Comments here are interesting. Of course it’s a copy, he was 12. For pretty much of all of human history, to make art you would first consume large amounts of it to develop taste and then make many copies to develop skill and style. The average modern-day artist (writer, painter, poet, etc) does this far less than ever before, to the point where we have forgotten that’s how learning works.
In fairness to the US system, it’s certainly better than the European system or pretty much all but a few around the world. Yes, there is corruption, inefficiency and the largest subsidies are often for huge corporations that obtained them by buying politicians, but! The US government still manages to fund the cutting edge in 2026 in countless fields, to fund real American manufacturing, if you want to get grants you have a real shot at real money regardless of who you are, etc. In China you’re not getting a dime without the right political opinions. In Europe you have to be part of a very specific academic-professional class. In the US you can be anyone.
The thing about China is that they’re more strategic with their money and have longer timelines and clear, achievable visions. If you read the Wikipedia page for Made in China 2025 you’ll get the wrong impression that their success is due to more recent pushes; the vision is far more universal and has existed for far longer. You don’t get to the forefront of advanced manufacturing from nothing in ten years. Look at the 5th and 6th Five-Year Plans, into the seventh… you see the groundwork laid for present day China. The US rarely does that sort of long term thinking or planning these days, and it’s not even about the political winds changing or short-termism as much as that we lack one unified vision. Without that unified vision you can’t plan long term and you also can’t correct glaring problems. For example, if we had a unified vision on manufacturing, an obvious issue would be the lack of an American JLCPCB. You could create one with a stick and carrot approach, tariff assembled PCBs, new rule that any imported assembled PCB has to prominently display “electronics made in China”, smart subsidies for US board houses that encourage scaling and cost reduction. But that level of cohesion and vision rarely happens in the US and so we get a chaotic hodgepodge.
$2900 seems pretty reasonable to me considering the size. Works out to $416/sqft, which is much cheaper than Bay Area real estate.
I never understood the draw of these huge monitors until I had to do CAD for work and now I understand. Giant monitor + SpaceMouse is a gamechanger. My current monitor is 36” and I could easily use more width.
The IRA wasn't a cure-all and Trump/the OBBBA didn't exactly kill it dead, either. 84% of IRA clean energy grants, or $96.7B, was protected from clawback by Trump [0.] The tax credits are largely intact. Many of the projects were already completed and fully funded.
Go through a list of IRA projects, though, and you'll see two things: 1) they weren't generally killed by Trump, and 2) the IRA still did not get the US competitive with China. Let's take the largest one, the Toyota plant in NC, which is operational, and impressive in that it makes the batteries from raw materials BUT it has an eye-watering cost for the capacity ($~14B for 30 GWh/yr vs. $>4B for 30GWh/yr at Hyundai's plant.) Compare that to a Chinese plant at $50-100M/GWh and you can see that despite huge subsidies at several levels - including a $35/kWh subsidy for domestic cell production - the US is far behind here.
Look at the others and you'll see similar stories. Ford-SK On just split up because the F-150 is too expensive, demand is soft, and SK On wants to do energy storage. If we could make the F-150 competitively, demand would be higher, but it's incredibly expensive relative to a Chinese EV.
The more promising story is in solar (panel manufacturing and installs) but again, not a cure-all under Biden or a catastrophe under Trump. The US solar industry just had its third largest quarter on record, and we can now make every part of a solar panel in the US in volume. Module production capacity is at 60 GW, up 37% from Dec 2024, and cell production is at 3.2 GW, up from 1.2 GW a year ago [1.] That's a much prettier picture (regarding manufacturing) than in the EU, where manufacturing capacity is far lower and not growing nearly as quickly, although the EU has a larger installed base of (mostly Chinese) panels.
Well said. The credit is with the model; you commissioned it but did not create it.
