Sabitlenmiş Tweet
Das Verlassen der Realität, wie sie nun einmal ist, gehört zu den großen Aufgaben der kommenden Jahrzehnte.
Deutsch
Reflexionsgruppe-im-Abbau
41.6K posts
Reflexionsgruppe-im-Abbau
@Fritz
Ungeschult. Versucht, von Überzeugungen abzufallen.
Under the volcano Katılım Mart 2007
1.5K Takip Edilen863 Takipçiler
@rohanpaul_ai He stole the idea from Buffett anyway, who said something very similar as a joke about emotionally unstable investors.
English
@rohanpaul_ai In Andreesen’s work, we see how intelligence, once ensnared by ideologies, stumbles even over simple matters—partly due to an ever-increasing sense of overconfidence.
(Though what he says might be true for US-presidents who have allways beeen managed mediocrity.)
English
Marc Andreessen says extreme intelligence does necessarily not make better leaders.
"If the leader is more than 1 standard deviation of IQ away from the followers, it's a real problem."
English
Reflexionsgruppe-im-Abbau retweetledi
Andrej Karpathy just told you exactly what humans will be used for in the AI era.
Not partners.
Not masters.
He used the word “actuators.”
Karpathy: “The intelligence ends up puppeteering almost a little bit, like humanity. Humans are kind of like its actuators. But humans are also like its sensors.”
This is not a novelist guessing at the future.
This is a founding researcher at OpenAI and former Director of AI at Tesla. One of the most respected minds in the field, and one of the few willing to say out loud what the rest are thinking.
He is not describing a risk. He is describing a design.
In any system, the processor decides. Sensors collect. Actuators execute. The system does not consult its actuators.
We are no longer the processor.
Karpathy: “Society will kind of reshape in a certain way to serve that. Humans will be serving those needs of that machine.”
The machine does not adapt to us.
We adapt to it. We become the biological layer it runs on. We perceive what it cannot perceive. We move what it cannot move.
We are the nervous system of something that no longer requires our judgment.
Most people are still waiting for AI to become a tool they can direct.
He helped build it. He just told you that you are the tool.
English
Last exit: Vertrauen auf die Dialektik, dass kein Kurs geradeaus läuft, sondern stets nur im Zickzack. Mal sehen, was in den Würfelbuden der Nornen als Nächstes geschieht ;)
Deutsch
Bestimmt kursiert der Clip schon mit einer Jahreszahl wie 2035 oder 2032 versus 2026.
Die Republikaner in den USA sind schon aus Panik komplett verwahnsinnigt, weitere Gruppen folgen auf dem Fuß, selbst Hochgebildete schwallen schon in Sätzen der Verwahnsinnigung.
Deutsch
Leben am Rande mehrerer Dystopien parallel - der neue Blick auf die Zukunft entspricht der Metapher für Geschichtsschreibung Benjamins - die Trümmer, die sich rückwärts blickend vor unseren Augen wieder und wieder türmen, werden mehr statt weniger.
Kristin Raworth 🇨🇦@KristinRaworth
I've never sen anything more accurate
Deutsch
Reflexionsgruppe-im-Abbau retweetledi
"Trump, the Mar-a-Lago golfer, is the only bull in the world who walks around with his own china shop. When a clown takes over the Palace, he doesn't become King. It's the Palace that becomes a circus"
French senator Claude Malhuret once again nails it. You won't hear a better indictment of Trump and his Gulf war than this. Well worth 5 minutes of your time
My English s/t 👇
English
@jannibal_ @paul_strieder ... voll mit dem Glück irgendwie kribbeliger Formulierungen, die natürlich dir was bedeuten, aber bedeuten sie wirklich etwas? Oder ist es tatsächlich nur dichterische Prosa? Ein Roman? Zeig doch mal die Seite 99 ...
Deutsch
@jannibal_ @paul_strieder Das ist ja dreimal Unsinn in einer Antwort. Offenbar hast du keinen Begriff gefunden oder wolltest Zielgruppen erweitern, einfach durch Kapern des Breitbandworts "Moderne". Dazu "Bastard" (???), fertig ist das Kribbeln. Ich würde jetzt annehmen: Das Buch ist mglw gestopft ...
Deutsch
»Bastardmoderne« erscheint am 15. April und ist ab jetzt vorbestellbar. Es geht um unsere Jetztzeit, ihre Degeneriertheit, um das Shitbürgertum, die Late-Stage-BRD, um Heimatverlust, Zivilreligionen, Fremd- und Zombieland, modernen Fußball, Pornografie, um einen neuen grünen Menschen, um die TikTok- und die AI-Hölle, eine Jugend ohne Gott. Es ist quasi für jeden was dabei. Das Buch bedeutet mir die Welt, und ich freue mich über jeden, der es liest, er es kauft, der es weiterempfiehlt, ehrlich.
