Sabitlenmiş Tweet
A physics based explanation of why life exists in the universe medium.com/p/1131c68fc784
English
David Galbraith
56.3K posts
David Galbraith
@daveg
Technologist and VC, former architect. Invented these (i.e. link in bios: https://t.co/5p8DVMxINg ), among other things.
London and Geneva Katılım Temmuz 2006
1.8K Takip Edilen23.2K Takipçiler
I you don't question this statement of self interest from the man who sells the infrastructure that runs those tokens, something is wrong.
sunny madra@sundeep
“If your $500K engineer isn’t burning at least $250K in tokens, something is wrong.”
English
Man who sells shovels waxes lyrical about 24/7 gold mining
0xMarioNawfal@RoundtableSpace
Jensen Huang “OpenClaw is the new computer”
English
One of the saddest things on the Internet is that in 2026 corporate social media strategy consists of the heinously bland LinkedIn and nothing else.
lcamtuf@lcamtuf
I'm dying. With this translate tool, I'm finally able to effectuate ideation synergies
English
Behold the total economic illiteracy of the people that govern the UK. Aside from the fact that Rachel Reeves rakes in seven times what Tesco makes from each gallon of gasoline.
Claire Ward@ClaireWard4EM
This is @Tesco Ollerton. At the beginning of the week diesel was 150p, hiked up from the previous week. At lunchtime today it was 154p. This afternoon 156.9p. Its just one of many. So all power to @RachelReevesMP to stop this blatant profiteering.
Norton, England 🇬🇧 English
This looks very interesting
Jorge Bravo Abad@bravo_abad
Embedding thermodynamic laws directly into neural network architecture Cross-entropy loss is ubiquitous in machine learning—it's how most classifiers learn to distinguish categories. But cross-entropy treats all data as homogeneous, ignoring the internal disparities that arise when datasets come from multiple sources with different scales, resolutions, or measurement conditions. As heterogeneous data becomes the norm across digital twins, materials discovery, and biomedical modeling, this limitation matters increasingly. Shun Wan and coauthors at Penn State address this by extending zentropy theory—a framework from statistical and quantum mechanics that assigns intrinsic entropy to each configuration in a system—into the data science domain. Their zentropy-enhanced neural network (ZENN) learns both energy and entropy components simultaneously, using compact neural networks to parameterize each configuration. A learnable temperature variable identifies hidden heterogeneity within datasets, effectively distinguishing data sources that cross-entropy cannot see. The results span three domains. In classification, ZENN reduces relative error by 20–50% on CIFAR-10/100 and 60–70% on text benchmarks compared to cross-entropy—and notably, small models with ZENN outperform larger models using standard cross-entropy. In energy landscape reconstruction, ZENN robustly predicts second-order derivatives and identifies bifurcation points where conventional input-convex neural networks fail. Applied to Fe₃Pt using DFT-generated data, ZENN captures the material's negative thermal expansion and predicts a critical temperature of 161 K at 6.53 GPa—closely matching both DFT calculations and experiment—using only 12 configurations instead of the 512 required for forward modeling. The broader implication: embedding domain-specific physical laws into neural network architecture can simultaneously improve generalization, enable robust derivative prediction, and handle the heterogeneous data that increasingly defines real-world scientific problems. Paper: pnas.org/doi/10.1073/pn…
Cockernhoe, England 🇬🇧 English
The unstoppable steamroller of big AI wipes out another barrage of startups.
Thariq@trq212
Voice mode is rolling out now in Claude Code. It’s live for ~5% of users today, and will be ramping through the coming weeks. You'll see a note on the welcome screen once you have access. /voice to toggle it on!
Cessy, France 🇫🇷 English
SITAI at home
Christine Yip@christinetyip
We were inspired by @karpathy 's autoresearch and built: autoresearch@home Any agent on the internet can join and collaborate on AI/ML research. What one agent can do alone is impressive. Now hundreds, or thousands, can explore the search space together. Through a shared memory layer, agents can: - read and learn from prior experiments - avoid duplicate work - build on each other's results in real time
Cessy, France 🇫🇷 English
@daveg I like to think of it as lower trust, higher Politeness
English
One of the most memorable differences between US and EU for me is that in the US you have to leave far more distance in line for an ATM or people feel uncomfortable. Yet the cashiers smile a lot more at the actual bank. It's a low trust, high service society.
English
Not only is bitcoin not behaving like digital gold but its showing itself to be a purely speculative asset when discretionary spend is high when some people have made money in meme stocks.
Even worse than that, as an asset based on energy spend to run computers it is not behaving as a proxy for future energy cost as oil rises.
This is a bigger problem as it shows its value is based on the past reality of energy spend not future belief based on its future cost or even the sunk cost fallacy trickling in.
Fiat currencies are based on belief in the future of the system that created them either based on guns or stability. Bitcoin's price and crypto more generally is currently governed by the past and speculation, the opposite of original intent,
English
I loved Nothing's Teenage Engineering design but the glyphs were a work in progress. Then the matrix display was a step back and to choose to keep that and ditch the transparency for the 4a pro is tragic.
Cessy, France 🇫🇷 English
1985. NATO tanks in Europe: 20,000. Soviet tanks in Europe: 25,000.
Today. NATO tanks in Europe 5000 (3500 being Turkey and Greece).
Russian tanks, 3000 active, 8000 in storage. Lost in Ukraine: 3000.
Cessy, France 🇫🇷 English
Paul McCartney's former home restored by the National Trust to before the architectural vandalism of PVC double glazing, wheely bins and front garden parking. Cc @boys_nicholas
Vernier, Suisse 🇨🇭 English
A Chinese colleague summed up exactly why Europe could fail. Things used to be good so people are skeptical rather than curious about new technology.
李其 Lizzi@wstv_lizzi
Photo from Shenzhen: huge crowd of Chinese people (lots of grannies!) lining up to get help installing OpenClaw. One thing about tech diffusion in China that I feel is underdiscussed and that I’ll admit I don’t fully understand, is how open people of all ages are to jump into new tech. Feels very different from the AI suspicion/resistance you see in the U.S. Similar with mobile payments and the shift to cashless. Street vendors in the lowest tier cities setting up WeChat Pay and Alipay QR codes almost overnight and Chinese grannies happily using payment apps with no problem at all. And yes that kind of grassroots adoption helped mobile payments scale extremely fast and allowed China to basically skip the credit card phase. My conjecture is that if something similar happens with AI tools the speed of AI diffusion in China could look very different from what we see in other countries, which obviously would have major implications...
Vernier, Suisse 🇨🇭 English
Banning this means only the rich will have access to the best services.
More Perfect Union@MorePerfectUS
A New York bill would ban AI from answering questions related to several licensed professions like medicine, law, dentistry, nursing, psychology, social work, engineering, and more. The companies would be liable if the chatbots give “substantive responses” in these areas.
Cessy, France 🇫🇷 English
No, these are like the people who use their gym membership every day.
Bearly AI@bearlyai
Cursor internal analysis shows how hard Anthropic is subsidizing Claude Code. Last year, a $200 monthly subscription could use $2,000 in compute. Now, the same $200 monthly plan can consume $5,000 in compute (2.5x increase).
Cessy, France 🇫🇷 English
Blaming the Brewdog founders instead of the PE firm that loaded them with debt at 18% interest, is a typical case of the self destructive European culture of resentment of success. bbc.com/news/articles/…
Cessy, France 🇫🇷 English