Nearly all evaluation steps for an LLM I produce some level of entropy. Yes you can evaluate an LLM model deterministically, but we don’t. We even tune the levels to get good results. Model cards have sampling parameters supplied to give “the best” results. For example, Qwen 3.8 flash next gives

We recommend using the following sets of sampling parameters for generation:

Thinking Mode: temperature=1.0, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0

Instruct (or non-thinking) mode: temperature=0.7, top_p=0.80, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0

Replaying the seed will only get you the same result using a particular engine and a particular random number generator. Run it on llama.cpp Vs Strata and you can get different results. Different libc, different results.

A compiler is systematically deterministic. No calls to rand()

AI is not a tool. Its an addiction

I totally agree. I’m watching people lose their drive to create and it’s so sad. They’re getting high off the satisfaction of “creating” without the work.

…and then they lose access to their wündertool and they can’t do anything.

I’m finding the allure of AI is particularly strong for people who have risen to positions of power where they had to stop doing the work themselves. Instead, they had to work through other people to get bigger projects done.

The allure is that you can just say what you want done, and it does it. If it’s done it “wrong” you can just tell it it’s wrong and not have to worry about arguing the technical differences. As such you feel like you get closer to what you wanted, rather than somebody else’s take on what you wanted.

The things is, when you are removed from the code, you can only work at a high level. That person you asked to do something is considering 10 other low level concerns that you didn’t consider. I think Linus’s recent patch series is a good example. He saw a potential optimisation, did a proof of concept but missed a race condition. He’s now handed it over to the person that normally works in that area and knows the details. Can the optimisation actually work? Who knows?

Nobody sane supports indefinite war. That’s not what is being said. They don’t support a peace treaty which will only last until Russia rearms.

This isn’t necessarily bad,

I’m of the opinion that it is bad, from the perspective of the user. They are now locked into that company. It not bad from the perspective of the authentication provider, and the service provider may get a positive outcome of it from fewer breaches due to bad passwords.

The problem is the authentication provider isn’t trust worthy.

Big silver?

Which most people don’t do.

Well if that’s possible, it doesn’t sound like a good basis for security.

Don’t believe it until the results are in.

The UK constitution is written. It’s just not in one document. In this case I think it comes from the civil war when the absolute powers of the crown were given to parliament.

Which is rather the point, there comes another act of parliament and more constitution is formed. What came before can’t stop that process. Even with a codified Constitution like the US one, there’s always a process of amendments which can change the previous versions. The fact that the US hasn’t had an amendment in a very long time is just a lack of political will.

So I think my question is a reasonable one. The founding fathers felt that the constitution should always be able to be changed, so wouldn’t state legislation also always be able to be modified by a future legislature?

In the UK constitution there are laws that state that no government may bind the hands of a future government and that a vote can always be superceded by another vote.

Does the US not have such laws?

It’s what was known as “Big Data”, but that phrase doesn’t get them hard anymore.

Trump is the last one to go for:

• Hegseth (Probably easiest to get momentum with)
• Roberts, Alito and Blanche (Restablish rule of law)

Then Trump and Vance.

Best way to avoid it is not travelling to/from the US.

It was gray. The used the concept of being outside of US jurisdiction to skirt the laws on illegal search and seizure. Somebody pushed back on this recently, so they made it less gray.

The Zig language has a no AI contributions policy. That’s for the language itself, but a lot of it’s community have similar views.

Codeberg has a no AI policy for projects hosted on its servers, so anything hosted there should meet your requirements.

You turned it off from their menu option? You’re a trusting soul aren’t you.

The TVos is the malware.

Does it have WiFi? It doesn’t need cables to look for other devices to communicate through.

…and then don’t live near anyone that doesn’t take the same precautions.