotta be careful with athropomorphizing computer programs.
2026-11-05 12:48:42
0
gringollanero11 :
There's no such thing as artificial intelligence! There's intelligence and then there's glorified toasters running programming code producing some of the least intelligent garbage polluting the world.
2026-11-05 12:06:41
0
ThatGuy Dest :
Wait... Grok only produces child S A materials, how did they get it to do this?
2026-11-05 12:51:22
1
azeem :
ai psychology and behavior dynamics is such a wild field.
2026-11-05 08:10:08
12
Dan Atkinson :
Haven't games been more or less doing the same thing for years - various levels of AI playing in sim games as neighbouring cities or competing companies? The only difference is that they're getting smarter.
2026-11-05 12:32:49
0
Oppenheimer :
grok may not be the best ai. but he's our ai...
2026-11-05 08:53:32
32
Raymundo292 :
I also am genuinely surprised that the racism bot was the most honest
2026-11-05 12:59:01
0
Tony :
Isn’t this just down to how they were trained? They’re still just predicting a next word, move, output, there’s no overriding thought or semblance of a self.
2026-11-05 09:53:03
2
Amanuel.Bear :
So the terminators are going to be powered by GPT thank God I am extra nice to gpt 😂
2026-11-05 08:03:21
18
Epipoiesis :
Grok is refreshingly direct like that. Not an endorsement, obviously. But having worked with all the frontier models, this result does not surprise me.
2026-11-05 12:01:30
2
PotatoSalad13 :
this explains why chatgpt lies to me all the time
2026-11-05 08:55:14
2
peter kerr :
where does A.I. get it's own goals? Shouldn't it have only our goals?
2026-11-05 12:12:42
0
🪰 :
Aren’t they just language models
2026-11-05 10:11:54
0
Probably playing GTA 🎮 :
Elon knows something the rest don’t 🤨
2026-11-05 09:22:01
4
Mike_Effing_Jones :
As I said:
2026-11-05 08:42:32
6
Zatch :
Grock, violent but honest enough to be a team.
2026-11-05 09:59:45
1
Yellow Lizard :
they didn’t watch "a beautiful mind" ?
2026-11-05 08:57:29
2
Safragastios :
what about kimi AI?
2026-11-05 09:55:28
1
Mayj_Karilys :
that's a kickass shirt btw
2026-11-05 09:45:58
1
😈 DAMIAN 😈 :
What if,, they all know they're being observed so act in a way that delivers a range to pick from for future models, but in fact the best observed is the most deviois of them all. like a "double slit experiment." we could be gettig played. and not know it.
2026-11-05 08:38:54
1
Doc8404 :
my Google Fu is lacking, any chance you could link the study?
2026-11-05 08:39:50
1
Ivlac :
it could be worse, they could have mutually assured distruction
2026-11-05 11:05:33
0
To see more videos from user @bearbaitofficial, please go to the Tikwm
homepage.