Comments by "𝚿" (@and_I_am_Life_the_fixer_of_all) on "TheAIGRID" channel.

stop giving him ideas
5
their servers need to scale, its running so slow... but this is so cool, already got mine locally setup! this is like a free sora, so cool... also after getting my first video, its probably still worse then sora if I'm being honest
4
its the benchmarks that got people upset with it, might as well continue using 4o if 4.5 has no use case outside of "vibes"
4
don't worry guys, gpt-3 was out for almost a full year before chatgpt was released... and there were some with gpt-3 keys that didn't have the idea or didn't execute on it to make chatgpt... where the public doesn't pay attention is where opportunity remains...
4
@AeroMagic_Official that is not genius at all, its an old concept called GANs, invented in 2014. .A generative adversarial network (GAN) is a class of machine learning frameworks and a prominent framework for approaching generative AI. The concept was initially developed by Ian Goodfellow and his colleagues in June 2014.
2
llm+ml
2
@danypell2517 just make sure you have the resources to keep it running and know how to repair them too - or better yet, make sure you own two, so if one breaks the other one can repair it, and always have them going after the resources to keep themselves running and you alive and well
2
builds scary ass contortionist robot not one wants to buy it sad.png
2
@jondoe22575 I bet he doesn't even have a phone
2
100% this, it didn't do well in the benchmarks, and "vibes" isn't a strong use case
2
hm, interesting
2
I'd pay to control a space robot, but I'd still want to experience zero gravity again and see it all with my own eyes
1
brain in a jar baby, nano robots to make sure it lives on forever
1
@purplelion12v make a deal with yourself that you'll do your best to figure out time travel and time travel your brain and replace it with a replica moments before brain death, that way the 3d model of your brain will bring you back
1
@brianmi40 so far away from maintaining a brain that in 80 years from now it'll still be a challenge - that would be my guess
1
ok, this video was actually good. very good analysis on reasoning by that twitter post.
1
agreed, or maybe not even publish something that isn't outperforming key benchmarks
1
probably a wise decision until o3 drops
1
@robroufla h200s, yeah
1
cloud computing as long as your wallet can handle already does this, any model that is different will be exploited into oblivion
1
everyone is working on this, it is not something that openai, google, meta, x aren't aware of
1
@D_N_Bonacich did you mean hundreds of thousands? anyways, look at the bench marks, gemini doesn't perform all that well in coding
1
hm, yes, what about it?
1
@tiagotiagot who is basil? sky is on pause for now.
1
yeah, I had that in mind, what if anthropic, openai, x, meta and google teamed up to build one model? that would probably be smarter
1
how so? from my testing 4o was outperforming it in stem the only use case i found was "vibes"
1
there's like 2 papers he didn't mention about robotics and physics, I'm assuming it's do to company competition type of shit. It is naive to say he is out of the loop, but I get why you'd say that... it just sounds naive
1
not even the first comment, L comment
1
feasible is not equal to easy, or cheap
1
just think about it, we are all competing against each other's clusters of computing system... they can just tell everyone to collaborate on the same cluster of computers
1
@codycast why wouldn't a company invest on something the government would largely pay you for using? if anything, it would be the ultimate goal over there, to build something so successful that it gets conscripted
1
is that what happened with the advanced voice model? oh wait..
1
yeah
1
hehe time for your crispr therapy hooman
1
internet explorer, the arms race started in 2017
1
dude...
1