aktenlage

The models can predict any conversation, so it probably doesn't matter to them. For us humans it would be much better to let them write in the third person (e.g. "the coding assistant needs to verify XY, it does it by calling `grep -i abc`). It would decrease the anthropomorphism.

There is no need to have a final one. It's no problem to keep supporting older codecs and add new ones in new products, as long as there's a "common ground" of widely supported codecs.

Well said. I wonder whether AI games from low-effort prompts will play like most AI written blog posts read, superficially great but on some level not connecting with the user.

There are high stakes customers who will want to use the best model for any price. Military, secret service, ...

Maybe future models will just be trained to give wrong answers instead. That should be harder to overcome.

I for one wouldn't trust abliterated model providers, unless it's clear that they have a clear incentive not to sneak in unwanted behavior, as described in that top story last week.

It may solve the problem of anthropic and open ai going out of business with all major consequences that would have for the rest of the US.

It's not black and white though. There's a scale from "easy to censor without collateral damage" to "need to take the whole nation offline to suppress information flow".

And there is a scale from "trivial to circumvent, if only (you know) you want to" to something like "need to be very tech-savy and use custom hardware"

You are right, my answer didn't account for the specific citation you were answering to. Sorry.

If it can be shown that left-wing politicians in key positions of their party want to abolish core concepts of the democracy or constitutionally guaranteed rights: sure.

Centrist and far-right views have failed. Spectacularly so in Germany. Maybe spectacularly is not the right word for WW2 and the Holocaust?

And even then, you have just the word of the producer that there wasn't any poisoned data used for training. For that you would need "reproducable builds" (i.e., training). Which makes no (economic) sense, given that the training is the expensive part.

Thank you, great read.

I got it running, I just didn't get how to do it with podman. Thanks for the hint to use dockge, I'll look it up. I'm using wire guard with my router for outside access.

I am sure nixos works perfectly reliable. I meant a specific setup (I've the configuration as home server with immich, paperless-ngx, jellyfish, samba, ...) as I am sure there still are better and worse ways to configure such a system. Is there a good "library" of NixOS configurations?