OPENAI JUST DROPPED A MATHNVKE ON US
OPENAI JUST DROPPED A MATHNVKE ON US
>>109426268
>>109426290
>>109426296
idc
>>109426303
you're just a codemonkey at best you don't need to care. the people who tell you what to do and how to do it, do.
你顶多就是个码农，用不着操心。真正操心的是那些告诉你怎么做、做什么的人。
>>109426303
unironically this, wake me up when AI cures cancer or invents interdimensional travel
说真的就是这样，等 AI 治好了癌症或者发明了跨维度旅行再叫醒我
Cool. Can't wait until this improves my life in any way whatsoever. Any day now...
酷。真等不及看这东西哪天能改善我的生活。随时都可以来...
>>109426382
Why the fuck should AI even try to cure a disease it can never get? Do it yourself.
AI 凭什么要治一种它永远得不了的病？你自己去治啊。
>>109426268
Aside from non sofic groups all of those are little more than a curiosity
除了非 sofic 群之外，那些都只是图个新奇。
>>109426499
Non sofic groups included. Fuck back to sci or plebbit with this garbage.
非 sofic 群也包括在内。带着这堆垃圾滚回 sci 或红迪去。
>>109426520
Nah you’re just an ignorant nigger
不，你就是个无知的傻逼
so now it officially proven: "group theory" is a bullshit
所以现在官方证实了：“群论”就是扯淡
>>109426684
Why would you want to end suffering when frogposters exist?
有青蛙帖人在，你干嘛还想终结苦难？
Only nonsofic groups are a game changer. The rest are just cool little findings that math PhDs cafe about and maybe have some very minor CS implications.
只有非 sofic 群才是颠覆性的。其他那些不过是让数学博士们兴奋一下的小发现，顶多对计算机科学有点边缘影响。
Non-sofic groups is legitimately the discovery of the century and changes the fundamentals of how we approach math and computer science from now on. Just off the top of my head I can already envision 20+ direct applications that change how things are done.
非 sofic 群真的是世纪级发现，它从根本改变了我们今后处理数学和计算机科学的方式。我随便一想就能想到 20 多个直接应用会改变现在的做事方式。
This is the first true breakthrough AI has done that will affect the lives of every average person in a huge way.
这是 AI 第一次真正的突破，会以巨大方式影响每个普通人的生活。
>>109426455
Wasn't the whole point of AI so that people didn't need to do it ourselves?
AI 的初衷不就是让人不用自己动手吗？
>>109426721
No.
>>109426712
>This is the first true breakthrough AI has done that will affect the lives of every average person in a huge way.
>这是 AI 第一次真正的突破，会以巨大方式影响每个普通人的生活。
Give some examples.
举个例。
>>109426361
You have no clue about the real world. Nobody cares about this shit except egghead academics nobody respects.
你对现实世界一无所知。除了那些没人看得起的书呆子学者，谁他妈在乎这破事。
>>109426712
>group theory bullshitting
>群论瞎扯淡
group theory is a pointless digging into infinity complexity
群论就是对无限复杂性的无意义挖掘
мoзгoвoй aнoннизм in Russian language
俄语里的 мoзгoвoй aнoннизм
>>109426268
>>109426290
usecase?
>inb4 IT COST OPENAI TRILLION DOLLARS
>未防有人说 OPENAI 花了万亿美金
Did it prove the https://en.wikipedia.org/wiki/Poincar%C3%A9_conjecture is wrong?
<----------------
>>109426728
I'm on mobile so I don't want to type everything out. However with nonsofic groups you can construct encryption that can't be broken by turing machines including quantum computers. Perfect cryptography is possible with this.
我在用手机，所以不想全打出来。不过对于非 sofic 群，你可以构造出连图灵机包括量子计算机都无法破解的加密。用这个可以实现完美的密码学。
It can also show the exact limits of neural networks so we will see exactly what tasks neural networks will never be able to solve and adjust the architecture to account for that.
