Official policy on use of AI

[Robbo99999]

It depends which AI model you're using and how well it's performing. One of the ways they've been improving performance while lowering running costs is to work out what skills the prompt requires and hand off to smaller, more specialised models. If you're running a model that does this, and it interprets your prompt correctly, you'll probably get the correct answer. If you're still using a pure language model, or the handover is wrong somehow, you'll probably get something that looks plausible but will be wrong if you check the calculation. And even with the same model it may depend on what backend tuning the operator has applied, perhaps preferring a cheaper but less accurate solution because the load is high or electricity more expensive just now. This is all opaque to the user, so if it matters you'd best get it to show its workings and check them.

ChatGPT 5 does that kind of auto switching. I've been happy with the auto switching so far. I've found I need to go with the paid model though because you don't get enough of the "long think" answers in the free version.

 

[respice finem]

Is that HAL 9000 by any chance? I'm not sure, just triggered something in my brain re that recollection.

BTW: I'm surprised no LLM provider called his LLM "HAL 9000"

 

[Robbo99999]

BTW: I'm surprised no LLM provider called his LLM "HAL 9000"

Ok, but it would have to be done in a very ironic way, HAL 9000 was nuts if I recall!

 

[respice finem]

Ok, but it would have to be done in a very ironic way, HAL 9000 was nuts if I recall!

Exactly - that's why it fits
At least no one could complain later, it weren't "written on the tin"

 

[Mart68]

Ok, but it would have to be done in a very ironic way, HAL 9000 was nuts if I recall!

There was a conflict in the way it was programmed. A bit like A.I

 

[respice finem]

There was a conflict in the way it was programmed. A bit like A.I

It went nuts because of having to deal with homo sapiens
/I.

BTW: We are just late to the party:

GitHub - kinzozi/HAL-9000: A command line GPT assistant for processing and generating commands using NLP.

A command line GPT assistant for processing and generating commands using NLP. - GitHub - kinzozi/HAL-9000: A command line GPT assistant for processing and generating commands using NLP.

github.com

 

[somebodyelse]

BTW: I'm surprised no LLM provider called his LLM "HAL 9000"

Movie companies are rather more litigious than your average author. They might get away with having it sing "Daisy, Daisy" as you close the window though.

 

[Timcognito]

Yep, I agree, I've experimented with it as a German conversationalist pal! I also used it for a while to try to teach me to play keyboard, well I'm still in the beginning phases after I had a pause due to some tendonitis I needed to let settle down, should start it up again time permitting!

FYI @Robbo99999 https://www.stacksocial.com/sales/skoove-premium-lifetime-subscription-2

 

[Cbdb2]

I've seen quite a bit of AI quoting recently, and I don't think it adds to Audio Science.
It's like parroting Wikipedia.
If you have to resort to wiki or AI to make your point, then you don't know what you're talking about, and should keep quiet.

You cant compare wiki and AI. Almost all of the technical/scientific stuff in wikipedia is correct. I use it to reinforce or give cred to my points. Cant say either for AI.

 

[Somafunk]

Because I just wrote them as I thought of them. Will formalize it soon....

But did you really?, or did you use an A.I. to co-write?

 

[Cbdb2]

I see ... Well, once we’re at it: What about banning any content that has been composed using details coming from »search engines« just the same? And also any text that has been automatically spell-checked?

Really? So you think doing research and using spell check is the same as having an AI do the thinking for you?

 

[Sono]

So you think doing research and using spell check is the same as having an AI do the thinking for you?

No.

 

[welwynnick]

You cant compare wiki and AI. Almost all of the technical/scientific stuff in wikipedia is correct.

No it's not, they both get things wrong, but that's besides the point.
The point is that people are lulled into a false sense of knowledge and understanding by using both.
What they have in common is that neither are a substitute for education and experience.

 

[SoySalsaPicante]

No it's not, they both get things wrong, but that's besides the point.
The point is that people are lulled into a false sense of knowledge and understanding by using both.
What they have in common is that neither are a substitute for education and experience.

There has to be a balance somewhere.

This is meant to be a hobby, not a PhD.

 

[tmtomh]

100% in favor of AI being disclosed - IMHO the lack of identification of AI-generated content in threads is a subset of a larger problem that is endemic with all online communication and in fact much of human communication, period: people are too unaware, thoughtless, and/or lazy to provide basic information and context that would help others understand what exactly it is they are saying and where it comes from.

As for "AI," we're talking about LLMs, which can be very useful in intelligently applied, specific applications but which cannot in any way think, and which are massively overrated.

As for this "being a hobby, not a PhD," sure - but peddling incorrect information or not caring whether information is correct or not isn't fun for me, and it isn't fun for a lot of other members here. Party of my enjoyment of the hobby is learning about and benefitting from what is accurate and what is not.

 

[Sono]

This is meant to be a hobby, not a PhD.

So what about renaming the whole thing to Audio Hobby Review? Thumbs up!

 

[JohnBooty]

I'm not sure what the ASR mods have seen, but as a mod of an audio-related subreddit.... we've had like ONE person trying to pass off AI slop as human.

I quite honestly doubt that many regular human users have any desire to do that. If you see AI slop I suspect it's some kind of bot account, trying to gain some karma/rep/whatever as a part of a plan to deliver spam.

Maybe so, but I am still waiting to encounter an AI quote that is actually useful for me.

There are uses besides having AI do the thinking and come up with ideas for you. I would agree: that is pretty bad.

AI can be an assistive technology, sort of like a spell checker on steroids.

• Consider people with conditions such as dyslexia
• Consider the relatively large percentage of technically-minded people who struggle with "tone" and nuance due to "engineer brain" and/or perhaps being on the spectrum
• Consider people whose primary language is not English.

 

[JohnBooty]

So what about renaming the whole thing to Audio Hobby Review? Thumbs up!

There is no conflict between "hobby" and "science."

 

[Sono]

There are uses besides having AI do the thinking and come up with ideas for you.

Why, sure. And those are the only uses we sincere ASR members will ever take into account.

Or do we?

 

[Mnyb]

HAL is also friendly jab at a computer manufacturer that dominated the market then , just move one letter further in the alphabet . Whats the next letter after H and the next letter after A and L