site banner
PaperclipPerfector  1mo ago (text post)   3295 thread views

Friday Fun Thread for September 26, 2025

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

3
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

Suno, the AI song generator, just released version 5. And now I think we are 100% past the uncanny audio valley. Version 4/4.5 was at the level of "Convincing, occasionally incredible, but still flawed enough to notice." Version 5 is 'tricking' my ear 9 times out of 10. Studio quality. We have a fully functional infinite music machine available for the monthly cost of a cheeseburger.

A few examples:

https://suno.com/s/N86w28eQjBWbI6fA

https://suno.com/s/BsKe5OnQpUhPj2Zx

https://suno.com/s/voPPxtsXxRjFRF93

https://suno.com/s/Yqe3pzUQHIPAQ4g4

I think people get too focused on the apparent 'slowing' of progress in the LLM space and think its proof that Machine Learning itself is not living up to the hype.

Meanwhile stuff like Video generation, Music, and Protein Folding/Drug Discovery are still improving rapidly.

Arguably LLMs are just the interface by which we can access these other powerful Djinns to provide us with the particular services we want, as we await the "one true superintelligence" that can do anything to arise.

There's probably a small window right now to write a Sci-Fi novel that features humanity invoking individual AI patrons that specialize in particular aspects of the world, in the same vein as 'old gods' (Stockfish God of Chess, Suno God of Music, Midjourney God of Aesthetics).

Anyway, if there was ONE arena you would want AI to reach superhuman capability, one particular application that would improve your life even if AI progress stalled out otherwise, what would that be?

For our purposes, lets just grant "customized pornography" as the killer app.

Me, I think I want the ability to produce bespoke episodes of older TV shows that I enjoyed but were cancelled or went off the rails and/or had horrible conclusions. GoT and Firefly are obvious examples there. But I have several others in mind.

It'd be cool to live in a world where the "Canon" of a given series was not defined by any particular "official" source, but instead you had a whole library of 'forks' in the plot and character development that fans can choose from, or generate their own as they like, with maybe some curation done by the rights holder to identify the entries they deem 'high quality' and consistent with the original vision.

Loved "Moonshot Rider". "Fight" isn't bad, either. You are correct; Suno 5 smashes the Turing Test.

But I was a fan of AI music even before this release. AI music didn't have to be as good as human music to be competitive; it just needed to be good enough to allow someone with creativity to create things that could have never existed otherwise. For example, demonflyingfox's parodies of Family Guy and Game of Thrones, or corridos like "Luigi Magione" and "Nariz Grande".

Just two months ago I played through a visual novel called Stains of Blue. It's still in development, but when you reach the end of the prologue, you get hit with the theme song and a montage of images from the game's various cutscenes. And even though the music is clearly AI generated, the way it was presented and tied up perfectly with the game's themes and story was so beautiful that I almost cried.

From Quarantine by Greg Egan:

I flop onto my bed, and switch on the room’s audio system. The controlling ROM I’ve been playing lately, “Paradise” by Angela Renfield, is one of hundreds of thousands of identical copies, but each piece it creates is guaranteed unique. Renfield has set certain parameters for the music, but others are provided by pseudorandom functions, seeded with the date, the time, and the audio system’s serial number.

Tonight, I seem to have chanced upon an excessive weighting for minimalist influence. After several minutes of nothing but the same (admittedly, impressively resonant) chord, repeated at five-second intervals, I hit the RECOMPOSE button. The music stops, there’s a brief pause, then a new variation begins, a distinct improvement.

I’ve run “Paradise” about a hundred times. At first, I could hardly believe that the separate performances had anything in common, but over the months I’ve begun to apprehend the underlying structure. I see it as resembling a family tree, or a phylogenetic classification of species. The metaphor is imprecise, though; one piece can be judged to be a near or a distant cousin of another, but the concept of ancestry doesn’t really translate. I think of the simplest pieces as being primordial, as “giving rise to” more complex variations, but beyond a certain point it’s an arbitrary decision as to who begat, or evolved into, whom.

I’ve heard some reviewers assert that, after a dozen playings, anyone who is musically literate should fully understand the rules that Renfield has chosen, making further actual performances unbearably redundant. If that’s the case, I’m glad of my ignorance. Tonight’s second piece is like a brilliant scalpel blade, prising away layer after layer of dead skin. I close my eyes as a trumpet line builds, rising in pitch, then mutates, impossibly, effortlessly, into the liquid sound of metaharps. Flutes join in, with an ornate, mannered theme – but already I think I can discern in it, hidden beneath the fussiness and decoration, hints of a perfect silver needle which will recur in a hundred guises; which will be honed, muted, then honed again; which will be held up for my admiration, one last time, then plunged into my heart.

Suddenly, four lines of glowing text appear at the bottom of my visual field:

[Backroom Worker:

Natural memory association.

Casey, Joseph Patrick.

Head of Security as of 12th June, 2066.]

