This is why AI in the US is so hated. Next to slurping up jobs, its #1 perceived use that we see is disrupting communications, distorting news events, and changing the narrative. Basic stuff on a forum gets disrupted by trolls, or someone starts spewing the same bullshit rhetoric that has no base in truth, but is echoed in choruses, verbatim, drowning all chances of meaningful discourse, and driving people into creating ever more Discord servers.

It used to be that one solid post on USENET was as good as another. Now, with all the bots and AI slop, finding someone meaningful to interact with is next to impossible unless introduced by friends.

The edge cases are getting cleaned out. These are not the big ones, but the ones like parking, private lots, reflection and glare, and others. This may not bring around much in the way of news, but it is helping out.

Where I see self driving cars making an impact is how prevalent they are starting to become. I was surprised how many people are driving vehicles without a hand on the steering wheel in traffic these days, letting the vehicle do the work.

Of course, there is work to do, but it is getting there, and if it is successful, it will be a very positive thing.

I think the difference between people who -think- they are good drivers, versus actually being good drivers is a pretty wide gap. I don't know anyone who thinks they are a crappy or below average.

Probably the best way to tell is by the amount of wrecks and tickets, however, that metric isn't even good, especially if one is dealing with areas where speed limits vary, and the area has poorly designed roads. Put enough miles in, and statistically something will happen.

This is deliberate. However, just last week, Waymo got the go-ahead to allow their FSD taxis onto highways in Austin.

I have seen varying stats on this, and the numbers can be rigged. For accidents, is it a fender-bender, or does it result in deaths? One can say that only a few accidents are alcohol involved, while most deaths have a high chance of having alcohol as a factor.

Overall, FSD does drive better than people in most circumstances, but it requires a lot of edge case training. It is getting there. I think it is a good thing, because it has faster reactions than a person. There are also things that can be done (provided there is security placed so it can't be hijacked) for V2V communication or V2G which can make easier for autonomous vehicles to navigate and figure out the best lane positioning. FSD also is a good thing for people who don't really want to hang up their car keys because they have to go to work, but are worried about reaction times, and here in the US, the public transportation system is spotty at best.

I'm all for FSD. Just because it always pays attention and has faster reactions.

I agree with you -- even on antediluvian Linux kernels and kernel drivers, RISC-V boards to well, just because of the constant evolution. However, what the Raspberry Pi brings to the table is consistency. You can buy a Raspberry Pi from over ten years ago still, which is important for businesses who use them, where a board replacement is doable. No need to buy a bunch of stock and sock it away in a warehouse until all customer warranties run out.

I do wish RISC-V would get to a similar point, but it will get there, just because it has a lot of innovation behind it.

I'm beginning to think we need to do like VeraCrypt does as an option. Three algorithms. One optimized against QC, and two proven, solid ones that are somewhat resistant to that. Yes, it means three times the CPU to validate/sign the same stuff, but it protects us against catastrophic algorithm failure where one discover might reduce a keyspace to something trivial.

Downside of doing this is that some dedicated cryptographic processors have their die designed around the mathematical functions of that specific algorithm, and it can get expensive to design for multiples.

Thie key thing is accreditation. Ages ago, even ITT was accredited. When that was lost with the "show cause", the money earned for the degree becomes worthless.

Same with this AI school. You can teach what you want, but without having the seal of approval by some internationally vetted boards, the degree or certificates are not worth much, and we already have had students pummeled into the ground by one wave of student loan debt, so people are not really looking to go into the educational sector for anything other than licensed professions like engineering, architecture, finance, medical programs and law.

Bitcoin also has a bunch of regulators scrutinizing it. Combine this, with the pushback against data centers, be it AI data centers, or Bitcoin mining data centers, and it is expected to see BTC grow a bit, just because people can put money into it, perhaps as a long term "hodl" type of thing, but this doesn't mean it is going to hit the moon anytime soon. Especially if more data centers get hit by missiles and large swaths of Internet get blown away.

Even a blank VM is a no-no, especially with hypervisor escape exploits. Maybe a Bochs style VM, but that isn't exactly snappy.

The only way I could really do this is if I went with a VPC, ran their stuff on that, then powered off and nuked the VPC afterwards... but even then, there could be really bad stuff happening.

Instead of a ban, why not have an open source, social media app that is certificate based and allows each geographic region to dictate policies for the app on their turf, with the app using GPS coordinates and maybe other authentication methods to reduce spoofing? Yes, it would have boundaries in place (keep 18+ who are not family away from those below 18+, have geographic regions fenced off from each other), but why not offer something.

Outright bans don't help. However, having alternatives that do the same thing, maybe even better would bring better results. Otherwise, you will get a bunch of teens inventing new, decentralized, hidden P2P protocols to get around this.

Here in the US, we have been raised since days of the past of the machines turning on us. It could be Gort, AM, Skynet, GlaDOS, or many others. It is part of our national identity to fear terminators. Even our fiction, be it Star Wars, Star Trek, or Warhammer 40k [1] all fear robots and AI. In the real world, AI is (mis)used by businesses to take jobs, disrupt communications links via bots, elevating trolling to new levels, and obliterating the concept of an artist or musician. I was told by one businessperson that they wasted a ton of time with musicians, but with a Suno subscription, they can get "filler" music really easily and easily edit it. Art? Same thing. Just generate the art needed, and not worry about it.

China... completely different. AI has helped their economy by leaps and bounds, be it better manufacturing, better product design, better products, you name it. AI is viewed in general as a very positive tool, like an electric motor, or a battery for a cordless TCP/IP socket wrench.

These two different worldviews are why the US is wresting with AI like it is a bull out the gate, while China has it smoothly working behind the scenes.

With RHEL 10 adding more complex stuff like Lightspeed and immutable kernels, it is starting to have a lot more moving parts. RHEL used to be the production choice, but as time went on, there is a lot of oddball stuff, and oddball decisions made. For example, even though Fedora has btrfs, RHEL doesn't have btrfs, even as a -supported- option. (Yes, you can slap ZFS on it, but supported is the big thing).

Debian, on the other hand, doesn't have snaps, flatpaks, or other subsystems present by default. It "just works" out of the box. Plus, other systems like Proxmox can be installed on vanilla Debian with ease.

I would assert that because that fewer moving parts are needed, it makes things easier. You can take filesystem snapshots without Stratis, for example.

By making it harder for US companies to use AI, how does that harm China? Already, businesses are feeling the squeeze. I'm just wondering if this is regulatory capture on a higher level, ensuring the squeezes on hardware prices by US vendors can't be fixed by other countries.

Already, the RAM and SSD price hikes are not affecting other countries nearly as much as the US.

I think this might be a good idea, because getting a 4.0 GPA can be a lot different than actually mastering the material. However, the potential for abuse is present as well.

It would be nice though -- one could focus on material, not just trying to memorize via rote to copy/paste from memory to a test, then after the final is over, everything is forgotten.

It isn't perfect, but it can be tuned. Perfect is the enemy of the good.

I sort of wish it had some passive RFID abilities so it could work as a tag even when its battery is depleted. Even if the range is at most a meter, this might be useful if a device detects it as it goes past.