它还能展示神经网络的精确极限，这样我们就能看到哪些任务神经网络永远解决不了，然后调整架构来应对这些限制。
The last is that we can now proof if a system will experience emergent properties and how far these will scale. Like Conway game of life and other cellular automata, we can now see the limits of these systems and proof how far they can scale. This is the closest we can ever theoretically get to solving the Turing halting problem.
最后一点是，我们现在能证明一个系统是否会出现涌现性质，以及这些性质能扩展到什么程度。就像康威生命游戏和其他元胞自动机一样，我们现在能看到这些系统的极限，并证明它们能扩展到多远。这在理论上是我们能最接近解决图灵停机问题的程度了。
Ironically we can use this math to directly prove if LLMs will scale to AGI and ASI or not. And if not, what exactly the pitfalls would be and how to mitigate them.
讽刺的是，我们可以用这套数学直接证明 LLM 是否能扩展到 AGI 和 ASI。如果不能，还能明确指出问题出在哪，以及怎么去缓解。
This essentially guarantees we will reach AGI even if LLMs/transformers won't pan out
这基本上保证了我们能达到 AGI，即使 LLM/transformers 这条路走不通。
>>109426443
>How does this benefit ME PERSONALLY?
>这对我个人有什么好处？
Man I'm starting to think the NWO are right about the whole "useless eaters" thing.
我开始觉得新世界秩序在“无用之人”这件事上是对的。
This implies that they're sitting on a pile of proofs they're choosing not to release at this time.
这意味着他们手里攥着一堆证明，只是现在选择不发布。
>How does this benefit ME PERSONALLY?
>这对我个人有什么好处？
300year of "group theory" bullshitting pissed off
300年的“群论”扯淡把人给惹毛了。
>>109426832
The ones that contribute to AI won't be shared because it would help competitors.
那些能推动 AI 的成果不会公开，因为会帮到竞争对手。
RSI is less than 12 months away and the race is heating up.
RSI 离现在不到 12 个月了，竞争越来越激烈。
are you feeling the AGI yet, anon?
老哥，你感觉到 AGI 了吗？
because we're only just getting started
因为我们才刚刚开始。
>>109426832
collatz
N3 is like: all the group theory disproved
N3 就像说：所有群论都被证伪了。
>>109426832
more likely the most important thing in the loop is getting a human to read and understand the proof and verify it's correct
更有可能的是，这个环节里最关键的是让一个人去读并理解证明，确认它是对的。
>Grandious claims of mathematical discovery that haven’t been subject to peer review and publishing in a western journal
>对数学发现的宏大宣称，还没经过同行评审和西方期刊发表。
I’ve seen this episode before.
这集我看过。
>>109426953
soon and AI opinion will be the only valuable review
很快 AI 的意见就会成为唯一有价值的评审。
>>109426268
is math getting its string theory moment?
数学是不是迎来了它的弦理论时刻？
>>109426953
AI has no peers. The best we can do is expert human review, and they already did that in private with leading experts of the respective fields.
AI 没有同行。我们能做的最好的就是人类专家评审，而他们已经在私下和各领域的顶尖专家过了一遍。
>>109426953
seeing as every other claim they've made has been right, yes, we have seen this episode before
既然他们之前说的每个宣称都是对的，那确实是，这集我们看过。
have fun coping tho
不过你就慢慢自嗨吧。
>>109426953
There's 2MB of Lean for each problem, for starters
每个问题有 2MB 的 Lean 代码，才刚开始。
>>109426824
> Internet flooded with bot traffic and slop garbage
> 互联网被机器人流量和垃圾内容淹没
> Job market fucked up
> 就业市场一塌糊涂
> Huge corporations and oligarchs get even richer while normal people get poorer
> 大公司和寡头越来越富，普通人越来越穷
> Advances in technology being hyped up while it's clear that all it's going to result in is more efficient cat video and spam slop generation, more economic trouble for everyday people, and even more money into the pockets of huge corps and oligarchs
> 技术进步被吹上天，但很明显只会带来更高效的猫视频和垃圾生成、普通人更多经济问题、以及给大公司和寡头口袋里塞更多钱
People have legitimate reasons to be pessimistic
人们有充分的理由悲观。
The only thing that I care about is if it can build Fully Homomorphic Encryption.
