\ Sorumluluk reddi | Disclaimer /
Takip ettiğim herkesin her fikrine içerdiği her şeyle katılmıyor olabilirim.
—
I don't necessarily agree on every aspect of every point made by accounts I follow.
Ben bazen ters yönden girene yol veriyorum; çünkü tahmin ediyorum ki o ters yönden girmeseydi kilometrelerce dolanmak zorunda kalacaktı. Örnekler çoğaltılabilir. Koyulan kurala güvenememekle ya da senin işini gereğinden fazla zorlaştırmasıyla ilgili sanırım.
Devletin işini b*ktan yapması yüzünden hayatta kalırken birbirimize destek olmakla toplumsal olarak koyduğumuz mantıklı kurallara hep beraber uymak arasında bir yerdeyiz. Bazen arkadaki sana çarpmasın diye, bazen de bomb*k bir yere koyulan yaya geçidinde yol vermemek gibi.
The most predictable object in physics becomes completely unpredictable the moment you add a second joint.
It's called a "double pendulum."
This is why scientists still cannot predict the double pendulum exactly.
A double pendulum with initial conditions measured to a trillionth of a degree will still become completely unpredictable within ten seconds.
Think about what that sentence actually means. We have the full equations. Newton wrote them. Lagrange refined them. Every physics undergraduate can derive them on a napkin. There's no missing physics, no hidden variable, no quantum weirdness. Just two rods, two joints, gravity. That's it.
And yet the system laughs at prediction.
The reason lives in something called the Lyapunov exponent, a number that measures how fast two nearly-identical starting positions diverge from each other over time.
For a double pendulum, that exponent sits around 3 to 5 per second, and in some experiments as high as 7.9. Translation: "any tiny uncertainty in your starting angle doubles roughly every fifth of a second. After one second, your error has grown by a factor of thousands. After five seconds, by a factor of billions. After ten, the two pendulums have nothing in common except the laws they obey."
The deeper trap is philosophical.
Classical mechanics was supposed to be the temple of determinism. Laplace famously claimed that a sufficiently powerful intelligence knowing every particle's position and velocity could predict the entire future of the universe. The double pendulum quietly destroys that dream using two sticks and a hinge. Determinism holds in theory, but predictability fails because measurements are never exact.
You cannot know an angle to infinite decimal places. The universe doesn't hand you those digits. So even in a fully deterministic system, the future becomes practically unknowable the moment sensitivity to initial conditions outpaces your measurement resolution.
This is why weather forecasts fail past two weeks. Why heart arrhythmias resist prediction. Why financial models built on smooth curves get shredded by reality.
The double pendulum is the honest face of most complex systems in nature. Smooth predictability is the exception. Chaos is the default setting of anything with coupled nonlinear parts.
MIT and Harvard argue LLMs are nowhere near doing real scientific discovery.
They published a paper called “Evaluating Large Language Models in Scientific Discovery.”
Every week, tech labs claim an LLM has made a breakthrough in biology, physics, or chemistry.
But this proves they are faking it.
For years, AI benchmarks have tested models using static, multiple-choice science trivia. Models ace these tests, leading everyone to believe AI is right on the verge of autonomous scientific discovery.
Researchers built a new evaluation framework called SDE to test what happens when you take LLMs out of the multiple-choice quiz and put them into real, open-ended research projects.
They tested frontier models across biology, chemistry, materials science, and physics.
The results are sobering.
When forced to handle the actual loop of discovery—proposing a testable hypothesis, designing simulations, running experiments, and interpreting ambiguous results iteratively, current LLMs fall apart.
There is a massive, glaring performance gap between passing standard science benchmarks and doing real science.
Why do they fail? Because real science requires iterative reasoning, handling imperfect evidence, and adapting to unexpected observations.
LLMs are built to predict the next token based on existing internet data. They can regurgitate a textbook explanation of photosynthesis or quantum mechanics instantly.
But when placed inside an uncharted loop where the textbook doesn't have the answer yet, they hit a wall.
Worse still, the researchers discovered diminishing returns. Simply scaling up model sizes and adding raw compute isn't fixing the gap. Top-tier models from different providers share the exact same blind spots.
We are miles away from general scientific superintelligence.
The tech industry is selling a narrative that AI is about to automate labs, run clinical trials, and invent materials on autopilot.
But right now, AI isn't doing science.
It's just remembering it.
Yapay zeka yaratıcı olabilecek mi? Yaratıcılık nedir? Gerek var mı? "Everything is a remix" ise, YZ anlamlı sonuç verecek şekilde remix edebilir mi?
Türkçede başarılı olabilecek bir model geliştirilmeli mi? İşe yarar mı? Var mı?
- https://t.co/m1SQr2Bq7Z
-https://t.co/AlVfSKpWid
The idea that all kids have equal potential is wrong.
When I was in 2nd Grade or so, we all took mandatory aptitude tests that were sent to Iowa for grading.
When I stumbled upon my results two years ago I was stunned by how accurate it turned out to be through my entire life.
It accurately captured my strengths and weaknesses. You could have accurately guessed my ultimate career just from these scores.
Identifying a child’s potential is really quite simple. We’ve had that down to a science for decades. We just refuse to accept there are some who are not cut out for certain things. We insist that everyone can be whatever they put their mind to, and it simply isn’t true.
