I think the only safeguard against that is to do your own moderation. The only tool I can think of that would even come close to detecting “power tripping” would be AI, and we all know that would end terribly. The best you can do is remove incentives (somehow, idk) so mods aren’t rewarded for a bad ban.

I had a Samsung smart watch, and it was honestly hard to use without an account. Half the features are locked behind signing in. I stopped using it when weather updates stopped working without logging in.
Now I have a dumb mechanical watch, and it’s glorious.

I’ve been on EndeavourOS for the last year or so, and I have to say it’s been rock solid. The only issues I’ve had have been around display scaling, since different programs and games handle it differently. Some things work better in X11 mode, and others better in Wayland. It’s weird some games report a weird 1706x960 resolution with 225% scaling on my 4K TV.

HDR and variable refresh rate just worked out of the box though!

I’m curious what kinds of projects you’ve seen firmware written for. The YouTube projects I’ve seen usually don’t require more than a handfull of if-statements on an Arduino to accomplish what they need.

In my experience, LLMs are capable of solving the kinds of problems you could have copy-pasted from StackOverflow, but as soon as you try and do something too complicated or unique, it falls apart.

The trap is that it seems capable at a surface level in areas people aren’t personally an expert in and then assume it can accomplish real tasks on its own like an expert. It can be great as a tool for getting started, or for otherwise throwaway code, but I’ve yet to see any model produce code that’s complex yet maintainable in the long term.

Historically, there’s been a bit of a whiskey trade on the Denmark-Canada land border.

Well, it looks to be plugged in and charging via a cable. Possibly connected to Android Auto / CarPlay. From the photo they could probably argue it’s a battery bank or something like that.

I’m 50/50 on if they really don’t have a place to rest their phone vs they were previously using the phone and just put it down between messages. This fine seems a little excessive, but it’s consistent with the law there, so /shrug

An interpreter is significantly easier to sandbox than native execution, and it means they can make guarantees the host system won’t crash no matter what the program running is.

I could see it being useful for allowing third parties to run experiments on their hardware in space without having to manually verify the code safety.

Yeah, I said transformer because that seems to be the state of the art in AI architectures, but purpose built neural network hardware might not actually benefit from the same architecture.

A neural network made from analog hardware could theoretically replace a significant portion of an LLM’s processing and not be limited by things like floating point precision or clock speeds. Who needs floating point when you can literally just multiply voltages together with a couple transistor junctions at the speed of light?

If they don’t have a warrant, it’s not admissible in court anyway. Let em have it.

Oh it’s definitely admissible in court if you willingly hand it over. If they have no warrant, they have no legal right to enter your home, and you don’t need to talk with them. But that doesn’t apply if you invite them in and let them start developing probable cause of a crime.

Me: I need some new hardware to run my transformers on

Mom: We have transformers at home

The transformers at home:

Joking aside, I think you’re right, using discrete floating point math to simulate a transformer architecture will never be able to approach the efficiency of a “native” analog system like actual brains. I think eventually we’ll see someone come up with a hardware transformer that doesn’t require full synchronized clock signals.

We’re at the point where you can fit an entire computer inside a USB cable end without it looking any different (beware of keyloggers if you’re not using your own cables, they can even fit a wifi antenna in there)

It’s not a $20k repair bill I’m worried about, it’s the potential of $100k+ in medical liability that I’m really buying insurance for.

In my area there’s plenty of expensive cars driving around too, I somehow doubt the minimum $25k insurance would cover even half the cost of a totaled car + everything involved.

Sarcasm or not, the people dealing with localized datacenter heating during already hot weather don’t find it funny.

"AI datacenter are creating “heat islands,” warming the land around them by up to 16 degrees Fahrenheit, and making life hotter for more than 340 million people. "

Yes, all that emitted radiation does cool down the sun. It’s why it has a mostly stable temperature instead of getting hotter infinitely.

There’d be nothing less than gold plated B300s in a space datacenter. If they’re spending billions flying it up there, they’re not going to be putting mid-range consumer GPUs in there. The gold plating probably doesn’t even do anything for radiation shielding, the AI just told them to add it to help prop up the AI bubble.

The advantage of datacenters in space is that the peasants can’t break in and sabotage your equipment. Only a very small set of nations would have the capability of blowing it up or somehow jamming its communications.

It literally only makes sense if you’re a billionaire worried about the growing unpopularity of your AI datacenters, or you’re using it for war and don’t want it easily bombed…

As far as I know, there are no more orbital deadzones. We’ve got so many satellites, if not Starlink, I’m sure some other satellites could do the job of relaying to ground stations. We get live video of boosters landing in the middle of the ocean after all.

This. Honestly they’re amazing on airplanes, since you can still talk and hear all the announcements / flight attendants without the background noise.

I might have to rewatch The Fast and the Furious. Now it’s just AI hardware instead of VCRs or whatever it was in the movie.