More importantly, tritium has problems as a fuel for a commercial reactor because it's not a naturally occurring isotope. It has to be generated, usually by bombarding lithium with neutrons. The only economic source of those neutrons is a nuclear reactor. Currently we do this with fission reactors, where tritium production is a minor byproduct. You can produce some tritium from a fusion reactor, but I can't see how you can get even close to one tritium per fusion reaction in any real-world design.

The net result is that it just isn't practical to use D-T fusion as the main energy source because we can't have enough tritium for that to work. We need excess energy from a reactor that uses primarily deuterium (of which there is plenty to fuel everything for millions of years) for fusion to be a practical power source for the whole world.

The same thing is true of any film print. The only time you see exactly what's on the film is projected slides from slide film. Any printed photograph involves a complex process to convert the original film image into what winds up on paper. Not to mention that a lot of changes can happen by careful choice of development process. Again you have an image that's based on the light that came in through the lens, but there's a complicated process for converting that light into the final image.

You should read Ansel Adams's Examples: The Making of 40 Photographs to learn about the complicated process involved in developing a printing a film photo. He spends a lot of time talking about developing and printing, which he saw as at least as much of the process of producing a photography as composition and exposure. He never distinguished what happened before he pressed the shutter as a distinct activity from what happened after. All those steps were essential to producing a photograph.

More generally, Amazon is destroying its value as a seller by acting as a marketplace. They've deliberately made it as difficult as possible to tell the difference between goods sold by Amazon and ones being sold by third parties using their system. That makes it easy for scammers to take advantage of the system. They could try to weed out the scammers, but they won't. They're worried that eliminating scammers could give some company that's less discriminating a chance to break into the market. I get the impression that Temu is already doing this. At the same time, I'm sure the scammers are good for Amazon's short term profits. Not their long term health, though; they're getting so worthless for so many things, people are going to go back to buying from individual sellers who don't have the same quality problems.

I think there are two possible failure modes. You're right about the danger of people accepting photos uncritically, but there's also a danger of people rejecting accurate information out of hand because they don't want to believe it. If there's a real worry about fake pictures, it just makes it easier for people to ignore any picture they don't want to accept. We need to do something to try to make news photos harder to fake to counteract that danger.

It's true the scammers used them because they're ambiguous, but courts are used to dealing with ambiguous communications. Lots of people who want to avoid legal consequences use deliberately ambiguous language in the hope of hiding their intent from the law. The court will usually let the triers of fact in the case determine what ambiguous things really meant.

I think the main use of this would be for news photography. The photographer would send the signed, unedited photograph to their employer (or sell the signed, unedited photograph if they're a freelancer), with the signature serving as evidence the picture was unaltered. The employer/buyer would then crop, color correct, etc. the image for publication, but would still have the signed original to serve as proof the image was legit.

This also explains how they'll deal with a lot of the problems with the system people have pointed out. This isn't a 100% perfect solution to catch fakes. It's a system intended to keep people honest. If a news photographer gets caught faking pictures, their career is on the line. This is intended to increase the chance they'll be caught. The stuff about AI fakes is there because AI is a hot topic that gets attention, not because it's the primary use case.

I don't think this is a huge problem as long as you make the keys specific to each camera. The camera owner can then fake pictures by extracting the key from their camera and using it to sign fakes, but the images are still traceable to their camera. They could also create a signed fake by printing it at sufficiently high resolution and photographing it, which you can never stop. It's not a perfect system, but it substantially raises the bar for creating fakes. The important point is to make fakes traceable to the specific camera so the camera owner can be held accountable. If the camera manufacturer is responsible for public key distribution, they can even revoke the key of a camera that's been used to make fakes so everyone will know they can't trust those pictures.

The whole point of good cryptography is that the algorithm can be widely known without the system being compromised. I'd have to look at their implementation, but this is probably a fairly standard digital signature system. Basically, the camera takes a hash of the image and encrypts it with a private key. Anyone can decrypt the hash with the published public key and compare it to a hash of the file they received. If the two are identical, it's evidence the image hasn't been tampered with.

Public key crypto and hash algorithms are all well publicized and have been thoroughly tested. It's possible they'll be cracked eventually- we've gone through a whole series of hashing algorithms as old ones were cracked- but for the purposes of verifying news photographs that's not a huge problem. The biggest risk is of faking pictures of current events, and news photographers can always get new cameras or updated firmware that replaces vulnerable algorithms with up-to-date ones.

This won't completely eliminate the possibility of fake photographs. There are two obvious methods. The low-tech approach is to print the fake image and take a photograph of it. You would have to print it at high enough resolution to hide any printing artifacts- you don't want someone seeing the dots from your printer in the final photo- but it's not that hard. Another method is to extract the private key from the camera and use it to sign your fakes. Either one will only allow you to fake photos from a camera you control, since the keys will be specific to the individual camera.

