I love the way “the future” is largely reified as gamer lighting 😂 it also ends up constituting quite a harsh criticism of generative AI - the most futuristic thing it can show you is a nonspecific, slightly worse pastiche of the present
I don't know much about Grok specifically, but most AI machine learning is based on what already exists, especially in larger numbers as to what represents popular culture - therefore these systems seem to excel when you ask it to make a trap beat with pop vocals, not only because those are musically easy and formulaic, but because also it's been fed such material in excess to learn from that it's effectively mastered many "modern" styles of popular music. However, ask these systems to compose something in the style of classical (Mozart, Beethoven), it will get some of the features correct (it seems to understand tonal harmony no problem), but it can struggle with orchestration and will conflate things like "solo violin" with a full string orchestra or even a solo woodwind instrument. Ask it to compose in the style of later/romantic-era, and then 20th century experimental era like Berg, Penderecki, or Messiaen, it doesn't seem to really understand this all too well, which I can only assume has to do with the fact it has barely learned from their music to understand the possibile idioms behind their writing. What it will do, however, is create something pretty random - which on the surface, might seem what those later composers sound like, but this takes no account for the structures and concepts that fueled their writing. So it's not surprising to me that it struggles on the one hand with notation which is sadly fallen out of favor for DAW-style piano rolls - it has likely not been learning much from this, just as many modern humans don't seem to learn it either lol. And on the other hand, it doesn't surprise me that it can't peer into the future with a satisfying (and perhaps surprising) result - rather, showing us something which looks like a modern day "beatmaker"/producer using a DAW with those fairly typical neon pink lights everywhere (and oh god, just as you called out, all the cliché succulent plants!). Because this is what it has learned from as "modern", it seems to have difficultly imaginging beyond what it has been taught. The one around 17:15 made me laugh because if you look closely at all the screens, these actually look exactly like modern day video editing and color grading applications - and with all the visuals open, it seems a composer of the future is being conflated with a modern-day video editor lol. Another funny thing I observed in all the modern and "future" images the keyboards are essentially minature portable synth-style keyboards. Is that a prediction that composers of the future will minimize the importance of a piano as an instrument and compositional tool (which in some sense, seems to be the case with many producers)? Rather than simply getting smaller, I think in 100 years the piano-style keyboard may diminish in favor of touch surfaces and VR style control over note input - but because I have no idea what this looks like, it seems AI can't imagine it either 🫠
Thanks, Jonathan. Among other things, you anticipated (or overlapped with) some of my concerns written up in a comment that I wrote after yours but didn't see first.
I got some scary good results on another AI platform with conditions like "serialism", "vocal", "chamber" and then typed in my own text. It spat out something like Berg. The orchestration was also quite interesting. But it's not composing. I did do another test with just solo violin which I then orchestrated around the audio that Udio created.
I view Grok as not a prognosticating or analyzing tool but rather a new age and amazing art program. Every one of the images which were created in this video were magnificent and I'd love to frame them all and put them on my wall. Great for album art. The earliest ones with their strange manuscripts were quite magical, stuff of dreams. That was fun.
I get that these easily digestible ""jokes"" serve well their purpose as engagement devices in platforms such as these, however I do think there are better ways to go about displaying stereotypes to an audience other than validating them by putting them into practice live, an standard to which I held your content. Sorry if my demeanour comes off as upstart over minor things, but as a person with Asperger's this "sense of humour" admonishment comes off as a very poor excuse for secondary discrimination against composers with such unpopular stances nowadays as refusing the trends and believing the potential for tonal music is far from over, not merely as an inert foundation on which to build radically new genres and styles, nor as a dead weight to shake off in order to mandatorily pursue "uniqueness" in the craft, but as a tradition that still endures in its own right and deserves to be revitalized.
@FugaxContrapunctus If you don't have a sense of humor (and I'm not judging anyone on the spectrum here) then maybe don't feel compelled to post replies to comments/posts which are CLEARLY meant as a joke. If humor is not your thing (which is totally fine), then such humorous comments are likely not your thing either. Just ignore them and move on to other comments/posts with which you are better suited for interaction. It is not necessary to interject replies to all comments/posts that pop up.
