You inadvertently bring up a good point. If this is actual AI rather than "AI" they're creating AIs that are wildly misaligned, which is on the same level as creating new kinds of flu viruses with CRISPR

And before someone jumps on it, by "real person" I mean in the form he is presented, not that he didn't exist or wasn't black.

People arguing this is acceptable are using a motte and bailey. The bailey is that this is a real person. When shutting down factual arguments proving this wrong fails (and discrediting Wikipedia and its woke bent in the process) they switch to the motte that it's okay for the game to take this liberty with history because it's just a game.

Allow me to counter the motte. Yes, it's okay to make any kind of game you like with any kind of thing in it you like. But when you do so for a blatantly political reason following a long series of blatantly political choices in an attempt to rewrite history that acts in concert with many other attempts to rewrite history in the same way, I'm going to call you out on it and a lot of people are going to hate you and your game. That's not because I'm racist. It's because I hate lies, the rewriting of history, and your political agenda.

The fact that parent comment was modded down says a lot about Slashdot moderation. He makes a valid point and rather than argue it, you try to hide it. Shame.

Okay, can you link to even one fully automated refueling or recharging station anywhere in America? How about plans? Is there one being built anywhere in America? How do you figure one will be planned and built in the next few years given that the existing ones are very much in the research stage, let alone enough for general long haul trucking?

Yup, fully automated long haul trucking is decades away. They need a whole lot more than self-driving trucks. They need weigh station automation, they need to figure out security, minor contingencies like blowing a tire, major contingencies, getting the paperwork good enough that it can be automated, including purchase and sales papers that currently need to be signed off by humans for security reasons, docking by voice, many ways a truck can accidentally drive on non-truck routes, etc etc.

That said short haul point to point routes also are a bit disruptive when automated, so it's something.

We should value flashes of humor and personality like this one. Life is already hard enough; rigidly enforcing seriousness, especially over something as harmless as a cartoon dragon is, IMO, harmful to society at large. This wasn't an attack on the court's dignity; it was likely an attempt to inject humanity into the process.

The human spirit requires a chance to express itself, to add a touch of personality and maybe even some humor to the everyday. A quirky dragon logo that makes something more memorable is just that sort of thing. A demand for strict conformity --even within a formalized process -- is a demand for a dull, gray world. We should champion these sparks of character that reflect the diverse tapestry of human nature, rather than striving for an environment stripped of all personality and surprise.

This is laudable. As a gamer I would far rather have a studio put out something decent in a year or two than put out whatever it has now and patch later.

Weight isn't enough. To be fair you need to tax by the weight/mile.

Roads have long been subsidized by a tax on gasoline. As that goes away something has to replace it: the roads must, if not roll at least be repaired. If not a tax on cars then what? The interstate system at least must be funded by the federal government somehow, and the states will need to fund their networks. Eliminate the gasoline tax for one thing, it no longer serves its purpose of balancing miles with cost. For the states implementing a registration charge by car milage may be the best way. I don't like the idea of tracking total miles driven, but it's fair. It can also finally allow a proxy for road wear based on weight. Those behemoths people drive are harder on the roads and should be taxed accordingly. So for states maybe a registration fee based on car weight and miles driven.*

As for the federal government a sales tax on new tires would do the trick, based on weight capacity of the tire it would not only properly integrate the real cost of those 18 wheelers, pushing economics toward trains where they belong but fairly tax electric and non-electric cars alike. All without intrusive monitoring.

*Obviously cars would tend to be put up for sale before the registration is due, but that just pays the tax in the form of lower sales prices, with the rest of the money being paid for registration by the new owner. Even state-to-state sales arbitrage wouldn't be too hard to deal with.

Check out this site for some science fiction that is sadly being pushed by tech CEOs. It's not only beyond ethical boundaries to make this kind of claim, it's beyond all reason. But on Youtube and other social media with niches there are communities, fairly large ones, which actually believe the singularity is coming some time in the next five years based on these lies. There's a line between hyping your product to maintain fiduciary responsibility and outright fraud, and these guys crossed it still accelerating months ago.

Retail food margins are .5% to 5% for most items. Margins over the internet are higher, but this is an important distinction: the people most affected don't buy their food online.

"value" already includes quality.

> in principle, nothing would stop Claude or ChatGPT saying "I need to find out about relationships involving pizza and cheese", and the reasoner could tell it.

This is where you're going wrong. An LLM doesn't output questions like this unless prompted to, at which point you're performing multiple LLM queries for each whatever you're doing, and you're going to have to customize those queries, automatically or otherwise. Pretty soon you're really in the weeds. LLMs are inherently limited in this way.

I'm planning to implement retrieval augmented generation with a program using an LLM to augment in a way that should mitigate in-context forgetting. Does that count? I'm going to use a fuzzy vector search to find related terms and feed that to the LLM with the rest of my text.