Aron Brand

@HSwlmy66573 @eyalhayoun @Litalsun @Roy_Iddan @grok האם הטענה בנוגע ליהודי נכונה ?

@eyalhayoun @Litalsun @Roy_Iddan נכון, גם לא רוצח מתועב שרצח את השכנה. אנחנו נלחמים באויב של מדינת ישראל, בשביל זה החוק. גם נגד יהודי שהצטרף לחמאס, וגם אם הוא רצח אזרח ערבי. נגד עבריינים, אנרכיסטים, רוצחים, גזענים, אנסים וכו', יש ספר חוקים שלם.

זה לא ״חוק עונש מוות לפלסטינים״. זה חוק למחבלים שרצחו אזרחים. כשאפילו התקשורת הישראלית מעוותת ניסוח - אל תתפלאו איך העולם מבין את זה

Microsoft’s Researcher for Copilot hints at a bigger shift: the model labs may be the only players who can’t use the best reasoning workflow. Labs mostly have self-play: model vs copies of itself. Same priors, same blind spots. Platforms can run true cross-model critique: OpenAI generates, Anthropic challenges, another model verifies. Once workflow owns quality, the underlying model becomes interchangeable. Power moves from labs to platforms.

Vibe coding collapses three roles into one loop: developer, product manager, and user. That collapse is the point. The old system was a sequence of translations, each adding latency and distortion. Now the loop runs at thought speed: idea → artifact → felt experience → correction. That speed is not just acceleration, it’s a phase change. When iteration becomes nearly free, you stop optimizing for execution and start optimizing for judgment. The question shifts from “can we build it?” to “is this actually good?” Taste, not throughput, becomes the limiting factor. So the builder’s role changes. Less implementer, more conductor. You don’t fully specify upfront. You interact, probe, adjust. The product emerges through dialogue with the system. Decisions move from abstract reasoning to lived experience inside the loop. But collapsing the loop also collapses perspective. When you are the user, it’s easy to overfit to your own mind. The loop gets faster, but narrower. The real skill is holding both: operating at high speed while maintaining enough distance to see what others will actually experience.

@sweatystartup This take is nonsense, it applies to the handful of frontier model builders burning cash, but for everyone else using AI as a tool it’s the opposite of a house of cards.

Investing heavily in AI at your company will backfire. You are becoming dependent on something that is unsustainable. The VC money will dry up once they realize nobody is going to make any money in the long run except NVDA and the power companies. The subsidies will stop. And your costs will 5x. There is no moat in AI. Switching from GPT to gemini to grok to claude takes seconds and you don't miss a beat. Its a house of cards.

@sundeep He can triple his productivity with $5,000. With $250,000, he can drown his entire organization in 100 billion tokens of unverified technical debt

“If your $500K engineer isn’t burning at least $250K in tokens, something is wrong.”

@TFTC21 He can triple his productivity with $5,000. With $250,000, he can drown his entire organization in 100 billion tokens of unverified technical debt

Jensen Huang: "If that $500,000 engineer did not consume at least $250,000 worth of tokens, I am going to be deeply alarmed. This is no different than a chip designer who says 'I'm just going to use paper and pencil. I don't think I'm going to need any CAD tools.'"

@edandersen He can triple his productivity with $5,000. With $250,000, he can drown his entire organization in 100 billion tokens of unverified technical debt

Software engineers will not be trusted to spend 50% of their salary on variable opex costs with no guarantee of productivity, unless they are executive level. this is a pipe dream to sell GPUs

@Rahll He can triple his productivity for $5000 . With $250,000 he can generate an unmanageable bottleneck of 300 million badly tested tokens a day

Wait wait wait waaait a second.... He's saying he wants his $500,000 engineer to actually cost him $750,000 by using $250,000 worth of AI tokens. Wasn't AI supposed to make things cheaper, not cost 50% more?

@Pirat_Nation $250k in API spend buys ~108B tokens a year (~296M/day), far beyond what any human can realistically review or validate. AI can generate at that scale, but until agents can reliably review and ensure correctness, it’s mostly a nice dream for Nvidia shareholders and leadership.