There are a lot of elements to human intelligence, aren’t there? Those crazy images reminded me of something that I saw years ago. I taught junior high instrumental music for over thirty years. I once had a principal who came to school on a weekend and hand painted musical symbols and fragments of music on the wall next to my classroom door. While I appreciated his attempt to brighten up the drab hallway, my principal artist was not a schooled musician. Since our music studies included some theory instruction and assignments, I gave some thought to assigning students to identify ten errors in notation in Mr. P’s art. Alas, the idea was more amusing than practical and I imagined that insulting my boss wouldn’t have been a good idea.
"I tried GROK. Here’s what happened." This text, on UA-cam's page, was accompanied by a generic monster. I clicked on the image and there was no monster. It is not what happened, it was just click bait! Stop already. Don't become a hustler. We have more than enough of them.
X (Twitter) is licensing the new Flux image generator to produce images. When you ask Grok 2 to produce an image, Grok 2 merely hands down (sometimes refining it a bit) your prompt to the image generator. What the image generator produces has little bearing in what it is Grok 2 - the language model - understands about your request.
I had some fun with ChatGPT (DALL-E), using your prompts. It was rubbish at producing sheet music, much worse than Grok. No matter what I did it came out smudged and pretty unreadable. However, it did a much better job at drawing avant garde and futuristic composers. It seems that in the future composers will mainly use some kind of light rays when creating music, and in 1000 years composers will themselves be luminescent.
Something worth noting: since you stuck to image prompts, the results you got were not actually from Grok (which was very likely trained on tweets that speculated about the future of what music composition would look like), but from api calls to a separate image gen model Flux, which was only trained on images and whatever tags and descriptions accompanied them. Hopefully this illustrates why the knowledge base of Grok does not seem to be influencing the image outputs.
Brilliant concept Sam! I, for one, would be over the moon if you made a regular series/playlist out of this, and tested different models in these was, as they continue to likely improve!!
I work for a tech company and Samuel’s video is by far the most interesting thing I’ve seen yet in regards to AI generated responses. Love it! Please revisit in a few months.
1:50 What's interesting about this result is that it seems to imply that the source materials which it is drawing upon include both photographs of older handwritten Western staff notation and at least a few East Asian sources, specifically either Japanese shakuhachi notation or Ryukyuan kunkunshi-the few semi-legible letters appear to be the hiragana お and ゐ, the latter of which is obsolete in standard Japanese but would appear in Classical Japanese texts and is sometimes used in Okinawan and other Ryukyuan languages and thus would show up in kunkunshi; there are also suggestions of certain kanji which I think are common to both. That said, the fact that it is presented horizontally rather than vertically is something you see in Western notation and some Classical Japanese writing, but traditional Japanese forms of notation are, to my understanding, entirely vertical and right-to-left. Very odd result.
My first thought about the illegible symbols after the treble clef, is that Grok was attempting to imitate the bass clef's two dots, thinking that was some sort of rule.
I tried MidJourney a year ago, and I was getting results that reached roughly the same impasse that you have today. In short, our expectation is for imaginative results, but AI never has any comprehension of the symbolic meaning of the representations its working with. The brief, therefore, is to build an image based on the language used in the prompts, where the language indicates a link to the 'machine-learned' material, not the actual meaning itself.
Fun video. Grok is good a summarizing things in text, and generating fun images. Google has an AI that can write in the style of Bach. "Celebrating Johann Sebastian Bach Doodle." I wrote a piece with it and expanded / tweaked it, you can listen to it, it's called "Nine-Part Chorale for Human and Mouse" by BipTunia.
That clean shaved white haired composer closely resembled John McLaughlin. Maybe referring to specific examples (" in the style of") during the request could yield more desired or more interesting results.
It got that part right. Its mistake is to think that there are any contemporary composers worth considering at all. Male or female. You're barking up the wrong tree.
All it's doing is pointing a mirror at society. It's simply showing us the sort of cliché image of a composer that composers themselves and pop culture have helped create. If composers have something interesting to say then maybe AI will too.