我只关心它能不能搞出全同态加密。
>>109426712
>This is the first true breakthrough AI has done that will affect the lives of every average person in a huge way.
>这是 AI 第一次真正的突破，会以巨大方式影响每个普通人的生活。
Lol how much are you getting paid per post ?
哈哈，你发一条帖能拿多少钱？
>>109427068
>Bro we got the RL Q* equation that is AGI, it's way too dangerous to ever be released we call it project strawberry
>兄弟，我们搞到了就是AGI的RL Q*方程，太危险了，永远不能发布，我们叫它“草莓计划”
>Yeah, it's AGI 100% wayy too dangerous, I'm actually scarred
>没错，100%是AGI，太危险了，我都吓坏了
>Released and everyone forgot about it
>发布了然后所有人都忘了这回事
Damn true they have never lied about anything.
靠，还真是，他们从来就没撒过谎。
>>109426268
well at least this means GPT 6 is likely to be a minor improvement rather than a large breakthrough that solved all math problems
好吧，至少这意味着GPT 6很可能只是个小幅改进，而不是什么解决所有数学问题的重大突破
>>109428340
jewish nonsense
犹太人的胡扯
>>109426268
Usecase?
>>109426268
Wow Jewish nonsense
哇，犹太人的胡扯
>>109426268
fuck off with this heckin' velocidensity buzzword bullshit
滚蛋吧，少来这套“速度密度”他妈的花哨废话
So far llm in math seems to have real value for having all of worlds math papers in its heads which is impossible for humans
到目前为止，LLM在数学上确实有真实价值，因为它能把全世界所有数学论文都装进脑子里，这是人类做不到的
>>109429259
Unfortunately, when it’s allowed to store all scientific data in its head, it becomes racist.
不幸的是，当它能把所有科学数据都塞进脑袋里时，它就变成种族主义者了。
I can't. These luddites... Their cognitive dissonance... I just can't.
我受不了了。这些勒德分子……他们的认知失调……我真受不了了。
>>109430304
>it's not a PURE LLM because it's in a harness (tool environment) and a human told it "keep going, you can do it!" every 3 hours
>它不是纯粹的LLM，因为它被装在一个工具环境里，而且每3小时有人类跟它说“继续，你能行的！”
lmao
So can someone explain what nonsofic groups are or why they matter?
那谁能解释一下非sofic群是什么，或者它们为什么重要？
>>109430751
>>109430751
>nonsofic groups are or why they matter?
>非sofic群是什么或为什么重要？
a nonsense (just proved)
一个没意义的东西（刚被证明）
>>109431040
one more Turing machine variant
就是又一种图灵机变体
>>109426823
We know neural networks can scale to AGI simply because human brains exist.
我们知道神经网络能扩展到AGI，就因为人类大脑就是例子。
You could also point to something like MIT the organization and call the organization an example of ASI.
你也可以指个类似MIT的组织，然后说这个组织就是ASI的例子。
>>109430751
it doesnt matter all of these small maths things being solved are academic circle jerk and they arent solved because no one wants to waste time on them
都没屁用，这些被解决的小数学问题都是学术界自嗨，以前没被解决纯粹是因为没人想浪费时间
>>109431781
>neural networks are just like human brains!
>神经网络跟人类大脑一样！
lol
>>109426290
>copy pastes
>复制粘贴党
>>109428340
wow generalizations were retarded?
哇，概括能力是脑残的？
crazy
>>109431953
the generalization achieve nothing
这种概括屁用没有
>>109429689
>351 results found.
>找到351条结果。
（如果你觉得这篇文章有启发，可以点击这里付费）