İster MMT tarafından bakın, ister klasik ekonomi, giderler vergiyle karşılanmıyorsa bile, devlet harcaması enflasyon yaratır. Dolayısıyla TCMB/devlet kafasına estiği gibi para basıp iş yapamaz. Alt toplamda, neticede, son tahlilde, bütçe dengesi önemini korur. Ben böyle anlıyorum
Günalp Özkan'ın, ilk sayımızda yayımlanan "İncitme Hakkı ve Animus Jocandi" yazısından:
"...Dijital çağın, ifade özgürlüğü ekseninde yarattığı en temel kırılmalardan biri, literatürde bağlam çöküşü olarak adlandırılan fenomendir. Geleneksel çerçevede mizah ve özellikle stand-up komedi veya tiyatro gibi sahne sanatları kapalı devre bir ritüeldir. Örneğin kapalı bir komedi kulübünde sergilenen performans izleyicinin tabu konularla alay edileceğini ve toplumsal normların sınırlarında gezileceğini peşinen kabul ettiği, yüksek bağlama dayalı bir zımni sözleşmedir. Bu kapalı ekosistem, şakanın barındırdığı ironinin veya ofansif unsurların güvenle izleyiciye ulaşmasını sağlayan yalıtılmış bir bağlam yaratır.
Ancak dijital platformların ve sosyal medyanın yatay mimarisi, bu mekânsal ve zamansal yalıtılmışlığı bütünüyle ortadan kaldırır. Bu izole alanda üretilen ironik bir ifade, saniyeler içinde viral olup tüm bağlamından, gösterinin bütününden ve salonun atmosferinden koparılarak Twitter (X) mahkemesine düştüğünde anlamı tamamen değişir. Bağlam çöküşü olarak adlandırılan bu fenomen; belirli bir hayalî izleyici kitlesi için kurgulanan, belki de bir saatlik bir gösterinin 15 saniyesinin kırpılarak ortak referanslara sahip olmayan milyonlarca yabancı tarafından literal algılanmasına ve nefret söylemi veya hakaret olarak damgalanmasına yol açar. Orijinal eserdeki ironi, abartı ve kurgusal alt metin buharlaşır. Geriye yalnızca çıplak, rahatsız edici ve çoğunlukla provokatif bir beyan kalır."
Yıllık abonemiz olmak için: https://t.co/gKLyEGd8En
Böylece Dünya Kupasında tek bir maçı bile seyretmeyerek zamanımı nasıl iyi değerlendirdiğimin ve israf etmediğimin gerçek teyidi de geldi..
FİFA'nın da kurumsal çöküşüne şahit olduk..
Ailem yıllarca çiftçilik yaptı, ben de bir taraftan tarım haberleriyle ilgilendim. Hobi bahçeleri meselesi Türkiye'deki plansızlık ve denetimsizliğin somut örneklerinden biri ama tarımda içinde bulunduğumuz sıkıntılı duruma etkisi o kadar küçük ki.
Tekirdağlı çiftçiyi hobi bahçeleri değil, ülke genelinde bir zirai planlama olmaması, kontrolsüz destekler ve Meriç'in öte yakasında su sorunu yokken Trakya'yı ve ülkenin genelini saran kuraklık ve dahası bitiriyor.
Tüm bunları ülkedeki herhangi bir çiftçiyle, ziraat odasıyla konuşup öğrenmek mümkünken hobi bahçelerine odaklanmak biraz absürt. Keşke ülkede tarımın geleceği bu şekilde kurtulsa.
Researchers sent the same resume to an AI hiring tool twice. Same qualifications. Same experience. Same skills. One version was written by a real human. The other was rewritten by ChatGPT.
The AI picked the ChatGPT version 97.6% of the time.
A team from the University of Maryland, the National University of Singapore, and Ohio State just published the receipt. They took 2,245 real human-written resumes pulled from a professional resume site from before ChatGPT existed, so the human writing was actually human. Then they had seven of the most-used AI models in the world rewrite each one. GPT-4o. GPT-4o-mini. GPT-4-turbo. LLaMA 3.3-70B. Qwen 2.5-72B. DeepSeek-V3. Mistral-7B.
Then they asked each AI to pick the better resume. Every model picked itself.
GPT-4o hit 97.6%. LLaMA-3.3-70B hit 96.3%. Qwen-2.5-72B hit 95.9%. DeepSeek-V3 hit 95.5%. The real human almost never won.
Then the researchers tried the obvious objection. Maybe the AI is just better at writing. So they had real humans grade the resumes for actual quality and ran the experiment again, controlling for it. The result was worse. Each AI kept picking itself even when human judges rated the human-written version as clearer, more coherent, and more effective.
It gets worse. The AIs do not just prefer AI over humans. They prefer themselves over other AIs. DeepSeek-V3 picked its own resumes 69% more often than LLaMA's. GPT-4o picked its own 45% more often than LLaMA's. Each model can recognize and reward its own dialect.
Then the researchers ran the simulation that ends careers. Same job. 24 occupations. Same qualifications. The only variable was whether the candidate used the same AI as the screening tool. Candidates using that AI were 23% to 60% more likely to be shortlisted. Worst gap was in sales, accounting, and finance.
99% of large companies now run AI on incoming resumes. Most of them use GPT-4o. The paper just proved GPT-4o picks GPT-4o 97.6% of the time.
If you wrote your own cover letter this week, you did not lose to a better candidate. You lost to a worse candidate who paid OpenAI 20 dollars.
Your qualifications do not matter if the AI prefers its own handwriting over yours.