Nvidia CEO Jensen Huang said he would be deeply alarmed if a $500,000 engineer did not consume at least $250,000 worth of tokens annually. “If your $500K engineer isn’t burning at least $250K in tokens, something is wrong.” Huang also said that if an engineer reported only $5,000 in token spend, he would "go ape something else."

@mmasarax1 @DiscussingEcon1 @maayantal87 אמירות גזעניות כמו ״בבון מהרי אטלס״ אסור לנרמל . הדוגמה שבראש הפוסט הנוכחי לא נופלת בקטגוריה הזו

@aron_brand @DiscussingEcon1 @maayantal87 מספיק או שאתה צריך עוד?

האישה הזאת, במדינת ישראל, בשיא יוקר המחיה, חסכה כסף והצליחה לצאת לפנסיה בגיל 31 רק עם השקעות בבורסה ואורח חיים פשוט. אותה אישה, לא מפסיקה לטנף על המדינה ועל כמה שרע פה ועד היום היא אנונימית מישהו יודע למה?

@mmasarax1 @DiscussingEcon1 @maayantal87 תן דוגמאות

@aron_brand @DiscussingEcon1 @maayantal87 זו לא הביקורת. זה הסגנון. והציוץ הזה שלה עוד עדין. הבחורה/בחור מפיץ רעל ושנאה באופן קבוע ובסגנון אלים וקיצוני. אם אתה חושב שהסגנון הזה סבבה זה שלך. אבל אם אתה בסדר עם זה אתה או צבוע או אלים בדיוק כמוה.

@mmasarax1 @DiscussingEcon1 @maayantal87 "לטנף״ - מילה נרדפת לביקורת עבור מי שהביקורת לא מוצאת חן בעיניו

@DiscussingEcon1 @maayantal87 בוא נעשה סדר. היא מטנפת על ציבור שלם שוב ושוב. מטנפת על המדינה, על הממשלה, על העם. תקרא לזה איך שאתה רוצה. היא (הוא) חלאה אלימה ואם אתה מצדיק את השיח הזה אתה גרוע לא פחות ממנה

This is porn for AI engineers... Rubin Ultra #gtc26

Jensen just rolled Vera Rubin onto stage . 40 million X performance in 10 years . 100% liquid cooled with 45c hot water . Groq 3rd gen in volume production #gtc26

At #gtc26 . Unstructured data is the context for AI, says Jensen .

@UMD_PSC382 @kiaran_ritchie Disagree. Privacy laws like GDPR require data export. If your history, preferences, and memory can transfer, the emotional lock-in disappears.

@aron_brand @kiaran_ritchie I’ve seen the exact counter argument also put forward: that these models will know you so well and you will establish a personal connection with your chatbot to the point where the costs of switching for your well-being is so high that you’ll be locked in (Econoclasts podcast)

I don't see how Anthropic, OpenAI or any of the model providers have any hope of defending their moats. And consequently, I think they're going to get wiped out. Right now, in early 2026 they have a meaningful advantage in terms of model capability. But far cheaper and open source models are not far behind. How long can they maintain a meaningful advantage? For the vast majority of use cases, we don't actually need much higher intelligence. It doesn't take 140 IQ to automate Turbotax or powerpoint. Eventually we will be saturated in cheap, local models that are "good enough". Of course some scientific labs and frontier research will always want the latest and greatest. But that market is orders of magnitude smaller than these company valuations can justify. What am I missing?

People aren’t exactly fond of Microsoft Copilot (to say the least). Meanwhile, users can’t get enough of Claude Code and Cowork. That’s what makes Microsoft’s latest move interesting. Instead of fighting Claude’s momentum, Microsoft is leaning into it. Microsoft isn’t trying to win the model race. Right now that race is burning through money with no clear path to profit. The real goal is to control the profitable layer above the models. If Microsoft owns the workflow, the interface, and they are the undisputed leaders in enterprise distribution, the underlying LLM matters much less. In that world, power shifts away from the AI labs and toward the platform where the work actually happens. And that platform is Microsoft Office.

ChatGPT with the 5.4 thinking model just spent 24+ minutes on a single prompt and created a full, well-designed PowerPoint presentation, complete with images, ready to download as in .pptx format. That’s not “answering a question”. That’s doing the work!