I remember Slashdot before all the whining about American politics in the comments..

It is far more expensive, but there are also some other benefits. Energy production is more reliable because of more dependable wind velocities at sea. There are also fewer issues with available space in some regions (but I guess not really Texas).

Indeed, I wish I will be in such good health when I turn 95. If I make it that far.

But if you look at one of Nvidias examples from Resident Evil: Requiem (first image in article):
https://arstechnica.com/gaming...

and look at the Cigarettes sign on the building on the right, you can see obvious mesh distortions in the DLSS 5 version. How is this caused by lighting alone?

Calling it a rogue AI implies intent, which implies some level of awareness. So why not "Incompetent Engineer Placing Too Much Trust In AI Agent Triggers Serious Security Incident At Meta"?

It's very obviously more than mere lighting, it's clearly changing textures, removing fog, doing the equivalent of changing meshes etc.

I thought he planted potatoes in Martian soil, but I guess I was mistaken.

This looks like cool tech demo, not like something I would actually like to run in most cases. It needs significant control from game developers, for example to only apply the effect on certain objects and only before e.g. lighting and shadows are rendered. Definitely not whole scenes.

but but 0.1 mm thinner

That won't really work at least in the EU where MFA is needed for online credit card transactions.

The whole "upload an ID" idea is extremely dubious. It would be a lot better if there would be a standard system for providing age category verification online without exposing any other personal information. So you get redirected to the verification site, do strong authentication there, and the verification service just sends back something like "this person can view R-rated content" to the site and nothing else. Not their age, true name, address etc.

Sure there is still the age verification service provider to worry about, but that could be much more tightly controlled than people uploading scanned IDs to identity phishing porn sites.

The issue with your logic is that the whole AI model would not generate anything useful from your brilliant prompt had it not been trained with copyrighted material. And even if it's covered by copyright, why would it be your copyright and not the company's who trained the model?

If I brilliantly describe an evocative scene in detail in text and then a painter paints a brilliant picture following my description exactly, I still don't own the copyright to the resulting painting, only to my original description.

Yes but

1) you can't copyright mere names.See https://www.copyright.gov/help... ("Names are not protected by copyright law. Some names may be protected under trademark law.")
2) you don't have to copyright your works, they are copyrighted by default, although in 1987 the US still required a copyright notice to be added, BUT
3) the 1st edition of "Operating Systems: Design and Implementation" where Minix was first published, HAS a copyright notice: (c) 1987 by Prentice Hall inc.

A PDF of the 1st edition can be found here: https://github.com/olegslavkin...

I guess you mean trademarked? You have to keep using and protecting your trademark for it to stay valid, I doubt he'd have wanted to bother with that.

You don't have to copyright anything, you have a copyright on all works you release automatically unless you declare it public domain. But the key concept here is "works", you can't copyright something short line a product name or a simple slogan. So there would not even have been a way for him to prevent someone else using the name Minix just on the basis of copyright.

Ooh edgy strawman. The article is literally about air pollution in neighborhoods.

Yes, it's clear that some tasks can be automated well with AI when using specially trained models. Classification tasks and such comes to mind.

It's also clear that there is some productivity increase overall possible by smart use of AI. I found for example that instead of reading Microsoft's documentation, it's often more efficient to ask Copilot about it. Not surprisingly Microsoft's AI is pretty decent at knowing Microsoft's own stuff.

My worry is having to waste time fixing someone else's AI vibe coded slop in the future..