In the old platform there was the "who posted' feature. When you clicked on the number of total posts of a thread, a list of all members posted there would pop up together with a number of their total posts. Then choosing one would display their posts only. I'm really missing that feature...
A "histogram" would display with bars corresponding to each participants # of posts, ranked from most to least. Just kidding, but it's a computer might as well give it something else to do. An AI might frown on that...
On Threads that you have commented on before, it shows your tiny avatar on top of the Starter's avatar. By clicking on your avatar it will list all your previous comments on that thread just as refresher on what was said before for reference.
diyaudio.com seems like the perfect deep, mature, knowledge repository for training an AI framework so natural-language, questioning searches can be done. Instead of using just the current crude word/phrase searches. There are several open source AI frameworks out there. The AI model should train itself exclusively from diyaudio content, but don't include the Internet at large, it will just go off on un-necessary tangents.
Is an AI framework integration on the development roadmap for diyaudio.com?
Is an AI framework integration on the development roadmap for diyaudio.com?
AI?
Smart people recognize the word ID10T. AI does not understand the meaning.
'AI has to know its limits' ('quote' stolen from Dirty Harry movie, and modified to taste).
Just my pessimistic opinion
Or, sounds like something from the state of Missouri (the show me state).
Smart people recognize the word ID10T. AI does not understand the meaning.
'AI has to know its limits' ('quote' stolen from Dirty Harry movie, and modified to taste).
Just my pessimistic opinion
Or, sounds like something from the state of Missouri (the show me state).
AI can greatly improve searching
6A3sUMMER,
I understand where you're coming from, the market hype over AI being like Einstein is ridiculous. But I've worked with it in my industry (truck mfg) in truck diagnostics, and design, it has merit, lots of merit and usefulness. Heres how, basically its still garbage in garbage out like any software. But when a neural network is trained entirely from a single source of truth, like diyaudio, or a companies engineering and service documentation. It becomes an uncannily and scary good "librarian". I believe that the folks on diyaudio are largely very smart and creative, so what better training bed for a neural network to back up the currently pitiful forum search. A neural network can focus threads down to the question of interest. For example try searching for knowledge on the forums specifically having to do with "CCS as used in anode circuit". Hundreds of unrelated posts that briefly mention "CCS" at all result. The result is basically useless unless you have all day to now manually read them all. A neural network however will indeed discern discussions about CCS usage in HV anode circuits from cathode and other usage. It might even answer your question specifically. Or just ask it a question like "list all schematic diagrams that use a 2A3 tube for me", it will do that too. It does it for our 2,000 dealer technicians in other ways when they need harness or other schematics on older product.
But the neural network must be trained from a somewhat reliable source, not the Internet at large. At work all of our dealer technicians now diagnose the 50 or so embedded controllers on our trucks using a neural network trained in house, the feedback is anything but "its an 1d1ot". But I fully understand where that comes from if one uses ChatGPT to try to do a diy audio hobby. A neural network sitting under the reliable talent, wealth and years of knowledge here at diyaudio can be a scary good librarian for diyaudio, I guarantee it.
6A3sUMMER,
I understand where you're coming from, the market hype over AI being like Einstein is ridiculous. But I've worked with it in my industry (truck mfg) in truck diagnostics, and design, it has merit, lots of merit and usefulness. Heres how, basically its still garbage in garbage out like any software. But when a neural network is trained entirely from a single source of truth, like diyaudio, or a companies engineering and service documentation. It becomes an uncannily and scary good "librarian". I believe that the folks on diyaudio are largely very smart and creative, so what better training bed for a neural network to back up the currently pitiful forum search. A neural network can focus threads down to the question of interest. For example try searching for knowledge on the forums specifically having to do with "CCS as used in anode circuit". Hundreds of unrelated posts that briefly mention "CCS" at all result. The result is basically useless unless you have all day to now manually read them all. A neural network however will indeed discern discussions about CCS usage in HV anode circuits from cathode and other usage. It might even answer your question specifically. Or just ask it a question like "list all schematic diagrams that use a 2A3 tube for me", it will do that too. It does it for our 2,000 dealer technicians in other ways when they need harness or other schematics on older product.
But the neural network must be trained from a somewhat reliable source, not the Internet at large. At work all of our dealer technicians now diagnose the 50 or so embedded controllers on our trucks using a neural network trained in house, the feedback is anything but "its an 1d1ot". But I fully understand where that comes from if one uses ChatGPT to try to do a diy audio hobby. A neural network sitting under the reliable talent, wealth and years of knowledge here at diyaudio can be a scary good librarian for diyaudio, I guarantee it.
Last edited:
Intriguing idea, @Windcrest77
While I can’t see Jason putting that on his plate anytime soon, I’m sure it would work well.
An old friend of mine is head of service documentation for a private jet manufacturer. I wonder if he is doing like you? I’ll reach out to him after the new year.
While I can’t see Jason putting that on his plate anytime soon, I’m sure it would work well.
An old friend of mine is head of service documentation for a private jet manufacturer. I wonder if he is doing like you? I’ll reach out to him after the new year.
As far as I know yes, each person can only send a message back to you. They can not reply to each other.
What happens next though (if you then reply back to one of them) I'm not sure about. I suspect they all will then see your reply to that one individual as all are still in the 'group'.
What happens next though (if you then reply back to one of them) I'm not sure about. I suspect they all will then see your reply to that one individual as all are still in the 'group'.
This is the case for regular members, not forum sponsors. I apologise for any confusion.
I had once asked, years ago, whether the search parameters could include the option that only comments containing an attachment (PDFs, images, etc.) be returned from the search. To which the answer, at that time, was no. Is that answer still true today. I, for one, would find such a search feature very useful.