I’d forgotten that I’d asked for staff records, too – or I would have excluded them. I think about waiting for the music to finish, but there’s no point; I know full well that I’d be unable to enjoy it. I hit the STOP button, and one more unique incarnation of “Paradise” disappears forever.

Yeah.

To add to the shock, OpenAI just put out Sora 2, which is also gobsmacking me with how good it is.

https://sora.chatgpt.com/p/s_68dc49d67ce0819194ea5d9f24bdb28e

This video is completely 'convincing' to me, between the reflections, the dog, and the traffic in the background and road noises. I can still reason out that it's AI, but my natural intuition is not picking it up automatically anymore.

It is also pretty damn good at quality-looking animation and voice-acting. albeit in very short bursts.

Basically, as these tools improve, the amount of actual creative skill and free time needed to create 'passably decent' media drops by like 50% every 6 months.

Someone's going to figure out how to hook Suno, Veo and/or Sora, together with a 'director' LLM and make full on music videos or contiguous scenes with soundtracks and everything.

https://suno.com/song/106b0e6b-6345-42ba-91b6-0e2e0d6c5d82

This is a jam - they would play second stage at Ozzfest cut a 1999-2003 and have a following.

Color me impressed.

My favorite ways to use the tech thus far:

  1. Producing truly weird Genre mashups. Japanese Folk Music + Bagpipes? Mariachi Sea Shanties? Oops, almost forgot: Heavy Metal Ska.

  2. Producing novel content in the vein of certain genres or bands of the past that I miss/wish had more content.

  3. Converting songs between genres with as few changes to the melody as possible to hear how the emotional tenor of the song changes even if the lyrics and melody (mostly) don't.

Strange sense of nostalgia to hear something that very easily could have come out during your childhood, even though you know with 100% certainty that you never heard it before this moment.

Your sea shanty sounds Irish more than anything, at least the vocals.

I specifically requested an Irish accent for maximum dissonance.

I haven't played too much with asking it to do different languages with different accents, but it does a pretty good job of adding Indian-accented English to songs.

Oh, neat.

Also, strong argument that almost all the well-known Sea Shanties are directly derived from Irish musical heritage.

Does this melody sound familiar?

(About 6-7 years ago I went on a kick researching European Maritime culture and learned a lot of interesting stuff).

My benchmark is an actually good language learning tutor LLM. The task seems pretty much perfect for the current crop of LLMs, and there is infinite VC funding for literally anything that involves LLMs. As long as something so obvious isn't mastered yet, LLMs are hardly more than a gimmick imo.

Depends on your definition of tutor? My go-to is:

[JAPANESE SENTENCE]

Can you pull this apart for me?

I'm at an intermediate level already but this can break pretty much everything down and explain the grammar points that I'm having trouble with.

There is a large abundance of beginner resources for any language out there, and yes chatgpt is fine for this sort of grammar explanation tasks. But this is not a human tutor that can converse with you or give you reading tasks with a specific goal in mind, noticing and focusing on your weaknesses etc. You have to be your own tutor and can get some help from chatgpt

Sure, I'm just saying it's a real game-changer for me. I've been stuck at a reasonably advanced plateau for a long time now because a lot of literary Japanese structures are hard to look up, and having something that can reliably figure out what structure is being used and what the referents are is incredibly helpful.

Personal Assistant.

"AI, please do my taxes, handle the car damage insurance case in a maximally profitable way, and tell me which area of work to focus on today. Also, which one of my friends have I neglected to catch up with the longest? Remind me to call them this evening, but not too late."

v5 is much better but it still sounds Suno... It's indescribable, something about the clashing sound (if I was a musician I could be more clear).

Is what I was going to say, till I found one that didn't sound Suno at all. Wow.

Anyway, if there was ONE arena you would want AI to reach superhuman capability, one particular application that would improve your life even if AI progress stalled out otherwise, what would that be?

Writing long-form fiction to even a high-human standard. Turn all my neverfinished story ideas into full stories, better than I ever could.

Writing long-form fiction to even a high-human standard.

I like this one. The ability to create a fictional world that has 'infinite' branching plotlines and arcs to follow, or to produce the exact sort of one-shot novel you've been craving.

I actually found it comparable to a lot of the early GPT3 written content. The models were able to produce grammatically coherent text that mostly stayed on topic, it just didn't go anywhere and after reading a few paragraphs you'd get pissed having been duped into reading essentially verbal diarrhea.

This music is much the same, there is no purpose and it doesn't go anywhere. Its like 10 second segments stitched together without much thought except having smooth transitions and staying in the sameish genre.

Perhaps there are decent songs produced and it's down to prompting and iteration, but I haven't heard them yet.

This music is much the same, there is no purpose and it doesn't go anywhere.

What do you mean? Why does music need a purpose? It just needs to sound good. A purpose and meaning are optional.

Now the lyrics, when they're AI, can have a samey, obvious vibe. But that's just the lyrics.

Principles such as tension and release, foreshadowing, thematic development, chord resolution are the essential tools of virtually every composer, regardless of genre and whether the composer is actively aware of it or not. While often discussed in a classical context, their application is universal and apply to pretty much all genres.