The part that's at all new for this application is where to store the digital signature. It's not a hard problem- the logical place to store it is the EXIF header- but it would require agreement between the camera manufacturers about how to standardize it. There's probably some trick just because you probably want to sign the metadata in the EXIF as well as the image data, and adding the signature to the EXIF changes it from the signed version. Even that should be solvable by stripping the signature from the EXIF before taking the hash.

LLMs do not make verbatim copies of copyrighted works that are them reproduced without permission

Except this isn't true. The article includes cases where ChatGPT would reproduce NYT articles verbatim with the correct prompts. For example, you could ask it to give you the first paragraph of such and such an article, and it would do so. Asking for subsequent paragraphs would get it to reproduce the whole article.

Additionally, if I or LLMs falsely attribute facts to the Times, there may be a violation of slander or libel laws. Without ill intent, a simple retraction of the statement is satisfactory.

That's not the way the law works. The kind of lawsuit you're talking about is supposed to correct a single big statement. If there's a pattern of misbehavior, the law allows the wronged party to ask for an injunction, where the court orders the wrongdoer to stop what they're doing. It's hard to see how you could do anything else with something like ChatGPT, since each instance will repeat the misbehavior of each other instance. It shouldn't be up to NYT to play Whack-A-Mole and find every instance of ChatGPT claiming its hallucinations are supported by NYT articles, especially since most uses of ChatGPT don't produce any public trail.

Maybe OpenAI bought a subscription and scraped their site, which is probably a violation of their TOS. More likely someone they get information from bought a subscription and passed it on as part of their sales. I expect this kind of argument to be one of the legally stronger ones against AI companies. If the AI companies are breaking copyright law in assembling their training sets, the copyright holders have every right to sue.

I also expect the misatribution part of the suit to have some legs. Claiming misinformation comes from NY Times when it was just made up by your chatbot is serious misconduct. It's not quite as bad as one of the bots slapping the Getty Images watermark on fake photos, but the courts are unlikely to look kindly on it.

I certainly think more companies should be trying genuinely novel stuff, but Hyperloop wasn't. People have been looking at ways of running trains in a vacuum for a long time, and every time they've looked, they've concluded it just won't work. I guess someone needs to come back and reevaluate that conclusion once in a while, but it should come as no surprise when it remains valid.

Meanwhile, conventional high-speed rail has decades of successful operation. The problems HSR faces in the USA are political, not technological, and new technology was never going to solve those political problems.

Part of the reason the US doesn't have a strategy for its bitcoin is because it isn't acting like a single entity. Instead, each agency that has bitcoin has them for specific reasons and will decide what to do with them based on those reasons. Maybe the IRS will decide to sell its bitcoin strategically in an attempt to maximize the money the federal government gets for them, while the Secret Service will return bitcoin to the people they were stolen from, and DOJ will liquidate bitcoin as each individual case involving them concludes. Meanwhile, CIA will distribute bitcoin to its assets to fund ongoing operations, but will be careful to buy them on the open market rather than get them from other parts of the US government to maintain secrecy. That's what happens when things just kind of happen rather than being the result of some overriding strategy.

The US has plenty of extradition treaties that allow either country to demand the arrest and transfer of someone in the other country. It's quite normal for ordinary criminals; high profile criminals will always be a mess, though. There are also some limits. Most countries will refuse to extradite people for offenses they don't consider to be crimes, e.g. the US won't extradite people for what we view as free speech. They can also refuse extradition if they believe the person will be mistreated by the judicial system, which includes countries without capital punishment not extraditing people who might be executed. And, of course, the person who might be extradited can always challenge their extradition through the legal system.

Absolutely. What's interesting here in the LA area- and I assume in other metro areas- is that the older areas that were build around public transit are still very walkable. City layouts are conservative, so neighborhoods built before cars were common still have the necessities of life within easy walking distance of the homes. Home prices there tend to be expensive because people love the lifestyle, but despite that we've written zoning rules that prevent building new neighborhoods like that. It's maddening and really stupid.

in most of our cities it's been cheaper to build outward than upward, so you get sprawly abominations that don't lend themselves to mass transit.

Except that isn't really true. Consider where I live: the Los Angeles metro area. It's considered to be the classic example of car-dependent sprawl, but we used to have a first-class public transit system before WWII. The Pacific Electric Red Car system went everywhere in the LA area, and there were street cars for local transit to get you to and from the Red Car station. The population density has gotten much higher since the Red Cars' heyday, which should make the area more amenable to mass transit, not less. So why did our mass transit system decay even as the city became theoretically more amenable?

It was a public policy choice. When we built new housing developments, we connected them to the city by roads but not by mass transit*. We spend a huge amount of money building freeways that could have been used to expand mass transit instead. We actually destroyed existing neighborhoods to provide a place for the new freeways, which require a lot more space than mass transit that can carry the same number of passengers. We decided we wanted cars rather than mass transit, so that's what we got.

*The Pacific Electric system was actually built as an adjunct to real estate development. The developer (Henry Huntington) knew people wouldn't move into his new housing developments if there were no way to get to and from them, so he built mass transit as part of the original design. If he had been building a generation or two later, he probably would have built roads instead.