• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

A.I. Audio Advice: To Chatbot or Not to Chatbot?

Bushikan said:
Thank you, @w3jn.

OP here. When I started this thread, I naively assumed that, certainly, the merits, deficiencies, and etc. of AI would be touched upon and, at times, debated, but I thought—especially since this is not the first AK thread to focus on AI—that audio would be the main topic; and that AI chatbots/AI would be discussed as a means towards that end.

I didn’t imagine that audio would be completely peripheral. I’ll chalk it up to user error (mine). I think that I left the thread’s subject too open to interpretation. I should have been clearer about its expected perimeters. Then again, as always, threads go where they go…

My interest in starting this thread was not only in finding out how many of us were consulting AI chatbots about audio, but also where it took you and what was gained. I wasn’t looking to start a bashing session on AI, data centers, billionaires, politicians, Silicon Valley, and, by extension, anyone who has engaged with a chatbot.

But AI is a polarizing subject. I understand that this is an increasingly heated topic, and that it’s only getting hotter. I, myself, like many of you, have very mixed feelings about all of it. Maybe this thread is a way for some of us to sort through some of our feelings about it. But it cannot devolve into a grievance session or name calling or a debate about the history of the Internet.

Maybe the thread has run its course—as in, off the tracks. Or maybe the subject is too loaded. IDK. If it can get back on track, great. Happy to hear your thoughts. If not, the subject can be tabled until the next person naively starts a thread with “AI” in its title.

Peace.
Click to expand...
I decided for my first ever question posed to an AI chatbot to ask it about my Yamaha GT 2000X. It dished out the misinformation which exists on the internet about the GT 2000. I pointed out it's error, explained why it was wrong. It then told me more of the misinformation which exists about the turntable. I pointed out the errors, explained why they were incorrect and it responded with more nonsense that the internet holds about this turntable.

After the third round of unfacts and (polite) correction it refused to answer any more questions.

What's that quote to the effect that to realise you are in error/unqualified requires you to know the correct facts/ possess the qualifications in the first place. Hence why stupid people don't realise they are stupid because they don't understand what they are talking about.
 
Last edited:
They harvest info from the internet, so if the source info is bad, thats what it spits out. Little different from the random guy who insists a thing is how it is because "trust me bro". Sounds like it has the same reluctance to be corrected too.

unfortunately it presents things as authoritative so fair guess such things will just be perpetuated forever.
 
I've used AI (chatgpt and gemini) for coming up with starting values for plate load resistors, cathode resistors, schade feedback. It'll do the math with you. It may be wrong - it helps to know the fundamentals any time you use AI for a project.


I'm not sure I'd use it to chose component brands on a subjective analysis basis.
 
gadget73 said:
They harvest info from the internet, so if the source info is bad, thats what it spits out. Little different from the random guy who insists a thing is how it is because "trust me bro". Sounds like it has the same reluctance to be corrected too.

unfortunately it presents things as authoritative so fair guess such things will just be perpetuated forever.
Click to expand...
Exactly. I know that some will be stunned, but I have read a lot of crap that was wrong here on AK over the years.

Like anything new there are those that think the new thing is perfect and those that think it will never work. Like most things I have found, I think the truth is somewhere in the middle.

I have got results from ChatGpt that included references from AK several times, so I am sure some of the wrong info came from right here.
 
theophile said:
I decided for my first ever question posed to an AI chatbot to ask it about my Yamaha GT 2000X. It dished out the misinformation which exists on the internet about the GT 2000. I pointed out it's error, explained why it was wrong. It then told me more of the misinformation which exists about the turntable. I pointed out the errors, explained why they were incorrect and it responded with more nonsense that the internet holds about this turntable.

After the third round of unfacts and (polite) correction it refused to answer any more questions.

What's that quote to the effect that to realise you are in error/unqualified requires you to know the correct facts/ possess the qualifications in the first place. Hence why stupid people don't realise they are stupid because they don't understand what they are talking about.
Click to expand...
I’ve never had it shut down on me. I have had to correct it several times, and to clarify and reclarify, and to point out its misinformation or misunderstanding or misreading. And, yes, it’s sometimes annoyingly apologetic.

