Renegade

All the Suno tracks I've heard have a similar style. Very procedural and formulaic. Calling it AI seams like a stretch.

Relevant article: https://lemmy.ml/post/12857742

Prompt engineering is a thing, but I wouldn't say it's much of a job title. There are people doing it: optimizing system prompts, preprocessing and postprocessing, llms are just one piece of a complex pipeline and someone has to build all that. Prompt engineering is part the boot strapping for making better llms but this work is largely being done by data scientists who are on the forefront of understanding how AI works.

So is prompt engineering just typing questions? IDK. Who knows what those people mean when they say that but whatever it's called there is a specialized field around improving AI tech and prompt engineering is certainly a part.

Containers are a really cool part of security. The security provided will depend on how the container is configured. For example if you give the container bridged networking permissions (or whatever equivalent term is used by your solution) then you're giving the container access to communicate with other devices on your local network. This would be the opposite of what you want to do to prevent an attacker from pivoting through your LAN.

Other threats just aren't within the set of protections that can be provided by a container. For example if you wish to protect your Minecraft world from being griefed the container won't have any affect on this. Another example is hiding your IP.

Basically what I'm saying is that whenever you are looking at a security technology think about what guarantees it provides and realize that no single security measure provides protection against all threats.

You're basically relying on the security of minecraft, and your ability to quickly patch. The Log4j exploit is one good example of the kind of threats you might face.

Another is just that revealing your ip can open an opportunity for various forms of harassment. Lots of us skate by on obscurity and luck without to many issues, but that's not a very robust solution.

Nothing in the article corroborated the claim in the title that human intervention made things worse, just that the problem goes deeper.

"AI Prompt Engineering Is Dead" long live LLMOps which is totally not the same thing /s

I thought it was a strange choice to use such a technical descriptor in their name. This makes sense.

Is this good? How does it compare to the existing tools?

I would be curious what the article means by AI. For example this might include some transcription and sentiment anaysis. Didn't see anything too complicated in their description of what the software does.

If I understand correctly, basically they are testing if the llm can create novel outputs.

Oh now i get it. Yes, exactly!

No actually, the water in spent fuel pools does not contain radioactive material. The water provides shielding. You could hypothically swim in that water just dont dive and also they would never let you do that because it would contaminate the pool.