Don’t ask for an image, that accesses a different model that lacks any learning of meaning. You might do better querying the text generation model that does map and learn meaning. This is true for chatgpt too.
If you try this again, I'd be curious to see what results you get when asking it to show you the notation of themes or motifs from famous works (Beethoven's fifth is probably the most obvious candidate). It would also be interesting to compare results from different text to image generators (Bing's AI image generator is also free, you just need an email from a Microsoft domain - outlook, live, hotmail, etc.)
I think this will run into the same problem as image generators have with generating text in images. Diffusion models have no semantic concept of what words, letters or symbols mean, so they get combined and distorted in very strange ways, leading to nonsensical, mirrored and incomprehensible text. The same thing is happening with the attempts at staff notation, which is really just another form of written language.
@@bahlalthewatcher4790 But there's a very good chance that there were some examples of the head-motif of Beethoven's 5th in the training data of some of the larger models. Most models can do a picture of a STOP, DANGER or WARNING sign, for example, and will even spell them right sometimes.
Even though i wasn't expecting much, this ruined my day. The thought that there is no way to stop the development of this stuff only makes me feel worse.
Radical avant garde is big beard like Brahms. Darwin and Marx fit this category too. If you look like you belong from 1880, then you are doing it right. Futurism is contemporary aesthetics with neon lights. Apply an 80s vaporwave colour filter for extra future. Wait a minute.
"Show me an image of a composer writing in a conservative tonal idiom. And show their workspace." Well, it looks like you got Ben Levin, but I'm not sure how conservative his music usually is. I can think of a number of reasons why the AI would be confused. A person cannot write inside of a "tonal idiom." A person can write inside a house, next to a mouse, in a box, next to a fox, etc. But you can't write music inside of an idiom. So the term "idiom" is probably ignored, along with "conservative" and "tonal." The AI will understand "workspace," but I'm sure that everybody in the 17th and 18th century would have asked for clarification. Radical notation: Etchasketch, Spirograph, Slinky toys, and yarrow sticks. Where will we be in 100 years with music? Very hard to say. Avant garde is supposed to be on (in tech terms) the bleeding edge of music. What will humans be doing that the computing machines cannot calculate? I've been writing software for over four decades, and while machines have been getting smaller and faster, I can't say the same thing about software. AI is bloatware, and takes a tremendous amount of energy (entire output of a coal-fired power plant) and hardware to function. So where will we be with music? Technological civilization could very well collapse under its own weight, and then we'll be tooting on bone whistles and banging sticks on rocks or logs.
Problem here is that you're asking the machine but you barely gave it feedback on what you were looking for so it could learn and correct it's imputs to tailor your needs and even to gain from your knowledge and experiences (which are not common).
"good artists borrow, great artists steal, the best artists ransack the www's public and not so public datasets." thank god these tools are indeed nothing more than expensive copying machines :)
"if i were to do this again in 3 months the results would be way better" This is an interesting thought to me. I imagine the results instead will be way worse.the models are beginning to consume their own output at a high rate. all the failed attempts to produce images of music notation for example will be included in training with no distinction that you were not satisfied with them. It's all a random amalgamation of aesthetics and contextless content that already exist.
At this stage, we don't know. It is possible that these technologies will continue to improve at an astonishing pace, and hugely disrupt every every aspect of our society. However, such techno-utopianism has been announced many times in many domains and failed to materialise. Is it all marketing hype? Or are we about to be subjugated to our transhuman successors? Time will tell.
I think your first iteration of the conservative, tonal composer was more accurate, in that they gave off more pop/film music vibes. The second version seems to have made the figure more conservative in the political sense (hence the 'I work in finance' look)
@@christopher9152 That is because Americans have been indoctrinated from birth to believe that successful capitalists must be hard working geniuses. Maybe Musk and Trump will finally teach them a lesson in reality. 😀
@@FugaxContrapunctus of all the things I've seen people get mad about on the Internet, "radical composers have big beards" is definitely one of the weirdest
If you understand Grok better than I do (or even if you are just as mystified as I am), let me know your thoughts about this below!