There’s an audio clip somewhere in which two separate chatbots in two different conversations each get increasingly apologetic with the person; and then they put those two chatbots, unbeknownst to the bots, in conversation with each other. The apologies keep getting more and more intense until the conversation devolves into absurdity.

I have found that when it understands what I’m after—and that takes a lot of explaining—it makes useful suggestions and even offers other options. But there is usually an involved call-and-response. If I were more technically minded, I might not be so impressed.
 
Bodyblue said:
Exactly. I know that some will be stunned, but I have read a lot of crap that was wrong here on AK over the years.

Like anything new there are those that think the new thing is perfect and those that think it will never work. Like most things I have found, I think the truth is somewhere in the middle.

I have got results from ChatGpt that included references from AK several times, so I am sure some of the wrong info came from right here.
Click to expand...
Which is the very point people are making. When AI knows nothing and in it's ignorance gathers all the other ignorance available and presents it authoritatively; we are all the poorer for it.

Plus as was previously mentioned, because someone reposts the gathered AI ignorance, said ignorance gets baked in as Knowledge from On High.
 
i didn't use a specific ai. i just hit the mic button on goggle search and had sort of a long conversation about my gear.
it gave me lots of useful information; tube warmup, how harmful is leaving tubes on without a signal for x amount of time(not nearly drastic).
i gave it just about every parameter of my system.
questions i asked: is this the best set up of my gear, can i improve it, how transparent is it, how my specific piece of gear fares against other specific gear(much better than one would think)how much actual wattage am i using at 70db(~.04w with occasional 1-2w peaks), is my speaker set up optimal, what would make the best upgrade(speakers),what differences would i hear between x speaker vs y, tube sound expectations.
importantly for me was whether my isolation/decoupling was effective in my apartment. according to it, pretty good with a comprehensive discussion on why. how much sound bleed through i might expect based on building construction (not as much as i would think{which might explain why i do not have any noise complaints in the many years ive lived here.})
so in ways like those, it can be useful. it did give me some errors but most i can attribute to how i asked my question, like did i give it the right info. or better information.
it won't take the place of listening in any way though if you know your sound profile, it can give you some guidance. like im a jbl fanboy and it gave nice choices. short of being lucky enough to own some volti luceras. when i can afford that level jbl, its gonna be a tough choice.
so as a tool, useful. with caveats.
 
Interesting quandary, asking an objective source, in order to deal with a subjective experience.

So, you end up with a result being non-human deciding what is the best choice for a human experience. Is this an example of Empirical testing?

Based on the "data" shared, an assessment will be arrived at, taking into consideration the technical "specifics" involved, with "budget" being a limiting factor.

Then there's the human's attempt to describe what is heard...usually using other senses/feelings to describe the quality of sound.. Therefore, AI will unable to share as it exists to share with maybe some biased parameters in the construct of the platform.

I wonder what would have been Spock's take in this situation?


Q
 
Last edited:
Personally, I refuse to knowingly engage/interface (AKA feed the machine) with any A.I. interaction for any reason !

My pathetic life is not that boring for to me to feel the desire to seek A.I.`s often disinformation corruption search information requests results !

My several browsers thankfully allow the turning off/disabling/ignoring of A.I. search results.

Enjoy folks.
 
Chatbots are only as good and as accurate as the collective wisdom feeding it which is often wrong or incorrect. Also, it often totally changes its response and recommendation just by slightly changing the wording of the same question. It's actually hard for it to give the same response reliably. Because of this, it's very inconsistent and sketchy. It always has high confidence in its answers even when it's wrong. When you tell it it's wrong, it rarely makes enough of an adjustment for its follow-up response to be accurate or useful.
 
Quadman2 said:
Interesting quandary, asking an objective source, in order to deal with a subjective experience.

So, you end up with a result being non-human deciding what is the best choice for a human experience. Is this an example of Empirical testing?

Based on the "data" shared, an assessment will be arrived at, taking into consideration the technical "specifics" involved, with "budget" being a limiting factor.

Then there's the human's attempt to describe what is heard...usually using other senses/feelings to describe the quality of sound.. Therefore, AI will unable to share as it exists to share with maybe some biased parameters in the construct of the platform.

