I asked AI to design a bass cab

[Admiral Akbar]

This thread is what concerns me about people and AI: it's not the technology that scares me, it's the misuse of it.

 

[JazznFunk]

This thread is what concerns me about people and AI: it's not the technology that scares me, it's the misuse of it.

All of it scares me. Not all technology is an improvement, or a step forward. The people part scares me because there's absolutely no pragmatic skepticism.

 

[Callused Finger]

I can't wait to read @agedhorse's comment

We are lucky he doesn't mind typing.

 

[agedhorse]

AI can be valuable as a tool screening out poor choices or identifying interesting candidates that might have been overlooked, but it doesn't replace the good judgement of an experienced designer who understands what their players are REALLY wanting (which includes a LOT of interpretation between the lines).

The problem is when folks blindly accept what AI delivers without considering that it may be nothing but well curated and presented bull-poopie.

 

[Stylee Mousse]

The real immediate risk to my mind is the short time it will take for the people with the expertise to monitor and understand AI output to age out, NEVER to be replaced.

I see a crossing X/Y axis…

 

[Al Kraft]

It's easy to design a crappy cabinet, it's difficult to design a really good cabinet.

The corollary to which is that virtually anyone, can design almost anything, that will work poorly if at all.

 

[Pep]

Vented speaker design is pretty well understood so the AI can crank through a lot of options which gets you into the ball park.

No it gets you to a championship golf course with the AI design and a properly designed cab 18 holes apart.
There are people asking AI medical questions, believing that everything from AI is the truth. Many have sadly taken it's advice and are much worse because of it.

 

[getBAK]

Wouldnt surprise me one bit if AI was already involved in some form in Cab and Amp design.

The line between automated CAD design and AI is very blurry. And really, why not use all of the tools at your disposal.

I Think of an AI as a really complex calculator (Gen Xr's, remember the TI-82). In the right hands it is an awesome supplemental tool. But if you don't know how to properly input or interpret the output data, it's just a screen that you can write 80085 on.

 

[agedhorse]

The line between automated CAD design and AI is very blurry. And really, why not use all of the tools at your disposal.

I Think of an AI as a really complex calculator (Gen Xr's, remember the TI-82). In the right hands it is an awesome supplemental tool. But if you don't know how to properly input or interpret the output data, it's just a screen that you can write 80085 on.

I disagree, it's a data parser that makes bad data look good by the way it interacts and presents its output.

The data it relies on (in an open trained model) is just as terrible as the information it trains with on the internet, as well as unproven literature. This is a big part of the GIGO model, not just the questions you ask but the information that the question are answered with.

As in all specialized fields, the more of an expert you are, the more quickly you will recognize how bad much of what's presented really is. Once the training is done on verified models (closed sources), and ignores "urban mythology data@, the better it will get.

 

[Pep]

I disagree, it's a data parser that makes bad data look good by the way it interacts and presents its output.

The data it relies on (in an open trained model) is just as terrible as the information it trains with on the internet, as well as unproven literature. This is a big part of the GIGO model, not just the questions you ask but the information that the question are answered with.

As in all specialized fields, the more of an expert you are, the more quickly you will recognize how bad much of what's presented really is. Once the training is done on verified models (closed sources), and ignores "urban mythology data@, the better it will get.

What he said!

 

[Al Kraft]

The line between automated CAD design and AI is very blurry. And really, why not use all of the tools at your disposal.

I Think of an AI as a really complex calculator (Gen Xr's, remember the TI-82). In the right hands it is an awesome supplemental tool. But if you don't know how to properly input or interpret the output data, it's just a screen that you can write 80085 on.

I think CAD works with a much more reliably curated and mathematically sound set of capabilities than the AI that most people are familiar with. Somebody posting absolutely wrong information on X or Reddit (which almost never happens...right?) won't be accidentally integrated into a CAD program result.

While MBSE tools like Cameo can often ingest data from all kinds of models and sources (I work with some folks that developed and are maturing an AI augmented capability for this purpose), it requires deliberate intent on the part of the modeler to select specific source data, not rely on refining prompts and hoping the sources are credible.

I am not a direct AI user in my work but am very familiar with integration of AI type capabilities into weapon systems, specifically multispectral sensor fusion to inform rapid tactical decision making. These AI capabilities are also trained, but on specific data sets so as to ensure the displays provide decision quality information to operators.

 

[Stylee Mousse]

Where I work there is a lot of emphasis on using AI. Unfortunately, it is not able to access the data it would need to use to actually be useful.

If it did have that access, I could use it for basic data retrieval and display, as long as it wasn't allowed to be rendered useless with the opinions of the many headed, the great unwashed, etc.

 

[TTA]

Out of (morbid) curiosity, I modeled the recommendation of 2CF tuned to 50Hz just to see how close the math and the word salad agreed... this isn't exact math but plenty close enough

View attachment 7440644

Average sensitivity is ~99.2dB/1W/1M
F3=80Hz
F6=50Hz
F10=42Hz
Mechanical power handling is about 200 watts RMS above 50Hz

So you can quickly see that the AI predictions are way off. Probably because it trained on poor, unrealistic designs that builders who are clueless have bragged about enough that it becomes training material. It's really pretty junky.

never used ai for such things, but garbage in = garbage out as they say.

i have used it for other topics i was trying to learn about, and an as a non expert i could easily identify issues/failed logic.

topics i do know about… i can verify ai is a mess. it can do basic info easily, but as soon as things get into complex design it falls apart.

but search engines are now defaulting to ai responses.

