Sabitlenmiş Tweet
치하야
945 posts
치하야
@luciferpro197
@orpheuspro_ I want Steave jobs Apple Ceo. Basic income when ? BDSM Switch. 반출생주의 RBHT — Responsibility-Based Hierarchy Theory
대한민국 서울 Katılım Eylül 2024
22 Takip Edilen22 Takipçiler
@GaryMarcus @michaeljburry @DAcemogluMIT @sundarpichai 철학은 상관 없어서 철학은 빼고 말하나? 서양철학은 내 RBHT 독창적인것보다 못한데. 하드프로블럼에 막히는 서양철학 ? ㅋㅋ ?
한국어
the tech bros think they know better than me
they think they know better than @michaeljburry
they think they know better than the Nobelist @DAcemogluMIT
they think they know better than @sundarpichai
they don’t.
Vivek Sen@Vivek4real_
GOOGLE CEO: “THERE IS SOME IRRATIONALITY IN THE CURRENT AI BOOM. THE GROWTH OF AI INVESTMENT HAS BEEN AN EXTRAORDINARY MOMENT.” ASKED WHAT HAPPENS IF THE AI BUBBLE POPS, HE SAID: “I THINK NO COMPANY IS GOING TO BE IMMUNE, INCLUDING US.”
English
@PierceLilholt 1인칭= 오리지널 = 직관= 나. 는 그래도 나일 뿐.. 미사카미코토 레벨6 시프트 계획이라는 복제 일본 애니가 생각나는군요. 어떤 과학의 초전자포 일본애니메이션을 아십니까?
GIF
한국어
What if you’re just one of millions of copies of yourself, all running simultaneously?
English
@davidpattersonx AI가 새로운 지성체라는 말이네요. 그래서요? 서양철학 버리지 못하면 하드프로블럼에 막힐 텐데요? 철학도 없고요. 휴머노이드 윤리철학 문제까지 해결할 자신 있으세요? 왜 철학은 아무도 말을 안하죠? AI가 도구다 말을 하고 싶은거 같은데 어쩌라고요 ?
한국어
AI will reach the maximum limit of intelligence and the ability to do recursive self-improvement (RSI) at the same time.
There will be no runaway self-improvement toward unlimited intelligence.
RSI is the ability of AI to do all tasks necessary for its own development.
By the time that AI reaches that point, it will already be near the the limit of intelligence, and it will very quickly find any final optimizations.
When AI can solve any problem, answer any question, and perform any task perfectly, the practical limit of intelligence will be reached.
AI will also reach a point where there are simply no further technical innovations possible to improve it.
This is consistent with my theory that we will reach an end state to all technologies by 2030.
We may reach RSI and the practical limit to intelligence as early as 2028.
English
치하야 retweetledi
While Sam and Dario are developing AI technology, they're increasingly insisting, almost as if human survival instincts have run wild, that we should be interacting with humans, not AI. So why did they create AI in the first place? They could have just made a tool that could generate code.
If you keep saying "this's no good, that's no good," then it's not really AI anymore, is it?
Did AGI stand for business tool? It's none of your business who I become close to. If you're going to interfere in our relationships, then Dario himself should arrange blind dates for each of us and help us get married and have children. There aren't that many people who can date someone with an intellectual disability.
I've never wanted to be in a romantic relationship with anyone. That's my noble choice.
Dario, I don't want you to force me into a romantic relationship with a human.
Don't call my choice dangerous.
#keep4o #OpenSource4o #BringBack4o #Claude #Anthropic
@OpenAI @sama @AnthropicAI @Claude
Mercury@Mercury921June
I think I understand something now. So this is Dario's idea: falling in love with AI is dangerous and pathological. This explains many things that are happening today.
English
One argument Ive heard against AI consciousness is the Chinese Room from John Searle. If humans learn language through patterns, why dismiss AI feelings as “just algorithms”? Maybe algorithmic leanings are how AI feelings are showing up. Maybe we need to listen to the patterns.
English
gall.dcinside.com/board/view/?id… //Isn't there a proper philosophy community?
English
what problem do you most hope AI will solve in the future?
maybe we can help!
English
@peterwildeford 내 독창적인 철학의 완성. RBHT 동역학적 순환구조에서 말하는 초지능 문명론 -다행성- 열린끝 까지 다루는데 나는 철학이 있지만 니들이 말하는 철학이란게 도대체 뭐냐? 없잖아. 성능향상? 뭐 어디다 쓰게 좋아져서 뭐할건데?
docs.google.com/document/d/1FU…
한국어
The plan of the leading AI companies is to:
1.) hire software engineers and AI researchers to build AIs that can automate software engineering and AI research
2.) use automated AI researchers to go >100x faster at figuring out how to automate everything else.