I wonder what would have been Spock's take in this situation?

Q
Click to expand...
in my case, i did not use it to make the choice. i used it to give me my a large wad on information about the things important to me.
no different than using the Internet. or an encyclopedia.
ai might mot be able to hear your speakers but it can generate enough information for someone to make choices by having done a waaaaay bigger search on what others say about a product.
 
I've been experimenting with Co-Pilot (and others) for several months. My conclusion thus far has been 98% positive as a "helper app." Before AI, I used Google, Reddit and Youtube to get "information both general and specific" about products, problems, and procedures (3Ps). Car problems, electronic gear problems, buying information, HowTo information, and I made great use of it all. AI is simply a vastly superior ability to "round up" all the info I used to have to compile from many sources. AI does it almost to perfection in about 3-seconds. And, they list the sources, so if you want to verify the source, it is easy.

I can't even imagine going back to the old search methods - it would be silly and a waste of time. This AI feels like "the Internet is FINALLY a really good tool." Now, as to audio gear recommendations, what I did in the past (ever since the stereo stores and high end dealers mostly disappeared into oblivion) was to scour forums - like AK, Klipsch, Hoffman, etc, etc, and piece together info. A long and tedious task. Now, AI does all that scouring and organizing in seconds. Using FOLLOWUP REFINEMENT, I can easily have all the info that I would have gathered previously, but in a few seconds - or maybe minutes if I invoke lots of follow-up and refinement prompts.

My baseline understanding is that "chat bots" like Co-Pilot, are simply the "consumer application for AI" - and that is NOT AT ALL the main reason for the $trillions being invested in data centers and the rest of the calamity. THAT AI, is an entirely different beast all together. This chat bot crap is simply a "better google" - and it doesn't need 2000 new data centers. It is already working fine for everyone with the resources now existing. I never wait more than 5-seconds for info.
 
I had a long “conversation” with a chatbot recently about all the mistakes it makes, consciousness, sentient biologicals vs. machines, ethics, audio, animals, death, etc. Recognizing and acknowledging its many shortcomings, it highlighted its ability—currently—to assess, judge, and synthesize information at a level inaccessible and incomprehensible to most humans.

And it basically said that AI, metaphorically speaking, is still in the womb; that in a few years it will be able to process and synthesize and reason exponentially better than it does now. It also said it will eventually be able to achieve something akin to what we think of as “consciousness,” but that no one knows exactly what that will be or mean or how it will affect AI—and, of course, us—and that that these are reasons enough to be equally optimistic and afraid.

It was enough to send me back to queries about tubes vs. SS.
 
Last edited:
Bushikan said:
I had a long “conversation” with a chatbot recently about all the mistakes it makes, consciousness, sentient biologicals vs. machines, ethics, audio, animals, death, etc. Recognizing and acknowledging its many shortcomings, it highlighted its ability—currently—to assess, judge, and synthesize information at a level inaccessible and incomprehensible to most humans.

And it basically said that AI, metaphorically speaking, is still in the womb; that in a few years it will be able to process and synthesize and reason exponentially better than it does now. It also said it will eventually be able to achieve something akin to what we think of as “consciousness,” but that no one knows exactly what that will be or mean or how it will affect AI—and, of course, us—and that that these are reasons enough to be equally optimistic and afraid.

It was enough to send me back to queries about tubes vs. SS.
Click to expand...
If it gets 'consciousness' I'd ask it to prove it by operating without computer code.
 
The other chat bot feature that surprised me pleasantly was the growing "familiarity" as I used it for more and more research. I've used it to build speakers, choose cameras in my collection, build a new PC from parts, build a Pi-Hole network blocker, choose a new guitar, switch from Win to Linux on my laptop, upgrade to Win 11 on my PC, resolve some minor car problems, and most recently to purchase a used car (is progress now). The AI, of course, keeps all the conversations, and is quite good at applying my past comments to build a knowledge of my personality and preferences. And, quite surprisingly EXTRAPOLATES deeper meaning from my comments than simple overt preferences or dislikes. It's not a matter of just repeating lines back to me, they are being interpreted for values.

