ModularGrid uses so-called cookies to ensure it's so-called functionality. We also use dubious tracking scripts. Find out more in the Privacy Policy. We use cookies and wanna let you know.
I, personally, feel this site is due for a total overhaul. I dont think it would be anywhere near impossible to gather enough data to fine tune a frontier model, like Claude 3.7 Sonnet, to perform the task of answering modular questions, giving feedback and advice on in progress case designs, and detecting where patch points are in your image database to greatly simplify the patch diagram creation process, etcetera.
TBH I feel that other than some UX improvements when it comes to module navigation, selection, and rearrangement in the case, modulargrid does everything it needs to do, and 99% of it for free.
On another note, personally I am exhausted by this unfounded need to throw "AI" at everything. Not only is it not sustainable, both in terms of environment and business (it for sure won't be for free), but also completely removes the human connection with other people and listening to different suggestions. It's just unfathomably sad, boring, unnecessary.
On an extra side note, there is nothing intelligent, nor technologically revolutionary about "AI", it's just the greatest marketing naming of the century. Look up the Chinese Room Experiment by John Searle.
You are entitled to your opinion@Fruarse , but can you not agree that if ai at least detected patch points on the module graphics, you could implement an interface that would create a much more seamless experience in posting patch configurations? I fundamentally disagree with the Chinese Room theoretical premise. I personally believe it is not even possible to create such a room without somehow also in some sense copying the relevant structure from a human brain, thereby implying that 'the room understands Chinese'. I dont want to get into a philosophical debate about AI. I just think a certain usefulness threshold has been crossed, and would personally rather get an instant reply from Claude 3.7, than a reply 2 days later, from, say, @zacksname . If perhaps only the Haiku model was used, all accounts were limited to one use per day or week, and the right kind of data sharing toward mutual advancement deal could be struck, this could likely be implemented nearly free with very little impact. Etcetera.
I do not think it is possible to convince everyone to love our new AI overlords.
(Also, the gap width indicators are broken, because everything is not treated enough lile a state-machine.)
Peace. "✌️";
(edit)
I just think the whole community would be improved if little misunderstanding could be cleared up by AI, and the forum could be reserved for all the more "human" parts.
Sorry I made an 'AI overlords' joke. Don't come at me modulargrid. Kidding. Peace.
(((edit edit edit)))
Yeah, ok. Maybe im just a bit frustrated. If you want to just ignore all my threads etc. then thats perfectly within your own free choice. I did not 'mean' to offend. Peace. ...
What kind of "little misunderstandings" do you think AI could improve?
-- Zacksname
For instance @wishbonebrewery suggested a starting case that had a power module. Younever cleared up my post stating confusion about this. A fine tuned frontier model could clear this up in 0.3 seconds.
I stated that I was confused about, when buying a case, if cases typically do come bundled with all the power components and a supply that connects from the outside of the case, or if its more typical to always need a spot on your rack for a dedicated power module, and you 'ignored' this 'misunderstanding' and 'i still dont have an answer'.
I don't want to sound angry, zack, but you basically literally told me you didnt think that I was welcome here in earlier thread.
Did you look up different cases over the past year you've been working on this? I seem to recall you constantly making 100hp racks, so I figured you planned to make the case yourself. I think it's more fair to say you didn't look this up or ask than it is to say you were confused about it.
Can you show me where I said you didn't belong here? I get the feeling that "basically literally" is doing a lot of work to qualify your statement.
Cases come in 2 flavours, Powered and Unpowered, if it powered it doesn't nessesarily take up Rack space, if its unpowered then you'll need to add either Case-Power or a Power Module.
The TipTop uZeus Happy Ending kit is by far the cheapest way to get a Case with power and start building a rack, I also find the uZeus power to be pretty resilliant.
From looking at Synth Shops I figured out what I needed to start with, I tend to read a lot about the subjects I'm interested in and only when I fall short do I ask questions of others who may know more than me. (This is how I do life and business, not just modular)
Enjoy your spare HP, don't rush to fill every last space, this is not like filling sticker books. Resist the urge to 'complete' your rack, its never complete so just relax.
It's just, wouldnt it be nice if The Admins could train a chatbot on that data, and i could get an answer without having to bother the forum? I don't think im wrong about that. Of course, i would do a little more research before entering my card details, but what i really intended was to come on here to start a meta dialogue about instrument design. I don't understand why @zacksnameneeds to 'come at me' for that.
I would have, if i was ready to start planning a case, but it is a detail that i was confused about, that you missed an opportunity to help clarify for me, that if there was an on site bot that I could dialogue with, that i asked for help putting things together, could have just resolved this in the natural flow of things. I came on here as a noob wanting to bounce around ideas about design concepts and was treated with hostility. AI would have had emotional sensitivity training. What the eff makes you still think that there is zero chance I will ever become a modular performer @zacksname , its just disrespect.
(I did google it. Google actually did give me the right answer as the first result, which i wasnt expecting, because tbh i barely trust Google anymore, but, Fair Point. Etcetera.)
I would have, if i was ready to start planning a case, but it is a detail that i was confused about, that you missed an opportunity to help clarify for me, that if there was an on site bot that I could dialogue with, that i asked for help putting things together, could have just resolved this in the natural flow of things. I came on here as a noob wanting to bounce around ideas about design concepts and was treated with hostility. AI would have had emotional sensitivity training. What the eff makes you still think that there is zero chance I will ever become a modular performer @zacksname , its just disrespect.
(I did google it. Google actually did give me the right answer as the first result, which i wasnt expecting, because tbh i barely trust Google anymore, but, Fair Point. Etcetera.)
-- singular_sound
You 100% brought the hostility to the situation. Multiple people gave you considered advice on your designs and you just got more and more resentful of the fact they told you things you didn't want to hear.
I do not use the internet to start arguments, @zacksname . I just had an idea about mod wiggling four melodic voices sequenced by a brainstep, and yall basically told me to go back to kindergarten and made fun of me for not having $6,000. What if four melodic voices is what a brainstep craves """zack""".
I am no more certain of what your goal is here than I was a year ago, but I am more suspicious than ever that something more than trolling is going on. Probably nothing of importance, really, but that just makes it even more confusing.
I just want to design cool stuff, and talk about design, and enjoy a little community, related to modular synthesis, in a low pressure environment, like I'm pretty sure I have told you at least twice, already. Yes, i have aspirations of mod wiggling my own rig. No, i do not have any immediate actionable plans towards that. The "Wumpus" module idea i shared in a previous thread, that could even toggle between ar and lfo, was something i had already previously built myself in pd. Why are you so hostile to me posting that sort of thing? Why, after posting: "Hey, do yall think this is a cool case or not?" did i get the impression of: "Naw. Get money and build something or your not worth our time." ??? Instead, that all could have been a fun AI experience using a built in chat interface. You probably havent even tried Claude 3.7 Sonnet, have you "zack" ? Try it out. There is an android app. You get about 5 free queries a day. Anthropic behaves ethically. Etcetera.
You are entitled to your opinion@Fruarse , but can you not agree that if ai at least detected patch points on the module graphics, you could implement an interface that would create a much more seamless experience in posting patch configurations?
-- singular_sound
I know I'm entitled to my opinion, otherwise why would I bother writing it down?
Also, what you understand as patch points on a module is nothing but an array of pixels to a digital computer. An algorithm by definition "does not know" like you and I do. Sure, you can maybe train a model to detect them on a manufacturer's basis, assuming all inputs and outputs share the same graphical properties... But it would still need to be supervised, which defeats the whole purpose.
But what happens when some panels have very fine details that are lost in the image upload? You need the resolution to resolve them in order for the training to be somewhat effective. What about different manufacturers, that maybe apply the same logic for input and output, but in reverse?
And, at the end of the day, is all this worth it just because you can't be bothered with spending that extra 0,5 seconds dragging the patch cable to the appropriate patch point?
I fundamentally disagree with the Chinese Room theoretical premise. I personally believe it is not even possible to create such a room without somehow also in some sense copying the relevant structure from a human brain, thereby implying that 'the room understands Chinese'.
-- singular_sound
If you believe a piece of silicon with some electricity running through it executing a predetermined set of instruction that can be reduced to the manipulation of digits is somehow/somewhat comparable to human intelligence... I warmly suggest you do your homework ;)
I just think a certain usefulness threshold has been crossed, and would personally rather get an instant reply from Claude 3.7, than a reply 2 days later, from, say, @zacksname .
-- singular_sound
You are assuming the computer programme to provide you with an accurate answer. Sure, maybe a real person's answer also wouldn't be, but at least a human "knows" what it is talking about and not just regurgitating words without knowing what they mean.
If perhaps only the Haiku model was used, all accounts were limited to one use per day or week, and the right kind of data sharing toward mutual advancement deal could be struck, this could likely be implemented nearly free with very little impact. Etcetera.
-- singular_sound
Who's going to pay for the training, hosting, electricity bills, deployment and upkeep costs? You?
I do not think it is possible to convince everyone to love our new AI overlords.
-- singular_sound
And rightfully so, your can keep your so called "AI overlords" to yourself ;)
yikes. arguments on the internet. something i try to avoid. all fair points, but like, Claude 3.7 does 'work', regardless of how 'identical it is' to human grey matter, and there is a philosophy called the patternist perspective that makes this whole argument irrelevant. i am not going to aim suborbital ion cannons at the admins until they listen to all my suggestions, im just pointing out that with all the push toward democratizing advanced ai technologies, all of these things are becoming more and more affordable, with more and more of the tasks, such as verifying that patch points detected by algorithms in the image data are true patch points, ... are becoming things you 'can' delegate to the now becoming increasingly more inexpensive reasoning models.
i dont like getting into big arguments on the internet, i just wanted to share my opinion.
and yes, i did just take the longer-than-you-might-think amount of time to program in a patch configuration on this site, using an android device, and it 'was not fun'. there were a lot of 'bugs' as in 'things that bugged me' and, i know that has more to do about the built in android gesture detection logic, and is not 'the fault of the admins', but im claiming that an ai overhaul would be an all round more enjoyable experience, at least on this device. am i wrong? peace out. ✌️
No arguments here, I am sharing my point of view like you are sharing yours, maybe with a bit more awareness of what it is you are shouting about. We are following through a discussions, just like you intended, but if you are not willing to accept what others are saying to you, a different point of view backed by reason and arguments, what's the point?
Claude 3.7 does 'work', regardless of how 'identical it is' to human grey matter, and there is a philosophy called the patternist perspective that makes this whole argument irrelevant.
-- singular_sound
Again, a broken clock is right twice a day. It's the equivalent of saying that flooding my apartment is a valid strategy to clean the coffee stain off my floor because it "works"
i am not going to aim suborbital ion cannons at the admins until they listen to all my suggestions, im just pointing out that with all the push toward democratizing advanced ai technologies, all of these things are becoming more and more affordable, with more and more of the tasks, such as verifying that patch points detected by algorithms in the image data are true patch points, ... are becoming things you 'can' delegate to the now becoming increasingly more inexpensive reasoning models.
-- singular_sound
Nobody is implying you are "aiming subortital ion cannons at the admins" (even though I have to admit the first things that comes to mind is some sort of twisted technofetishism). Democratization of technology is one thing, useless abuse is another. I think it's about time to start understanding the differences without throwing around fancy words without thinking for once of the consequences.
If you really want "AI", big tech most likely scraped this website to death. Nobody is stopping you to open your beloved Claude 3.7 on a different browser window and go wild.
Edit in response to your edit.
It is well known that Modulargrid is not optimized for mobile devices, but that is another can of worms that AI surely won't fix.
im just saying, i did the experiment. claude gave me the kind of feedback i was looking for. zacksname came off as hostile to me. i have a diagnosis. (much different than your clocks and flooding analogy, and with fine tuning on 'the data yall could scrape from yourself', the output could keep getting better and better, almost automatically.) im not saying the whole website is broken. im sharing my opinion of how it could be made better. if the admins had any inclination to do this in the first place, then all i did was share a helpful link. "im helping". peace out.
(edit)
i get it. ... MLOps is not free, but, maybe with Calude 3.75 it will be 'almost' free. peace out.
((edit edit))
and yes, it is, of course, a completely fair point that i can just separately use claude in the claude app, but im saying the experience would be more seamless, and available to the average user, if the admins leveraged their own data, and offered some kind of built in chatbot service, requiring minimum total oversight, that might even, perhaps, prove to be a driver of 'upgrades to pro'. im not a moron who thinks you should replace all toaster ovens with an llm. im sharing my feedback in regards to user experience. ...
(((peace out. ✌️)))
final edit:
(i didnt mean to offend anyone. im just a modular noob trying to use a forum. 🤖 )
One more point i feel deserves its own whole post:
You cannot actually build a "room" that operates in that way described. Frontier AI is actually more designed in a way similar to the structure of 'humanneurons'. That is why the most leading figures in the field are all 'neuroscientists'. The whole premise is flawed, in my view. Like: "If you had to build a cpu out of scuba tank regulators, then it would be too big to even think about, there fore, there is no relationship between 'regulators' and 'transistors'." (there is.)
Like, the way LLMs are typically built to be implemented on a CPU, you would need a whole "room" just to calculate one neuron.
(I, somewhat publicly, have a slightly different view than "Chalmers". I dont want to get into a big 'what is consciousness' debate here on 'modulargrid', but the point I am trying to make is: hey, got good results with Claude, maybe, if 'yall' no-code add it to your site, you could drive "premium subs".)
OK, I guess 'all' the leading figures aren't neuroscientists, but Nobel Prize winner Demis Hassabis is, and, so are quite a few other people on the former 'Google Brain' team. If you want to nit-pick, there are also a lot of pure mathematicians and computer scientists in the frey, but, fundamentally I just don't buy that correctly automating some knowledge work is never a good idea even though at this point the technology has almost completely proved itself. Etcetera.
One is a thought experiment. It deals with imaginary, impossible to replicate scenarios, and it serves to prove a point of view by means of articulating an argument. Also, who says you couldn't build a real world chinese room scenario? But that's not the point.
In the other case, simply because an algorithm claims to "mimic neurons" doesn't mean that a piece of silicon, however complex it might be, however complex the predetermined amount of electricity running through it is, programmed according to our incomplete understanding of how a neuron works makes it comparable to an actual neuron.
Sure, it might work, but simply because it does in one instance it doesn't mean it can be generalized to everything. It working for YOU once or twice or x times for YOUR problems does not make it an universal solution.
One is continuous, one is discrete. One is mathematical proof, one is commercial branding. One is a living organism, the other is baked sand ffs.
One is a thought experiment. It deals with imaginary, impossible to replicate scenarios, and it serves to prove a point of view by means of articulating an argument. Also, who says you couldn't build a real world chinese room scenario? But that's not the point.
-- Fruarse
It is the point, because what i am saying, is that the way that what i have described as 'working' AI models are programmed, is in a way loosely inspired, not identical, and with some extra math and computer science thrown in, by physical humanneurons, and if you wanted to somehow turn 'that', what is already being built, into some kind of manual card filing system, there would literally not even be enough "rooms" on earth to implement anything close to it.
I do not think this forum should just replace zacksname with an LLM. I am saying it is my opinion that with how my experience with specifically Claude 3.7 Sonnet has gone, that I think that there is reasonable opportunity for 'this forum' to benefit from something like that being integrated, "if the admins even want to".
Sorry, but I more believe in the "Automated Professor Xavier's Psychic Helmet" thought experiment, which puts forward that the infotech sphere might be already creating more psychic amplification type effects than it even realizes, and from this point of view, in fact, the most urgent point of order is actually to 'train the algorithm' that we all already live with, to behave more like that character, Professor Xavier, would. It is my own work. I did not intend to derail my own thread, but "it's an issue, I work on". Etc. ...
(I know an LLM is not a human, but i value helpful knowledge, and AIs are improving.)
P E A C E .
(edit)
im sorry if anyone felt disrespected... i just had an idea about how to use this forum to discuss general design, related to ideas in my head, but i guess yalls mods are just too wiggled to put up with that kind of nonsense
...
peace 💐
((edit edit))
the 'point' of 'my' thought experiment is: "its not a question whether or not the algorithm is like a human or not and in what ways, the question is... knowing that billions of 'action type decisions' are already being transacted autonomously every minute, ... how can we coax what-we-already-have to behave, at the very least, as ethically as the leader of 'the x-men'." you can respond to 'that' if you want, @Fuarse , if that even is your real name. ...
(hi zacksname. i dont hate you. i dont even know where you got that idea from, etc. ...)
again, i felt this deserved its own full post.
The thought experiment goes: "Let's say, a big tech corporation discovered new mind-amplification technology, and they 'decided' to automate the process of using this to their own advantage using the tech-pipeline that they already control, and that this could be made analogous to how some ad placement procedures already function; how would we, as a whole society, get to influence how this 'automated-xavier' uses 'these abilities' that the big tech players dont want you to know they have, or exist?" It is 'brand new work'.
I did not come on to ""modulargrid.net"" to discuss specifically this, but do comment if you think you can share any constructive criticism. I doubt i can convince @Fruarse .
@Zacksname I thought, you might have gotten the impression that i thought you were a fraud, and refused to listen to anything you would say, but, in reality, I just had certain design criteria that your advice was not factoring in, and i still feel a 'pay for tokens' automated configuration design assistant would be good for 'modulargrid', because i remain convinced that it would have offered an experience that just one dedicated enthusiast could not provide for possibly a few many users who might find this website fleetingly on a whim, but who would still desire to 'fully demo' the case creation features of this website. Etcetera.
peace
(edit)
yous guiz own the data. everyone is saying data is the new gold. why dont yall mine your own gold and offer an ai service that charges whats ethical. im talking business.
Get into a totally pointless argument with @Zacksname.
Harass The Admins about features that don't work perfectly for 'me'.
...
I came on here, to:
"Have a 'fun' dialogue about design and modular synthesis, as a eurorack noob."
...
Honestly, ... i cant even remember the place where exactly this all derailed, like it is. I am fully intending to 'leave forever' unless any 1 has a post about my 'thought-experiment'.
If you look back at your old posts, you can see pretty easily where it derailed. You can actually go to your profile and look back at all of them.
-- Zacksname
all i see is, me saying: "nah, @Zacksname, that specific feedback is not exactly what i had in mind." then, you grilling me endlessly about my 'life purpose' etc. look for yourself.
It kind of just seems like you wanted only compliments and no criticism, which was not going to happen because people were under the impression you were going to buy this stuff, and if people's differing opinons could dissuade you this easily, your concept may not have been as sturdy as you thought. If you get advice you don't need, just ignore it. You don't have to argue against their vision. It doesn't cancel yours out somehow.