"Show me what a composer will look like in 100 years" Proceeds to display a skeleton.
I mean, where's the lie? Any composer today will be bones or dust in 100 years. 😂
I guess they...decomposed.
@@JimPea The Decompsong Compsoers. I feel a Monty Python moment about to happen.
I love the way “the future” is largely reified as gamer lighting 😂
it also ends up constituting quite a harsh criticism of generative AI - the most futuristic thing it can show you is a nonspecific, slightly worse pastiche of the present
Correct, these programs do not actually predict the future.
I don't know much about Grok specifically, but most AI machine learning is based on what already exists, especially in larger numbers as to what represents popular culture - therefore these systems seem to excel when you ask it to make a trap beat with pop vocals, not only because those are musically easy and formulaic, but because also it's been fed such material in excess to learn from that it's effectively mastered many "modern" styles of popular music. However, ask these systems to compose something in the style of classical (Mozart, Beethoven), it will get some of the features correct (it seems to understand tonal harmony no problem), but it can struggle with orchestration and will conflate things like "solo violin" with a full string orchestra or even a solo woodwind instrument. Ask it to compose in the style of later/romantic-era, and then 20th century experimental era like Berg, Penderecki, or Messiaen, it doesn't seem to really understand this all too well, which I can only assume has to do with the fact it has barely learned from their music to understand the possibile idioms behind their writing. What it will do, however, is create something pretty random - which on the surface, might seem what those later composers sound like, but this takes no account for the structures and concepts that fueled their writing.
So it's not surprising to me that it struggles on the one hand with notation which is sadly fallen out of favor for DAW-style piano rolls - it has likely not been learning much from this, just as many modern humans don't seem to learn it either lol.
And on the other hand, it doesn't surprise me that it can't peer into the future with a satisfying (and perhaps surprising) result - rather, showing us something which looks like a modern day "beatmaker"/producer using a DAW with those fairly typical neon pink lights everywhere (and oh god, just as you called out, all the cliché succulent plants!). Because this is what it has learned from as "modern", it seems to have difficultly imaginging beyond what it has been taught. The one around 17:15 made me laugh because if you look closely at all the screens, these actually look exactly like modern day video editing and color grading applications - and with all the visuals open, it seems a composer of the future is being conflated with a modern-day video editor lol.
Another funny thing I observed in all the modern and "future" images the keyboards are essentially minature portable synth-style keyboards. Is that a prediction that composers of the future will minimize the importance of a piano as an instrument and compositional tool (which in some sense, seems to be the case with many producers)? Rather than simply getting smaller, I think in 100 years the piano-style keyboard may diminish in favor of touch surfaces and VR style control over note input - but because I have no idea what this looks like, it seems AI can't imagine it either 🫠
Thanks, Jonathan. Among other things, you anticipated (or overlapped with) some of my concerns written up in a comment that I wrote after yours but didn't see first.
I got some scary good results on another AI platform with conditions like "serialism", "vocal", "chamber" and then typed in my own text. It spat out something like Berg. The orchestration was also quite interesting. But it's not composing. I did do another test with just solo violin which I then orchestrated around the audio that Udio created.
I view Grok as not a prognosticating or analyzing tool but rather a new age and amazing art program. Every one of the images which were created in this video were magnificent and I'd love to frame them all and put them on my wall. Great for album art. The earliest ones with their strange manuscripts were quite magical, stuff of dreams. That was fun.
"Make him look more conservative."
I was totally expecting GROK to put a red MAGA hat on him with an AR-15 in the background or something. LOL
Do you not have a sense of humour at all?
I get that these easily digestible ""jokes"" serve well their purpose as engagement devices in platforms such as these, however I do think there are better ways to go about displaying stereotypes to an audience other than validating them by putting them into practice live, an standard to which I held your content. Sorry if my demeanour comes off as upstart over minor things, but as a person with Asperger's this "sense of humour" admonishment comes off as a very poor excuse for secondary discrimination against composers with such unpopular stances nowadays as refusing the trends and believing the potential for tonal music is far from over, not merely as an inert foundation on which to build radically new genres and styles, nor as a dead weight to shake off in order to mandatorily pursue "uniqueness" in the craft, but as a tradition that still endures in its own right and deserves to be revitalized.
@FugaxContrapunctus If you don't have a sense of humor (and I'm not judging anyone on the spectrum here) then maybe don't feel compelled to post replies to comments/posts which are CLEARLY meant as a joke. If humor is not your thing (which is totally fine), then such humorous comments are likely not your thing either. Just ignore them and move on to other comments/posts with which you are better suited for interaction. It is not necessary to interject replies to all comments/posts that pop up.
That was my exact thought. 😂
@@FugaxContrapunctus I bet you're a lot of fun at parties.
I love it how we're all prone to being so courteous toward AI. "Please generate." LOL... I do it unconsciously too.
It just feels wrong not to say please
Thats part of how it works. In language if you want something you have to be nice, same thing with language models trained on human text.
There are a lot of elements to human intelligence, aren’t there? Those crazy images reminded me of something that I saw years ago. I taught junior high instrumental music for over thirty years. I once had a principal who came to school on a weekend and hand painted musical symbols and fragments of music on the wall next to my classroom door. While I appreciated his attempt to brighten up the drab hallway, my principal artist was not a schooled musician. Since our music studies included some theory instruction and assignments, I gave some thought to assigning students to identify ten errors in notation in Mr. P’s art. Alas, the idea was more amusing than practical and I imagined that insulting my boss wouldn’t have been a good idea.
It was a wonderful surprise to see you on Triggernometry today! Keep up the good work!
"I tried GROK. Here’s what happened." This text, on UA-cam's page, was accompanied by a generic monster. I clicked on the image and there was no monster. It is not what happened, it was just click bait! Stop already. Don't become a hustler. We have more than enough of them.
X (Twitter) is licensing the new Flux image generator to produce images. When you ask Grok 2 to produce an image, Grok 2 merely hands down (sometimes refining it a bit) your prompt to the image generator. What the image generator produces has little bearing in what it is Grok 2 - the language model - understands about your request.
I had some fun with ChatGPT (DALL-E), using your prompts. It was rubbish at producing sheet music, much worse than Grok. No matter what I did it came out smudged and pretty unreadable. However, it did a much better job at drawing avant garde and futuristic composers. It seems that in the future composers will mainly use some kind of light rays when creating music, and in 1000 years composers will themselves be luminescent.
That’s a future worth fighting for
Maybe need to ask what the Krell musicians work environment was/will be like.
A 20 mile wide room.
Something worth noting: since you stuck to image prompts, the results you got were not actually from Grok (which was very likely trained on tweets that speculated about the future of what music composition would look like), but from api calls to a separate image gen model Flux, which was only trained on images and whatever tags and descriptions accompanied them. Hopefully this illustrates why the knowledge base of Grok does not seem to be influencing the image outputs.
Thanks. What would you suggest I do to maximize the relevance of my prompts?
Brilliant concept Sam! I, for one, would be over the moon if you made a regular series/playlist out of this, and tested different models in these was, as they continue to likely improve!!
I’m not done with this topic!
I work for a tech company and Samuel’s video is by far the most interesting thing I’ve seen yet in regards to AI generated responses. Love it! Please revisit in a few months.
1:50 What's interesting about this result is that it seems to imply that the source materials which it is drawing upon include both photographs of older handwritten Western staff notation and at least a few East Asian sources, specifically either Japanese shakuhachi notation or Ryukyuan kunkunshi-the few semi-legible letters appear to be the hiragana お and ゐ, the latter of which is obsolete in standard Japanese but would appear in Classical Japanese texts and is sometimes used in Okinawan and other Ryukyuan languages and thus would show up in kunkunshi; there are also suggestions of certain kanji which I think are common to both. That said, the fact that it is presented horizontally rather than vertically is something you see in Western notation and some Classical Japanese writing, but traditional Japanese forms of notation are, to my understanding, entirely vertical and right-to-left. Very odd result.
Open the pod bay door Hal . . .
My first thought about the illegible symbols after the treble clef, is that Grok was attempting to imitate the bass clef's two dots, thinking that was some sort of rule.
Do a video where you ask Grok to compose a theme melody for a [fill in the blank] film.
I tried MidJourney a year ago, and I was getting results that reached roughly the same impasse that you have today. In short, our expectation is for imaginative results, but AI never has any comprehension of the symbolic meaning of the representations its working with. The brief, therefore, is to build an image based on the language used in the prompts, where the language indicates a link to the 'machine-learned' material, not the actual meaning itself.
Fun video. Grok is good a summarizing things in text, and generating fun images. Google has an AI that can write in the style of Bach. "Celebrating Johann Sebastian Bach Doodle." I wrote a piece with it and expanded / tweaked it, you can listen to it, it's called "Nine-Part Chorale for Human and Mouse" by BipTunia.
Grok? More like a Crok!
Considering what's out there to feed and train ai, and given the prompts, I think it did good! Thanks for sharing
Watching this is worse than a visit to the dentist.
To think we have "advanced' to this is beyond belief!
Do we want Grok to learn music notation and/or theory?? What could go wrong??
The composer with the avant garde notation has the head of a sixty year old and the hands of a twenty year old
Never worked a real job in his life
That clean shaved white haired composer closely resembled John McLaughlin. Maybe referring to specific examples (" in the style of") during the request could yield more desired or more interesting results.
if this is the state of AI in 2024 then I wonder if it will ever reach the escape velocity to become a true AI.
I doubt it
Grok isn't aware of the existence of female composers.
It doesn't want to trigger the anti-wokes lol
I noticed that too
@@SendyTheEndless Who is there that is really top-tier? Asking for a friend!
It got that part right. Its mistake is to think that there are any contemporary composers worth considering at all. Male or female.
You're barking up the wrong tree.
All it's doing is pointing a mirror at society. It's simply showing us the sort of cliché image of a composer that composers themselves and pop culture have helped create. If composers have something interesting to say then maybe AI will too.
have you tried udiio
Don’t ask for an image, that accesses a different model that lacks any learning of meaning. You might do better querying the text generation model that does map and learn meaning.
This is true for chatgpt too.
9.19 Grok: Show me John McLaughlin working out his guitar part on Shhh
"Huh, this is getting more and more abstract. And weird." is the current generation of AI's motto.
It’s now the conservative idiom to be “avant-garde,” and it’s now radical to be of the “tonal idiom”
If I was GROK I’d also be very tight-lipped about the future…
I'd love you to do this again, but try to play some of the music it produces!
This was fun.
Ok Dr. Wilson from Dr. House.
Grok has to be the 1984 and Future Shock version of composing and performing, yikes show him the door!
You need to try the full grok version, not mini. It came out yesterday, I think
I did. Doesn't seem any different.
You should try Midjourney, Samuel. It is waaay better than Grok for image generation.
If you try this again, I'd be curious to see what results you get when asking it to show you the notation of themes or motifs from famous works (Beethoven's fifth is probably the most obvious candidate). It would also be interesting to compare results from different text to image generators (Bing's AI image generator is also free, you just need an email from a Microsoft domain - outlook, live, hotmail, etc.)
I think this will run into the same problem as image generators have with generating text in images. Diffusion models have no semantic concept of what words, letters or symbols mean, so they get combined and distorted in very strange ways, leading to nonsensical, mirrored and incomprehensible text. The same thing is happening with the attempts at staff notation, which is really just another form of written language.
@@bahlalthewatcher4790 But there's a very good chance that there were some examples of the head-motif of Beethoven's 5th in the training data of some of the larger models. Most models can do a picture of a STOP, DANGER or WARNING sign, for example, and will even spell them right sometimes.
Interesting: for GROK "composer" = "man"
Every time
He is also looking at you annoyed. The future guy.
Even though i wasn't expecting much, this ruined my day. The thought that there is no way to stop the development of this stuff only makes me feel worse.
I think at the end Grok confused policing with "polishing"
Ask it to compose a whisper
@15:17 kinda leaning towards a Percy Grainger type figure
Here's not what happened: good results.
"Make him look more conservative"
Those are not AI generated, those are photographs from some image bank. WTF?
Radical avant garde is big beard like Brahms. Darwin and Marx fit this category too. If you look like you belong from 1880, then you are doing it right.
Futurism is contemporary aesthetics with neon lights. Apply an 80s vaporwave colour filter for extra future.
Wait a minute.
"Show me an image of a composer writing in a conservative tonal idiom. And show their workspace." Well, it looks like you got Ben Levin, but I'm not sure how conservative his music usually is. I can think of a number of reasons why the AI would be confused. A person cannot write inside of a "tonal idiom." A person can write inside a house, next to a mouse, in a box, next to a fox, etc. But you can't write music inside of an idiom. So the term "idiom" is probably ignored, along with "conservative" and "tonal." The AI will understand "workspace," but I'm sure that everybody in the 17th and 18th century would have asked for clarification.
Radical notation: Etchasketch, Spirograph, Slinky toys, and yarrow sticks.
Where will we be in 100 years with music? Very hard to say. Avant garde is supposed to be on (in tech terms) the bleeding edge of music. What will humans be doing that the computing machines cannot calculate? I've been writing software for over four decades, and while machines have been getting smaller and faster, I can't say the same thing about software. AI is bloatware, and takes a tremendous amount of energy (entire output of a coal-fired power plant) and hardware to function. So where will we be with music? Technological civilization could very well collapse under its own weight, and then we'll be tooting on bone whistles and banging sticks on rocks or logs.
No xenomorph, 2/10 clickbait see me after class
Gone all radical beardy Harry Partch on you
Problem here is that you're asking the machine but you barely gave it feedback on what you were looking for so it could learn and correct it's imputs to tailor your needs and even to gain from your knowledge and experiences (which are not common).
"good artists borrow, great artists steal, the best artists ransack the www's public and not so public datasets." thank god these tools are indeed nothing more than expensive copying machines :)
"if i were to do this again in 3 months the results would be way better"
This is an interesting thought to me. I imagine the results instead will be way worse.the models are beginning to consume their own output at a high rate. all the failed attempts to produce images of music notation for example will be included in training with no distinction that you were not satisfied with them. It's all a random amalgamation of aesthetics and contextless content that already exist.
At this stage, we don't know. It is possible that these technologies will continue to improve at an astonishing pace, and hugely disrupt every every aspect of our society. However, such techno-utopianism has been announced many times in many domains and failed to materialise. Is it all marketing hype? Or are we about to be subjugated to our transhuman successors? Time will tell.
Samuel, did you actually mispronounce Moog? I’m surprised.
Yes. Slip of the tongue. It happens!
Ah yes, the furistic composer Robert Downey Jr
What ! ? The first image when translated tells how to buld a free energy device and proves the existence of God
The only problem is to find out how to translate it😂
The composer of the future is a young muslim. Quite accurate.
el compositor es jhon Mc Laughlin
I think your first iteration of the conservative, tonal composer was more accurate, in that they gave off more pop/film music vibes. The second version seems to have made the figure more conservative in the political sense (hence the 'I work in finance' look)
Perhaps there is a message here to grow out your beard!
Apparently it understand music about as well as Elon Musk understands most things. 😀
Well, he certainly understands how to scam, coerce, or otherwise steer the American federal government into funding his often half-baked ventures.
@@christopher9152 That is because Americans have been indoctrinated from birth to believe that successful capitalists must be hard working geniuses. Maybe Musk and Trump will finally teach them a lesson in reality. 😀
Why would I bother to care about AI music crap? Nice vid though 😊
More radical composers have bigger beards lol
I might not have understood this had I not used Grok
There does appear to be a correlation
@@FugaxContrapunctus of all the things I've seen people get mad about on the Internet, "radical composers have big beards" is definitely one of the weirdest
@FugaxContrapunctus it’s a joke, relax!
@@whatsthatnoise5955I think he’s joking, and rather hilariously lol
ask it to show the listening audience for the avant garde composer