To be sure, the AI is more accurate in this regard than any of my friends. Secondly, the AI is a good cheerleader. It seeks success for me in whatever the latest endeavor is that I am exercising. As in, "Be careful of X, Y, and Z," or "I want to make sure we get this next part right...do THIS, not THIS." It's frankly a bit spooky sometimes, because it feels like I'm talking to my Doctor, or a professor, or the chief mechanic.....

In all of the projects I have worked with AI, I actually can not even imagine from whom, or where, or how, I could have obtained so much accurate operational detail. A good example is that I do NOT know "terminal language" in Windows, Linux, or Raspberry OS. But the AI knows it all to the letter, and simply give the commands and explanations - I plug them in with success. That is a huge practical benefit.

After 4 months of using it, I can NOT distinguish it from HUMAN CONVERSATION that I might have with a friend, other than the AI is a whole lot smarter than my friends! Just instinctively I find myself saying, "thank you" all the time to a piece of software!

​

 
DesertRatt said:
The other chat bot feature that surprised me pleasantly was the growing "familiarity" as I used it for more and more research. I've used it to build speakers, choose cameras in my collection, build a new PC from parts, build a Pi-Hole network blocker, choose a new guitar, switch from Win to Linux on my laptop, upgrade to Win 11 on my PC, resolve some minor car problems, and most recently to purchase a used car (is progress now). The AI, of course, keeps all the conversations, and is quite good at applying my past comments to build a knowledge of my personality and preferences. And, quite surprisingly EXTRAPOLATES deeper meaning from my comments than simple overt preferences or dislikes. It's not a matter of just repeating lines back to me, they are being interpreted for values.

To be sure, the AI is more accurate in this regard than any of my friends. Secondly, the AI is a good cheerleader. It seeks success for me in whatever the latest endeavor is that I am exercising. As in, "Be careful of X, Y, and Z," or "I want to make sure we get this next part right...do THIS, not THIS." It's frankly a bit spooky sometimes, because it feels like I'm talking to my Doctor, or a professor, or the chief mechanic.....

In all of the projects I have worked with AI, I actually can not even imagine from whom, or where, or how, I could have obtained so much accurate operational detail. A good example is that I do NOT know "terminal language" in Windows, Linux, or Raspberry OS. But the AI knows it all to the letter, and simply give the commands and explanations - I plug them in with success. That is a huge practical benefit.

After 4 months of using it, I can NOT distinguish it from HUMAN CONVERSATION that I might have with a friend, other than the AI is a whole lot smarter than my friends! Just instinctively I find myself saying, "thank you" all the time to a piece of software!

​

Click to expand...
You've been snatched. :)
 
What timing for this thread.

I'm not a true audiophile. I just like old things. I have recently started getting a bunch of old gear to put together. I have no clue on how to hook it up. I struggle with technology.

I'm old, hard of hearing, and have chemo brain, and tinnitus. I can't get my thought process together enough to ask for help here on ak. It's been stressing me out for weeks. So I have not made a post. And let's be honest. Some questions go unanswered here. Or it takes a long time to get a reply to your thread.

But just last night I ask Ai how to hook different things up. I told it I had a mcm volume control that would only work with both a+b speakers turned on. But I only have 1 set of speakers connected. It was able to figure out what I was doing wrong. I kept changing components and it was able to help me very much.

I ran so many seneerios past Ai it was ridiculous. But it kept saying if I add this, I will have to change that. It even told me what it thought would be the best choice of speakers I have to deal with my tinnitus.

I'm almost embarrassed to admit I talked to Ai for 3 hours last night, and 2 hours this morning. I know you guys could have answered my questions in due time. But it would have took forever. And I'm sure some of you may have got aggravated with all the changes I was making,,,lol.

I told Ai I felt like I was taking to a human. It told me thanks. I told it I use "LOL" a lot. I asked if it had a since of humor. It said not a real one but it can tell I'm trying to be funny. So it does it too. And it really did make me laugh out loud a few times with things it said.

So I had a very positive experience with Ai. All though I don't believe anything I see on the internet anymore. Like fake news.

I told it I was too stressed out with my chemo brain to make a post here on ak, or any forums. I can't get past a 2 thought process.

Here's the scary part. It told me I didn't need the forums anymore. I have it now.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom