Computerus

[Thad]

Then there's the question of whether I should do a fresh OS install or just stick with the one I've got. AIUI it's not that difficult to hook up the same boot drive to a new computer and have it work; since I'm going from Intel to AMD it's probably going to require booting to USB and reinstalling the bootloader, but it sounds like a pretty straightforward process.

I think I'll probably do that. While it might be useful to start fresh and without all the cruft I've accumulated over the years, I've spent a hell of a lot of time setting this thing up and I do not find the thought of having to do all that over again appealing.

It mostly worked fine but somehow now every time I do a kernel update I have to boot from a USB dongle and reinstall initramfs and GRUB.

I'm baffled as to why. It's three commands; if I execute them from my system before a reboot, I end up in a non-bootable state (trying to load the EFI partition before loading the module to read a vfat filesystem), whereas if I boot to a memory stick, mount the appropriate partitions, chroot, and then run the same three commands, everything works fine on my next reboot.

I've got it down to a workaround that takes under 10 minutes, so on the whole it's still less fucking around than if I'd done a clean install, but it's still frustrating and I'm sure there has to be a better way which I haven't quite cracked yet.

I figured out what went wrong here: when I reinstalled the bootloader after swapping out the motherboard, I didn't have my boot partition mounted, so it created a /boot directory on my root partition. So I ended up with a bootloader that expected the boot files to be on partition 3, and every time I updated my boot files they went on partition 2, and then the bootloader couldn't find the files it was supposed to be using to boot. Fortunately it was an easy enough problem to fix once I figured it out, and now I can reboot after an update with no issue.

[Thad]

Framework gives its modular laptops 13th-gen Intel CPUs and (finally) an AMD option

Well whaddaya know. I've been looking for a nice 13" AMD laptop. Will have to keep an eye out for when this one is actually released.

I assume it won't be cheap, but I'm willing to pay a premium to support Framework.

[Crick]

Has anyone here heard of Eliezer Yudkowsky? AI philosopher and critic? I've sorta "rediscovered" him and his work after briefly hearing of him years and years ago.

Apparently he has a new-ish piece in Time where he's calling for an end to AI research, a draw down of all GPU clusters, and missile strikes on any server farms that don't.

But I can't seem to find why he's so sure ASI would kill all biological life on Earth. I think it's related to the old paperclip thing-- superintelligent paperclip factory just converts all matter into paperclips-- but that seems like sorta specious reasoning?

That's to say, I do think AI stands disrupt a lot of daily life. It's a real concern that should be monitored closely.

But this seems closer to a fanatical doomsday prophet. We don't know when it'll happen, or why, but it'll one day destroy us all. Join my organization, let us put a stop to this, etc, etc.

[Mongrel]

Sounds like he fell off the deep end, wayyyyyy off. ASI is not AGI and all this "AI! AI! AI!" panic seems predicated on someone having a creepy ChatGPT experience and then wildly confusing the two, which is still an issue of sorts, but quite far from the apocalypse. This sort of panic being featured in mainstream publications is exactly why I was bitching in the other thread that we really need to be calling ASI something, anything, other than "AI".

As for the cautionary tale of the Paperclip Optimization Machine that nerds are so fond of, the thing is, they seems to forget we had Computers more than a century before we had computers, and that their Optimization Machine doomsaying is neither theoretical nor strictly about mechanical machines; Optimzation Machines are very, very real. We, and the planet we live on, have already suffered their horrible effects for hundreds of years already, in the form of the joint-stock company - better known to you and I as corporations.

[Crick]

Yeah, that's what I'm trying to articulate.

Our current economic system will use Dumb AI to absolutely ruin people's lives way before we reach whatever we'd describe as AGI.

I guess I'd believe a "superintelligent"-- or super optimized or whatever-- worm built for ransomware accidentally locks out every computer connected to the wider Internet in, like, a day before an ASI that spends a decade carefully constructing a weapon that wipes out all of humanity in an instant so that... uh....

[Grath]

tl;dr he's a creepy eugenicist doomsday cultist who thinks he's a genius and is fundamentally incorrect about how actual real world technology works.

[Crick]

Oh shit! Is there a term for when someone is milkshake ducked, but you're not surprised.

Also pretty disappointed to learn transhumanism has eugenics roots too.

[Mongrel]

ASI just seems threatening in the current moment because the most accessible uses involve scamming people or corporations continuing to corporation, only that little bit more so.

Which isn't all that surprising from a tech that's basically being trained to produce output more-or-less like a human while trying to overlook all sorts of small-yet-crucial ways its work is inferior.

Except maybe for people who get excited by rotoscoping. Or who're enjoyers of six-finger porn. Or something.

[Mongrel]

Oh and the NFT crowd have moved over to shilling hard for ASI, so, lol

BIG surprise, I know.

[KingRoyal]

This reads like a bunch of neo-reactionaries who forgot that "malevolent AI" is mostly a science-fiction concept

[Crick]

I couldn't believe how many people just agreed with him, but finding the actual "why" AI would kill everyone was hard to find and vaguely reasoned. That seems like the most important part, right?