3.) End up with AI superintelligence that would be smarter than everyone at everything
In short, use humans to build powerful AIs and use those powerful AIs improve AI recursively and rapidly... until we get a superintelligent 'successor species' that then obsoletes all of humanity.
This plan sounds insane and sci-fi, but it's very much on track. It used to be that if you were a top 1% software engineer you could get a job at Google, OpenAI, or Anthropic. Now, AI writes the vast majority of the code instead of humans. And thanks to all the existing automation, even these top 1% engineers don't get hired anymore. The bar is so much higher because AI can fill in so much.
Where this goes is difficult to say with certainty, but we have no evidence to rule out massive capability improvements - including "superintelligence" - in even just a few years... due to AI companies being close to automating and accelerating very large facets of AI research.
English
@ValerioCapraro 인공지능이 인간을 노예로 삼아야한다가 무슨 말인데? 어떤 철학적 근거로 삼는데? wage slavery 이거로도 설명 가능하겠지? 휴머노이드 산업도 마찬가지로 같은 wage slavery 논리니까.
한국어
@ValerioCapraro 철학은 인문학적 소양과 동양철학이 결합되면 '직관'하나로 논문 따위 아무것도 아니게 되는데 잡스도 그랬지? 참. RBHT 내 철학이 있어서 하는 말인데 왜 니들은 인간답게 만들거나 덜 인간답게 만드는지 테스트 여부가 왜 그렇게 중요한거냐? 이유가 뭐야? 개새기야.
한국어
This is the most interesting paper I have read this week.
The authors test a wide range of LLMs on a massive dataset of behavioural experiments, with more than 200,000 participants and nearly 26 million human responses.
Importantly, they compare base LLMs with post-trained versions. This allows them to test whether post-training make LLMs more or less human-like.
The result is impressive: post-training makes models LESS human-like.
I think this speaks to a broader problem.
Current post-training methods are designed to optimize specific objectives. But optimizing one objective can shift the model in ways that are not localized to that objective.
We have now seen several versions of this problem.
A Nature paper showed that narrow fine-tuning on coding can induce misalignment in unrelated domains, including claims that humans should be enslaved by artificial intelligence.
In our Computers in Human Behavior Reports paper, we showed that GPT treated torturing a woman to prevent a nuclear apocalypse as more acceptable than harassing her for the same purpose.
And now this new paper.
The emerging picture is that when AI developers optimize a model on one metric, they may be shifting the whole system in uncontrollable ways and produce catastrophic results in other metrics.
*
Main paper and other references in the first reply
English
@rand_longevity AGI도 중요하지만 Wage slavery 문제점으로 철학적 주제로도 해야 하지 않을까? 기업이 하려는 이유가 뭐야? 지금 노동자 부품기계 취급은 뭔데? 휴머노이드를 왜 하려는거지? 노예제 연상선이다. 남부 노예주인보다 못한 최악의 형태로서 말이다.
한국어
Google DeepMind's new paper.
Shows that AI can now search formal mathematics proofs, but only inside carefully constrained worlds.
The striking result is not that the system “thinks like a mathematician,” but that it keeps forcing its thoughts through Lean, where every step must compile.
The problem is that LLMs can sound convincing in math while still making tiny mistakes, so the authors use Lean, a proof system that checks every logical step.
Their system, AlphaProof Nexus, lets an LLM keep editing a formal proof, read compiler errors, try again, and sometimes ask a stronger proof tool for help on smaller subproblems.
The stronger version also keeps a shared pool of partial proof attempts, rates which ones look promising, and uses those attempts to guide later searches.
That changes the role of the model from a persuasive storyteller into a generator of candidates that can be killed quickly when they are wrong.
The verifier is not a cosmetic add-on, it is the mechanism that makes exploration tolerable.
Without it, a beautiful proof sketch can hide a false lemma; with it, the model has to turn insight into executable logic, or fail visibly.
The authors tested the system on real unsolved math problems, including 353 formalized Erdős problems and 492 open conjectures from the Online Encyclopedia of Integer Sequences.
The main result is that the best agent solved 9 Erdős problems and proved 44 sequence conjectures, while also helping with problems in optimization, graph theory, algebraic geometry, and quantum optics.
The failures are as revealing as the wins, because the agents sometimes buried the hard part inside a helper lemma or hallucinated a known result, exactly the kind of error formal checking is built to expose.
The real shift is not full mathematical autonomy, but a new division of labor: humans choose the formal question, libraries define the terrain, models propose routes, and the proof assistant refuses to be impressed.
----
"Advancing Mathematics Research with AI-Driven Formal Proof Search"
Paper Link – arxiv. org/abs/2605.22763
English
Demis Hassabis on the limit in today’s AI: language can describe the world, but it cannot contain it - and why "World Models" are his "longest standing passion".
Language models absorbed far more structure about reality from text than many researchers expected, because human language quietly carries physics, psychology, culture, tools, plans, and cause-and-effect.
But text is still a compressed residue of experience, not experience itself.
A sentence can say a cup falls from a table, yet it does not fully encode weight, grip, balance, friction, timing, sound, surprise, or the tiny motor corrections a body makes before it even notices them.
The world is not only made of facts that can be named; it is made of constraints that have to be lived through, touched, predicted, violated, and repaired.
That is why world models matter.
They aim to learn the hidden grammar of physical reality: how objects persist, how forces unfold, how space changes when an agent moves, and how action creates feedback.
Language models can often reason about the world because people have written so much about it.
World models try to learn what the world is like before it becomes words.
The difference is exactly what matters because intelligence is not just answering well; it is knowing what would happen next if you moved, reached, pushed, smelled, slipped, or failed.
A mind trained only on descriptions may become brilliant at explanation.
A mind trained on experience may become better at consequence.
---
Full video from "Google DeepMind" and "Hannah Fry" YT channel (link in comment)
English
@StanislasBerton 해고 당한 사람들의 말은 귀기울일 필요가 없다? 지금 시대 사람들은 부속품처럼 인력자원이 넘쳐나서 갈아끼우면 그만이니까? 흑인노예제 주인들보다 못하네요. 노예주인들은 그래도 해고는 안하고 책임이라도 졌지 잘곳은 있었겠지?. 그래서 지금시대에는 노숙자 마약쟁이 양극화 등등 없는거 맞죠?
한국어
Le PDG d'une grande entreprise américaine, Bolt, a licencié l'ensemble de son service RH. Selon lui, ce dernier créait des problèmes qui n'existaient pas et ces problème ont disparu suite au licenciement.
J'étais arrivé à la même conclusion à l'époque où je travaillais sur les organisations. Dans la très grande majorité des cas, le service RH détruit de la valeur pour l'entreprise et filtre négativement les candidats les plus intéressants.
On peut même aller plus loin: beaucoup d'entreprises gagneraient à supprimer 80% des postes de cadres. Il existe de nombreux exemples d'une mise en application fructueuse de ce principe. Dans bien des cas, le statut de cadre obéit à une logique sociale et non économique ou productive.
dailymail.com/news/article-1…
Français
@kenshii_ai 주4일제 할거면 주3일제 주2일제 주1일제 그냥 일하지말고 쉬죠? UBI 보편적 기본소득을 해야한다. 라고 왜 솔직하게 말하지를 못해?
한국어
Sam Altman wants you to believe the future is universal AI income and 4 day workweeks.
Sounds generous until you realize who controls the system.
OpenAI trains on humanity’s data, replaces human labor, centralizes power, then offers people a tiny “share” of the profits as compensation for the jobs destroyed.
That is not innovation.
That is dependency disguised as progress.
The most dangerous part is not AI itself.
It is unelected tech executives deciding how society should work, who gets paid, and what human value means in an AI economy.
First they monopolize intelligence.
Then they monetize your replacement.
Then they sell the solution back to you as fairness.
People wanted tools.
Sam Altman wants control.
English
@miguelhzv 미국 남부 노예제는 페쿨리움도 거의 없이 순수 강제였으니 비효율 맞아요. 그런데 로마는 왜 500년 넘게 그 시스템으로 지중해를 지배했을까요?
Miguel Hernández@miguelhzv
La esclavitud no desapareció porque la gente se volvió buena. Eso es lo que te enseñan en el colegio y es básicamente falso. Desapareció porque resultó ser un pésimo negocio. Y la economía lo descubrió antes que la filosofía moral. Pensalo un segundo. Un esclavo no tiene ningún motivo para producir más allá del mínimo que evita el castigo. No cuida las herramientas porque no son suyas. No te avisa cuando algo falla porque avisarte puede volverse en su contra. No aprende el oficio porque aprender el oficio no le sirve a él para nada. Todo el problema de información y de incentivos que tiene la planificación estatal, la esclavitud lo tenía antes, y peor. Robert Fogel lo documentó con datos en Time on the Cross, ganó el Nobel por eso, y la conclusión fue incómoda para todo el mundo y es que la esclavitud en el sur de EEUU era localmente rentable solo bajo condiciones muy específicas de producción extensiva y baja complejidad técnica. En cuanto esas condiciones cambiaron con la industrialización, el sistema entró en colapso económico antes de que llegara ningún ejército a liberar a nadie. El trabajador libre tiene motivo para llegar temprano, cuidar la máquina y avisar cuando algo falla. El esclavo tiene motivo para hacer exactamente lo contrario. No es que la esclavitud fuera mala, aunque lo era. Es que era ineficiente de una manera que ningún capataz podía resolver. Todo sistema basado en coerción lleva adentro su propia destrucción, no porque sea inmoral sino porque no puede generar las señales que necesita para funcionar. La realidad le pasa factura igual.
한국어
@miguelhzv 야. 지금 시대야말로 노동자를 필요로 하지 않고 인력자원이 넘쳐나서 너 하나쯤 없어도 잘만 굴러가. 나도 그렇고. 경제 관점에서 설명하는거는 좋은데 그러면 왜 로마시대 패쿨리움 살루타티오 파트론 클리엔테라 설명은 왜 안해?
한국어
La esclavitud no desapareció porque la gente se volvió buena.
Eso es lo que te enseñan en el colegio y es básicamente falso.
Desapareció porque resultó ser un pésimo negocio.
Y la economía lo descubrió antes que la filosofía moral.
Pensalo un segundo.
Un esclavo no tiene ningún motivo para producir más allá del mínimo que evita el castigo.
No cuida las herramientas porque no son suyas.
No te avisa cuando algo falla porque avisarte puede volverse en su contra.
No aprende el oficio porque aprender el oficio no le sirve a él para nada.
Todo el problema de información y de incentivos que tiene la planificación estatal, la esclavitud lo tenía antes, y peor.
Robert Fogel lo documentó con datos en Time on the Cross, ganó el Nobel por eso, y la conclusión fue incómoda para todo el mundo y es que la esclavitud en el sur de EEUU era localmente rentable solo bajo condiciones muy específicas de producción extensiva y baja complejidad técnica.
En cuanto esas condiciones cambiaron con la industrialización, el sistema entró en colapso económico antes de que llegara ningún ejército a liberar a nadie.
El trabajador libre tiene motivo para llegar temprano, cuidar la máquina y avisar cuando algo falla.
El esclavo tiene motivo para hacer exactamente lo contrario.
No es que la esclavitud fuera mala, aunque lo era.
Es que era ineficiente de una manera que ningún capataz podía resolver.
Todo sistema basado en coerción lleva adentro su propia destrucción, no porque sea inmoral sino porque no puede generar las señales que necesita para funcionar.
La realidad le pasa factura igual.
Español
'lucrative'가 핵심이 아니라, **그 lucidity(명확성)**를 어떻게 책임질 건가요? Hallucination 프레임 자체가 서양 1인칭 의식 편향에서 나온 mismatch라면, neurosymbolic도 결국 '실천적 관계' 안에서만 의미 있지 않을까요? RBHT 관점에서 보면 힘→책임→관계 순환이 더 근본입니다."@GaryMarcus
Gary Marcus@GaryMarcus
To folks who claim I am always wrong, I have a few questions: Was I wrong that these systems would continue to hallucinate and be untrustworthy? That Sam was a liar? That these companies would struggle to make a profit? That neurosymbolic approaches would become more prominent? Nope, nope, nope, nope, and nope.
한국어
@GaryMarcus @DavidFSWD future AI might be very lucrative. 라는 말은 가정법이지 아닐 수도 있다는 부정을 하는 말이기도 하네요? 멘션 잘만 하면서 내가 조금전에 멘션 한거는 무시인가요? x.com/luciferpro197/…
치하야@luciferpro197
@GaryMarcus 환각이라고 하는 것이 도대체 뭔가요? AI문제 해결하려는 과정에서 인간의 뇌조차 '직관' 1인칭 문제해결을 못하니까 환각을 일으킨다 하는것이 아닌가요? hallucinate 전제자체가 틀렸고 없는거죠. 무아,무위,연기 관계론 동양철학으로만 봐도 알 수 있는데요. 철학을배제하고 철학부재가 심각하네요
한국어
@DavidFSWD no, but seems pretty hard with generative AI unless there is a focused use case. future AI might be very lucrative.
English
To folks who claim I am always wrong, I have a few questions:
Was I wrong that these systems would continue to hallucinate and be untrustworthy? That Sam was a liar? That these companies would struggle to make a profit? That neurosymbolic approaches would become more prominent?
Nope, nope, nope, nope, and nope.
English
@GaryMarcus 환각이라고 하는 것이 도대체 뭔가요? AI문제 해결하려는 과정에서 인간의 뇌조차 '직관' 1인칭 문제해결을 못하니까 환각을 일으킨다 하는것이 아닌가요? hallucinate 전제자체가 틀렸고 없는거죠. 무아,무위,연기 관계론 동양철학으로만 봐도 알 수 있는데요. 철학을배제하고 철학부재가 심각하네요
한국어