With AI art... there is no passion, there is no pain, there is no emotion, there is no sex, there is no feeling, there is no reason. When Blaze Foley sang If I Could Only Fly or Nina Simone sang Stars or Bardot sang Je t'aime or Morricone wrote Se telefonando or Vermeer painted Zicht op Delft or Orozco painted his Epic of American Civilization or Maugham wrote Of Human Bondage or Stoppard wrote Rosencrantz & Guildenstern Are Dead or Cheever wrote The Swimmer there was a magnificent concentration of real feeling and a real reason that each of these things were made.
Could you imagine someone prompting a model, receiving the result, and then saying, as Cheever did about The Swimmer:
>It was a terribly difficult story to write. I couldn't ever show my hand. Night was falling, the year was dying. It wasn't a question of technical problems, but one of imponderables. When he finds it dark and cold, it has to have happened. And by God, it did happen. I felt dark and cold for some time after I finished that story.
---
To me, the reason for art is feeling, and the problem is that most things don't really provide feeling - if they do, it is a cheap and one-dimensional feeling. Almost all art and music and literature (and food, wine, architecture, poetry, photography, theatre, dance) that people consume today is _good enough_. It is correct, it satisfies. You listen to some hours of good-enough music on Spotify and the music is all correct and you come across "Chill77"'s AI-generated Papaoutai cover and you think that it is good. After all, it seems to have fooled a number of genuine Stromae fans. But the real function of art is not to satisfy. It is to reduce you to tears or silence or lust or anger or some beautiful cocktail of feeling. Of course, in the right context, with enough supporting factors, anything can produce emotion, but the best art needs little or nothing to make you feel. Bad art and good art are all around us, but the great is rare. That rarity is why people enjoy AI art: they forget the last time they felt, the AI is good, and that is enough.
The sad thing, of course, is that to make the great you must make a hell of a lot of bad and a fair amount of simply good art. And then there are those who have no delusions of grandeur but just make art for the sake of it. AI art cheapens those things; it makes them a trivial undertaking. The architect who would have become great on the completion of his two hundred and seventh building can now generate the first two hundred and six with the push of a button. The woman making fliers for her dance club - each one no great work of art, but certainly made with care and love, sees now that her work is useless and stops. We all lose.
>You can even combine the visual understanding of Gemini with the actions generated by Opus to take it a step further, by having the latter generate instructions and the former generates JSON DSL to that gets executed.
Yes, huge +1 for this. I do this in a different field and it's quite impressive. At first it felt weird using different models together but they really do have certain strengths/weaknesses (in January 2026.)
Also, fascinating how quickly things are evolving around PCB design. It was only six months ago that "vibecoding" a PCB was far off (look how basic this was: https://news.ycombinator.com/item?id=44542880) and now that may soon become possible!
In short: it plays far too well (~2500 ELO.) People think it originally played at a reasonable level and accidentally got more powerful as the seatback computers got more powerful; the same thing happened to the Mac chess app with the release of the M1.
That would be exceptionally sloppy development. Phones have had more than enough power for long enough. 4 core Skylake (Mac 2016) would be well beyond human capabilities, if it's just raw power.
The "thinking" (difficult) limit should be considered moves ahead, both depth and count. With a possible limit to time, if there is any time control.
> 4 core Skylake (Mac 2016) would be well beyond human capabilities
Not if the computer's time limit is set at 15 microseconds. It's not a question of whether the computers have "enough power"; just whether they are more powerful now than they were previously.
And yes, obviously that's a very sloppy and error-prone way to implement a difficulty control.
Even a computer from 20+ years ago will comfortably crush Carlsen, it really goes down to the specific engine used, chess engines have evolved a lot during the years.
Carlsen knows how to play anti-bot chess where some engines may struggle, but that only applies to amateurish engines.
> the same thing happened to the Mac chess app with the release of the M1
I fired up Chess shortly after getting an M1 and got destroyed a bunch of times. I thought that I was just extremely out of practice and quit playing for years. I guess it's better to find out late rather than never.
Precisely a Core Duo and a custom build with -O3 -ffast-math (a Chess engine doesn't requiere anything further from integers) and -march=$YOUR_CPU_THERE can yield crazy performance speeds without needing an m4 and a great match even for masters.
