It amounts to telling it: “Stop doing that thing you can’t stop doing.”

Mammals converged on systems that satisfy the same fluid-mechanics.

Substantive issues with this submission aside, it’s a mistake to have such long conversations with an LLM. The longer they go, the more likely they are to accumulate errors. The latest models all claim to be able to handle long conversations, but in my experience they still don’t do as good a job as just pasting your conversation into a new thread.

Does a “pledge” have more or less weight than a pinky promise?

Depends, were their fingers crossed behind their backs or not?

I’ve always been skeptical of niche archival formats, even ones as robust as this. Even if the technology itself is sound, companies go out of business and formats go obsolete. I’ve been using plain old mirrored spinning hard drives for years. And for parts of my archive, I give copies to friends and family for added redundancy.

FWIW I mainly go with offline HDDs and SSDs. And I basically now keep one source of truth, which I synch everywhere, for backups (and I verify the backups and make sure that my main source of truth ain't corrupted).

> I’ve always been skeptical of niche archival formats ... and formats go obsolete ...

The format is supported by Linux, that's never gonna be an issue. Not only can modern version of Linux read DVD or BluRay formats, should the support disappear, there's not a world in which in 30 years I cannot run an older version of Linux. There are, for comparison, people running Commodore 64 and Amiga hardware, today. You'll always be able to run the software, either on bare metal or emulated.

The issue is: will you find a drive in 30 years? As they are still built today and as many DVD readers from 25 years ago are still working today, I take it it's going to not be that hard to find a BluRay drive in 30 years and hook up to a machine running Linux.

And even on a BluRay, you simply do not store that much.

If one doesn't want to only rely on HDD/SDD and online storage, it's still probably a safer bet to go with tapes: you can store much more data, newer readers can read (up to limit) older tapes and these are battle-tested, supported for a long time, available, reliable. Because, well, it's not consumer tech but enterprisey.

> On Metaculus, a group of forecasters has taken to estimating when AIs will have the chops to out-predict an elite team of humans. Last January, they said there was about a 75 percent chance this would happen by 2030. Now they think it’s more like 95 percent.

I didn't change the title. The NYTimes did after I posted it.

I just reread E. M. Forster‘s 1909 short story “The Machine Stops.” I feel it should be assigned reading on an annual basis for everyone working in tech today.

https://www.cs.ucdavis.edu/~koehl/Teaching/ECS188/PDF_files/...