The AI bubble has reached its ‘fried chicken’ phase ft.trib.al/VGyrH8K
31.10.2025 20:07 — 👍 38 🔁 13 💬 3 📌 3@anystring.bsky.social
The AI bubble has reached its ‘fried chicken’ phase ft.trib.al/VGyrH8K
31.10.2025 20:07 — 👍 38 🔁 13 💬 3 📌 3I wanted to offer some thoughts on the Gates climate memo that has been circulating this week. While I can't directly speak for others, I can say that my own response is one of dismay & deep frustration (and that this view is shared by many climate/Earth scientists). [1/n]
30.10.2025 17:02 — 👍 1007 🔁 383 💬 23 📌 71Seems it had enough Nature links in the training data to learn the structure, but it didn't learn the exact links, and instead fills in the digits randomly. A simple explanation how these LLMs work is www.youtube.com/watch?v=wjZo...
30.10.2025 12:41 — 👍 1 🔁 1 💬 0 📌 0»Das ist eine Revolution: China überrollt die Welt mit Erneuerbaren«
»Teile der Union wissen aber nicht, wie sie moderne Energiepolitik machen sollen, ohne zu "grün" zu klingen. Also verkünden sie aus Trotz lieber das Gegenteil«
www.n-tv.de/wirtschaft/D...
You map from a domain of 4 values into a range of 5 possible values. Requiring that the map is ascending is equivalent to counting different orderings of the values only once. So these are essentially combinations with repetition, and their count is (n+k-1 over k) = (8 over 4) = 70.
28.10.2025 11:20 — 👍 1 🔁 0 💬 1 📌 0"Strömavbrottet sammanföll med en planerad uppdatering av Ellevios tjänst för att kommunicera om avbrott" www.svt.se/nyheter/loka...
27.10.2025 22:33 — 👍 0 🔁 0 💬 0 📌 0Apparently they are still in use, sort of. The latest printing was done from 1963 to 1968: web.archive.org/web/20090506...
25.10.2025 18:39 — 👍 0 🔁 0 💬 0 📌 0when I visited korea last year I went to haeinsa temple to see the tripitaka koreana, a set of 80k+ wooden printing blocks for buddhist scriptures, made in the 13th century en.wikipedia.org/wiki/Tripita...
25.10.2025 16:46 — 👍 1336 🔁 171 💬 20 📌 20Startlingly good effects of LNP-mRNA vaccines when given to cancer patients receiving immune checkpoint therapy. More vaccine research please. More mRNA!
23.10.2025 18:47 — 👍 63 🔁 24 💬 0 📌 1What is more, at the Nordic Tipping Week I‘m currently attending we learned that climate models that agree best with observations are the ones where #AMOC declines most by 2100 (as yet unpublished work).
23.10.2025 10:47 — 👍 95 🔁 13 💬 3 📌 0A Rössler attractor rendered with Manim. Two initial states with a distance of 0.0025 diverge over time. #mathsky
22.10.2025 19:32 — 👍 4 🔁 0 💬 0 📌 0Nexperia warns customers over chips made in China: FD www.dutchnews.nl?p=258863
22.10.2025 07:51 — 👍 1 🔁 1 💬 0 📌 1Very interesting article discussing what might lurk behind the Nexperia controversy.
bits-chips.com/article/lets...
Today I learnt that the significant portion on the internet that relies on AWS also happens to be the significant portion of the internet I seemingly avoid.
20.10.2025 09:08 — 👍 57 🔁 2 💬 3 📌 1“The U.S. military demonstration that shot live-fire artillery rounds over Interstate 5 on Saturday dropped metal shrapnel on a California Highway Patrol vehicle…”
https://www.latimes.com/california/story/2025-10-19/u-s-military-live-fire-demonstration-dropped-shrapnel-on-chp-vehicle
The floating French Tara Polar Research Station in Reykjavik harbor.
17.10.2025 20:30 — 👍 135 🔁 13 💬 1 📌 0I can just about cope with the idea that the amount of CO2 in the atmosphere has hit a new record -- it's what one would expect even if we were rapidly decarbonizing. But that the *rate of increase* has hit a new record is unbelievably depressing.
www.theguardian.com/environment/...
this is one of the best articulations of the AI bear case I've read
most of the AI companies are also massively wealthy and profitable hypercorps, but OpenAI specifically is a bomb planted at the heart of the American stock market helpfully labeled "this is a bomb"
prospect.org/power/2025-1...
Quick image created with Google Gemini on AI profitability model
Kun AI:ta pyytää visualisoimaan tätä epäpyhää kokonaisuutta, sohaisee se kyllä asian ydintä, mutta hermostuksissaan sitten typottaa urakalla. En siis edes promptannut näitä Compute Credittejä tai tätä toista kiertotaloutta 😅
12.10.2025 19:22 — 👍 4 🔁 1 💬 0 📌 0Now this is awesome! #hamr #hanradio
12.10.2025 10:29 — 👍 9 🔁 1 💬 4 📌 0Seems to have changed a bit lately. www.economist.com/britain/2025...
09.10.2025 18:39 — 👍 0 🔁 0 💬 0 📌 0Flying cars are the perfect example of a point solution: trying to solve a systemic problem (traffic) with an individual product (fly over the traffic).
But traffic is not a technology problem; it's a social problem. Remote work, congestion fees, and dense transit-connected housing solve it better.
Today is the 89th anniversary of The Battle of Cable Street when the people of the East End of London halted the march of Oswald Mosley’s Blackshirts through Stepney.
04.10.2025 15:08 — 👍 864 🔁 267 💬 18 📌 15I think for me the most compelling answer for "why fund basic research?" (and the one most relevant to the people doing the work) is that humans are curious and finding stuff out makes us happy and fulfilled. Science is a thing humans like. Life would be more dull and sad if we didn't do it.
end/🧵
Every few months now I re-read this "Who Goes Nazi?" piece from 1941 and am blown away by how it captures the people we are dealing with 80 years later.
harpers.org/archive/1941...
The base frequency is 14107.5kHz, and a search for that finds www.theskywaves.net/pactor.html . Maybe the call of a PACTOR modem?
27.09.2025 08:31 — 👍 0 🔁 0 💬 0 📌 0Eher deutet dieser Aktionismus auf Panik, man will mE den W unter Druck setzen, die Unterstützung der UA zu reduzieren, bevor die UA Angriffe auf die RU Infrastruktur und die zunehmenden Wirtschaftsprobleme zur Katastrophe führen
Die Lage in RU dürfte schlechter sein als von außen sichtbar
4/4
National Academies of Science panel finds that: The EPA was right in 2009 (when it found that climate change driven by society’s emissions of greenhouse gases are endangering human health & lives), and that everything we've learned since has only made it more right.
arstechnica.com/science/2025...
I initially missed the offset one, because the numbers on dice don't start at zero (and then I also missed the solution with the all-ones die). Starting to count at one always complicates things.
18.09.2025 08:59 — 👍 1 🔁 0 💬 0 📌 0One die with two each of 0, 1 and 2, and one with three each of 1 and 4.
Two of 1, 2, 3 and three of 0, 3.
Two of 0, 2, 4 and three of 1, 2.
Two of 1, 3, 5 and three of 0, 1.
And the trivial solution with one regular and one empty die.