 

[Al Kraft]

I'm not at all against technological progress. That said we have to remember that all new technologies bring with them vulnerabilities and consequences that may not always be well understood or even known in the early adoption phase. As an example, asbestos was a miracle product until we got a little more experience with it. We have also seen miracle drugs and agricultural chemicals that turned out to be worse that what they were fixing.

I'll simply offer that the AI community came up withe term "hallucinations" (not me) and here is a good article that addresses that aspect in a what I consider a balanced and accurate way based on my own experience: What Are AI Hallucinations? | IBM .

 

[agedhorse]

never used ai for such things, but garbage in = garbage out as they say.

i have used it for other topics i was trying to learn about, and an as a non expert i could easily identify issues/failed logic.

topics i do know about… i can verify ai is a mess. it can do basic info easily, but as soon as things get into complex design it falls apart.

but search engines are now defaulting to ai responses.

The search engine defaults give people a false sense of security and self confidence. This is perhaps the most dangerous (consumer) aspect currently.

 

[Al Kraft]

The search engine defaults give people a false sense of security and self confidence. This is perhaps the most dangerous (consumer) aspect currently.

+1...the reason the internet appears to be free to users is that the price we pay is well hidden from us. Companies can pay to have their results show up first when those results may be relevant but not even close to an optimal or even accurate answer. You also quite often get "most frequent/popular" (i.e., the one that sells the best) answers over an answer that might support you doing your own research to come to your own decision on what you want. I have met the people who do this work behind the scenes.

The internet is a unique type of business model without any real rules of commerce or ethics, albeit I would agree that is does have protocols for timing, data movement, etc. It serves the paying customers first in a way that will often be at the expense or to the detriment of those who think they what they are getting is "free access to a carefully curated, vetted, and/or peer adjudicated set of reliable information".

I'm not negative on the internet, or chainsaws,...I just have a healthy respect for their designs and accept that not exercising care in their use can have serious consequences.

 

[Aidenn7]

I was looking a panel speaker DIY.. That plus tactile transducer platform is the only thing I can think off to emulate the effect you after but there are obvious downside

 

[TTA]

The search engine defaults give people a false sense of security and self confidence. This is perhaps the most dangerous (consumer) aspect currently.

yeah exactly. i was discussing the topic with my wife today and actually used this thread as an example. i mentioned there are folks who build/design cabs/amps here for well known brands, and some very well informed folks. identify them and listen. AI does not know who is who… it just compiles data.

but take an ai and fill it with well known studies on all the design parameters/equations, and it will be able to give you what you want. then your ear just needs to be the deciding factor. And that accounts for a lot.

 

[TTA]

+1...the reason the internet appears to be free to users is that the price we pay is well hidden from us. Companies can pay to have their results show up first when those results may be relevant but not even close to an optimal or even accurate answer. You also quite often get "most frequent/popular" (i.e., the one that sells the best) answers over an answer that might support you doing your own research to come to your own decision on what you want. I have met the people who do this work behind the scenes.

The internet is a unique type of business model without any real rules of commerce or ethics, albeit I would agree that is does have protocols for timing, data movement, etc. It serves the paying customers first in a way that will often be at the expense or to the detriment of those who think they what they are getting is "free access to a carefully curated, vetted, and/or peer adjudicated set of reliable information".

I'm not negative on the internet, or chainsaws,...I just have a healthy respect for their designs and accept that not exercising care in their use can have serious consequences.

my line is "you can learn how to do brain surgery in the internet… the trick is sorting out the garbage first."

 

[TTA]

I'm not at all against technological progress. That said we have to remember that all new technologies bring with them vulnerabilities and consequences that may not always be well understood or even known in the early adoption phase. As an example, asbestos was a miracle product until we got a little more experience with it. We have also seen miracle drugs and agricultural chemicals that turned out to be worse that what they were fixing.

I'll simply offer that the AI community came up withe term "hallucinations" (not me) and here is a good article that addresses that aspect in a what I consider a balanced and accurate way based on my own experience: What Are AI Hallucinations? | IBM .

i was using ai to help me learn about high performance insulated walls. it helped a lot and i was learning about internal drying, external drying, code, dew points of condensation, etc etc.

it helped me identify some design flaws of my wall, but was also trying to tell me to tape foil backed polysio boards with the tape facing the wall. i called it out on it… and it was like… yeah you are right. i also asked it the same questions a month apart and if gave me different solutions. what was a very bad idea the first time it was saying was fine the second. The frustrating part was that most contractors did not understand the concepts either. so it came down to me doing my due diligence and saying "build this" which was not the process i envisioned.

and dont get me started on using ai to find corporate class etf's for canada. what a joke. but i repeated it two months later and it got it right.