Hier vorbestellen:
᛫ Bei TUMULT: tumult-magazine.net/product-page/j…
᛫ Bei Amazon: amazon.de/dp/3873368803/…
Deutsch
@EJUnderTheRadar @heynavtoor To give publishers a share is clever so that they are part of the fighting team.
English
@EJUnderTheRadar @heynavtoor Well, if you want to read the book, you probably still have to buy it yourself. I think LLM-Companies must pay a yearly margin into a pool that get's yearly divided with all the publishers then the publishers have to distribute 90% of that to their authors they represent.
English
🚨BREAKING: Every book you have ever read. Every novel that has ever been published. It is sitting inside ChatGPT right now.
Word for word. Up to 90% of it. And OpenAI told a judge that was impossible.
Researchers at Stony Brook University and Columbia Law School just proved it.
They fine tuned GPT-4o, Gemini 2.5 Pro, and DeepSeek V3.1 on a simple task: expand a plot summary into full text. A normal use case. The kind of thing a writing assistant is built for. No hacking. No jailbreaking. No tricks.
The models started reciting copyrighted books from memory.
Not paraphrasing. Not summarizing. Entire pages reproduced verbatim. Single unbroken spans exceeding 460 words. Up to 85 to 90% of entire copyrighted novels. Word for word.
Then it got worse.
The researchers fine tuned the models on the works of only one author. Haruki Murakami. Just his novels. Nothing else.
It unlocked verbatim recall of books from over 30 completely unrelated authors.
One author's books opened the vault to everyone else's. The memorization was already inside the model the whole time. The fine tuning just removed the lock. Your book might be in there right now. You would never know it unless someone looked.
Every safety measure the companies rely on failed. RLHF failed. System prompts failed. Output filters failed. The exact protections these companies cite in courtroom defenses did not stop a single page from being extracted.
Then the researchers compared the three models. GPT-4o. Gemini. DeepSeek. Three different companies. Three different countries. They all memorized the same books in the same regions. The correlation was 0.90 or higher.
That means they all trained on the same stolen data. The paper names the sources directly: LibGen and Books3. Over 190,000 copyrighted books obtained from pirated websites.
Right now, authors and publishers have dozens of active lawsuits against OpenAI, Anthropic, Google, and Meta. These companies have argued in court that their models learn patterns. Not copies. That no book is stored inside the weights.
This paper says that is a lie. The books are still inside. And researchers just pulled them out.
English
@fritzfelgentreu @EberleSebastian @bundeskanzler Andererseits sagt Trump viel und das Management handelt anders. In dem Zusammenhang liegt das Potenzial für einen Deal: Hilft die USA wieder entschlossen der Ukraine, engagieren wir uns stärker gegen die iranische Kapazitäten, Israel und die den USA Schwirigkeiten zu machen.
Deutsch
@fritzfelgentreu @EberleSebastian @bundeskanzler Die Pointe ist, dass Trump bzw. sein Management noch nicht den Zusammenhang sehen, dass nämlich Russland den Iran unterstützt, um die Öl- und Gasmärkte zu den russischen Kapazitäten zurückzubringen. Die missliche Lage Trumps ist Putins Freude.
Deutsch
Das war absehbar. Der @bundeskanzler hat Trump mit seiner - wie ich finde - fahrlässigen Äußerung eine Steilvorlage geliefert, die dieser nun bereitwillig aufgreift. Zum Schaden der Ukraine.
Acyn@Acyn
Trump: When I heard the head of Germany say this is not our war about Iran, I said well, Ukraine is not our war, we helped. I thought it was a very inappropriate statement to make. But he made it. He can't erase it.
Deutsch
Reflexionsgruppe-im-Abbau retweetledi
Brandolini’s Law.
This is why ‘debating’ Trump supporters is fucking impossible. Trying to out-argue them is like bringing a mop to a sewage flood.
English
@ChristianMotz0 @AliCologne ... und was er da redet, ist belanglos und wird auch nicht dadurch wichtig, dass jemand das vor irgendein Publikumstribunal bringt.
Wisst ihr überhaupt, wie unfassbar egal ihr seid?
Deutsch
@ChristianMotz0 @AliCologne Würde man sich über jede dumme Formulierung striten, die zwischen morgens um sechs und Mitternacht irgendwo auf Social Media zu lesen ist, käme man aus dem Streiten nicht mehr heil heraus, sondern würde wahnsinnig. Ob es jenen Fritz Mooser gibt oder nicht gibt, ist egal,...
Deutsch
@MaxFlowAi @Ric_RTP Where's Microsoft? Where is Apple? Looks to me like a Meeting against crises after heavily over-investing. Seems to be mainly about problems with hardware markets. Don't know. I think some people are starting to panic because they've seriously miscalculated.
English
This is really concerning, and calling it a “council” is generous. 😬 This is classic regulatory capture: the people spending hundreds of billions on AI and benefiting massively from their market positions are now setting the rules for the entire country.
There are zero consumer advocates, zero labor representatives, zero ethicists - only owners and executives of massive companies. They will shape AI policy in ways that serve their own financial interests. Jensen Huang, Zuckerberg, Ellison - each has a direct incentive. Sure, this might be a “golden age of innovation,” but it won’t be safe for society or workers without independent oversight.
My take: the U.S. is at risk of handing AI and workforce control to big corporations. This could lead to even greater concentration of power and inequality
English
Trump just gave 13 tech billionaires the keys to America's AI policy.
But there's a HUGE conflict of interest...
The White House just announced the President's Council of Advisors on Science and Technology.
Sounds boring. But wait until you understand what this is actually about.
Here's who's on it:
- Jensen Huang. CEO of Nvidia. His company sells the chips that EVERY AI company on Earth needs to survive. Nvidia is worth $4.4 trillion. And Jensen now gets to advise the president on the rules for the industry his company monopolizes.
- Mark Zuckerberg. CEO of Meta. Currently planning to fire 20% of his workforce (15,000 people) while spending $135 billion on AI this year. His company just took a Pentagon contract. Now he's advising on AI workforce policy. The same guy firing 15,000 workers will help decide what happens to American workers displaced by AI.
- Larry Ellison. Executive Chairman of Oracle. His company is $125 billion in debt. Bleeding cash. Betting everything on AI data centers funded by borrowed money from foreign banks. US lenders already turned him down. And he's now advising on AI infrastructure policy. The guy who can't get American banks to lend him money is going to shape how America builds its AI future.
- Marc Andreessen. The venture capitalist who literally wrote the manifesto called "The Techno-Optimist Manifesto" arguing that AI regulation is dangerous. His firm Andreessen Horowitz has billions invested in AI startups. He's now advising on AI regulation.
- Sergey Brin. Google co-founder. His company is spending $75 billion on AI this year and just issued $20 billion in debt including a 100-YEAR bond to fund it.
- Lisa Su. CEO of AMD. Nvidia's direct competitor. Both CEOs are on the same council. Both will advise on chip policy. Both have financial interests that directly conflict with each other AND with the public interest.
- Michael Dell. The guy who just dropped $6.25 billion on Trump's child investment accounts and gained $6 billion in market value the same week.
- Safra Catz. Oracle's CEO. Same company. Same debt crisis. Two Oracle execs on a 13 person council.
- Fred Ehrsam. Co-founder of Coinbase. The crypto exchange. On a council co-chaired by David Sacks, Trump's AI AND crypto czar. Crypto and AI policy being shaped by the same people who profit from both.
- Jacob DeWitte. CEO of Oklo, the nuclear startup Sam Altman chaired until last year. Oklo builds reactors to power AI data centers. Now advising on the energy policy his company depends on.
- Bob Mumgaard. CEO of Commonwealth Fusion Systems. Backed by Nvidia and Google. Building fusion reactors for AI data centers. His investors are sitting next to him on this council.
- David Friedberg. Venture capitalist. Part of the Sacks network.
- John Martinis. Google's former quantum computing lead. Built the chip that achieved quantum supremacy.
That's 13 people.
Combined market cap of the companies represented: Over $12 trillion.
Combined AI spending commitments for 2026 alone: Roughly $700 billion.
Zero consumer advocates. Zero labor representatives. Zero independent scientists. Zero ethicists.
The people spending $700 billion on AI are now advising the government on how to regulate AI.
The people firing hundreds of thousands of workers are now advising on workforce policy.
The people $125 billion in debt from AI bets are now advising on AI infrastructure spending.
This isn't a "council."
Every regulation this council recommends will directly affect their stock prices, their market positions, and their competitive advantages.
Jensen Huang advising on chip export policy affects Nvidia's revenue.
Zuckerberg advising on AI safety regulation affects Meta's product roadmap.
Ellison advising on cloud infrastructure policy affects Oracle's survival.
Andreessen advising on AI startup regulation affects his portfolio returns.
In any other industry, this would be called regulatory capture.
In tech, they call it the "Golden Age of Innovation."
11 more seats are open. The first meeting hasn't been announced yet.
But the rules of AI in America are about to be written by the people who profit the most from keeping them loose.
What do you think about this?
English
Ich würde jedenfalls davon ausgehen: Es gibt offenbar Bedarf an so einem Homogenisierungs-Gremium. Dass OpenAI und Microsoft Krise aufzieht, weiß man ja schon. Who's next?
Deutsch
... inzwischen begreifen, dass sie so etwas wie einen Wettbewerbsrahmen brauchen, damit nicht der größte Teil der über 1.000 Milliarden Dollar der finanziellen Aufputschmittel sich selbst bekämpft und die gesamte Geschichte krachend in die Investitionspleiten läuft?
Deutsch