The diagram says that (Cyrillic ∩ Greek) - (Cyrillic ∩ Latin) is 3 letters, П Ф Г but as the sibling comment says, Λ/Л, Δ/Д and Κ/К are similar enough. That only leaves you with Θ/theta (th as in thin), Σ/sigma (s as in soft), Ξ/xi (x as in fox), Ψ/psi (ps as in lapse), and Ω/omega (o as in ore.) A lot of those are close enough that you can sort of guess, if you know the English names for the letters!
I don't think it's particularly American; people these days read at a lower level around the world. Language aside, most bookstores in Europe and the US have a fair bit in common.
The history of Soviet electronics manufacturing is fascinating, but there are some huge differences and I actually don't think the private sector is the largest. One is the pace and type of innovation. In the 70s and 80s the landscape was incredibly dynamic and technology went through several huge changes. If you wanted to run a clone of the US tech industry then, you would need a distributed, dynamic effort across many fields and not a top-down directed Manhattan Project. In 2025 we do have rapid technological change, but things are much more consolidated. In terms of strategically important recent innovations I can only think of EUV and AI. That's much more Manhattan-Project-able.
The other difference - which is even more significant - is that China is already far ahead in advanced manufacturing. The US was lightyears ahead of the Soviets in advanced manufacturing, which is what allowed us to win in the 70s and 80s. Now, we're so far behind it's not even funny. Sure, the West still makes some ultra-precise machines for EUV, but look where most of the components in those machines are made...
At the start of the microchip age, the US wasn't that far ahead. The techniques for manufacturing microchips weren't anything special and the Soviets could do so easily. The problem was the top-down mandate to clone, not lack of internal capability.
I don't know when you want to define as the start of the microchip age (50s? 60s?) but the Soviets were always behind the Americans in advanced manufacturing. In 1931 Stalin said "We are 50–100 years behind the advanced countries" [1.] In WW2 the Soviets relied on the US for advanced machine tools and in 1943 Stalin said: "The most important things in this war are the machines.... The United States is a country of machines. Without the machines we received through Lend-Lease, we would have lost the war." Khrushchev echoed this in his 1964 memoirs and there is a 1963 recording of Zhukov concurring. Now onto the 50s; the US invents NC machines, transistor manufacturing, chemical processing, miniaturization. The Soviets really only have the lead here in a few metal-related subjects: titanium production and rocket engine metallurgy. This is probably the closest the Soviets came; at this point they did indeed mass produce more tonnage of simple heavy machinery, but cannot mass-produce almost _anything_ cutting edge; look at their cars or consumer electronics of this time period. Onto the sixties; now the gap widens once more, the US pulls far ahead with ICs, microelectronics, CNC manufacturing, advanced alloys and composites, extreme precision manufacturing and precision manufacturing on a large scale. Look at the Saturn V vs the N1 or anything else in the aerospace/space industry at the time; the US was lightyears ahead. They could do things the Soviets could not even dream about. By the 70s the game was lost entirely; the US had advanced electronics manufacturing on a massive scale, laser cutting, even wider deployment of CAD/CAM, and was building the Space Shuttle. The Soviets essentially stagnated and were relegated to making cheap copies years later that looked similar but didn't work well in practice; see the Buran, TU-144, etc.
The reality is that the Soviets never really managed cutting-edge mass manufacturing. There are only a few countries that have: first Britain from the screw-cutting lathe in 1800, then the US, then later Japan, Germany, China, Taiwan and South Korea. In 225 years it's a fairly short list of countries, and most only made the list for a few decades!
The closest the Soviets ever came, of course, was aerospace, but to use that as an example of advanced manufacturing leadership would be a stretch. Sputnik was impressive, but advanced manufacturing it was certainly not; it was a simple sphere with a simple radio transmitter. Look at the tolerances, finishes, materials, manufacturing methods, etc of N1 vs Saturn V, Buran vs the Space Shuttle, TU-144 vs the Concorde.
reply