This seems like a consistent weakness in at least current AI-generated music, which often does well with short-term musical grammar but fails to build a cohesive, long-form structure, resulting in a piece that feels aimless and unsatisfying, IE "it doesn't go anywhere".

These same concepts of building expectation and providing resolution are fundamental to other time-based art forms, such as literature and film, and to effective communication in general.

I think this feels similar to how earlier text generating models wrote and similar to how they still struggle to write longer form text. Perhaps the issue is an insufficiently large context window or perhaps it's something else more fundamental to how the models function, I don't know.

Its like 10 second segments stitched together

I'm pretty sure it's more or less exactly that: short segments stiched together.

The idea of relying on the feedback loop of remixed AI slop for entertainment and it drowning out genuine good stuff evokes in me disgust that is hard to convey.

This is the part where the whispering earring tells me "better for you that you take me off".

Why do you hate fun?

"Oh no, hundreds of millions of humans will be able to express themselves artistically in ways that were never possible before, resulting in an explosion of creativity and human expression"

You don't have to scroll the Facebook newsfeed or Tiktok brainrot. Both of which were absolute slop even before ChatGPT 3.5

It's not like HBO is going anywhere. In fact, I would assume AI lowering the cost of VFX, etc will result in more prestige TV, not less, as lower costs allow for increased scale and more risk taking. Plus lower barriers to playing with art may mean more people discover/develop latent artistic talent and get into the industry.

Low quality fan fiction has been available in infinite quantities for decades, and yet there are still high quality books being published. Low quality digital art exists, and there are still excellent artists. Quality art will always rise to the top (AI will probably help you discover it).

All I can think about when people get cranky about AI art is what portrait painters in the 1800s must have sounded like sneeding over how cameras aren't real art because it requires no skill as the camera does it for you.

I am not sneeding about no skill, I am sneeding about no expression - or at the very least, significantly less expression. How many prompters are going to painstakingly describe every detail they just don't have the painting skill to put on canvas, and how many are going to go "hot woman in cyberpunk armor, in the style of studio ghibli" or whatever? At least the photographer actually has to discover/set up the shot.

How many prompters are going to painstakingly describe every detail they just don't have the painting skill to put on canvas, and how many are going to go "hot woman in cyberpunk armor, in the style of studio ghibli" or whatever?

Very few, but who cares? The person who has the hot cyberpunk studio Ghibli waifu picture is now happier than they were before. Maybe 1 in a million will be so inspired (or so good) by this that they'll get super into prompting and become a legitimately good "chimera" artist who blends AI gen and human taste to make actually great art. That's a (tiny) win for the human race as a whole.

I have no interest in seeing their stupid waifu, but I also don't go on DeviantArt or whatever so this won't effect me at all. Maybe I'll see their great art 15 years later in an art exhibition on AI digital art.

I agree this will lead to massive explosion in slop, but human-generated slop was already at functionally infinite levels prior to AI, so I'm not sure if there's a net loss here.

human-generated slop was already at functionally infinite levels prior to AI, so I'm not sure if there's a net loss here.

I expect the ratio of slop:epic to rise by at least an order of magnitude and I do consider that a loss.

That's fair

The idea of relying on the feedback loop of remixed AI slop for entertainment and it drowning out genuine good stuff evokes in me disgust that is hard to convey.

I can't find myself caring one bit about it because the good stuff slowed to a teeny tiny dribble over two decades ago. As far as I'm concerned there is nothing left for AI to drown.

Tell me you haven't listened to The Alex Jones Prison Planet without telling me etc...

Meh. That just sounds like a collection of shitty djent cliches.

I can't find myself caring one bit about it because the good stuff slowed to a teeny tiny dribble over two decades ago. As far as I'm concerned there is nothing left for AI to drown.

My perspective is that this isn’t strictly true. It just seems that way because algorithmic media (Facebook to Netflix), instead of expanding the window of exposure, narrowed it and homogenised everything within it. Thirty years ago in my small town in Europe, I had more choice in my local video rental store (where there was a whole section of foreign language films, that was updated frequently) than I get on all my home screens and feeds now, no matter how far I drill down.

Edit: I believe the “good stuff” is still being made, but its audience and distribution network is not part of the great algorithm fork. It’s elsewhere, it’s curated, and it’s often offline. Interested in the new avant garde. I suspect it’s here, it’s just not where 50-98% of people spend their time.

Edit: I believe the “good stuff” is still being made, but its audience and distribution network is not part of the great algorithm fork. It’s elsewhere, it’s curated, and it’s often offline. Interested in the new avant garde. I suspect it’s here, it’s just not where 50-98% of people spend their time.

Yep. You have to be willing to dive beneath the surface, long enough to find the pockets of original and specifically high quality work that the indie scene is putting out.

You have to be willing to dive beneath the surface, long enough to find the pockets of original and specifically high quality work that the indie scene is putting out.

So, how many hundreds or thousands of hours would you say is it acceptable to use to find more than a tiny handful of such gems? Please give a serious answer with actual numbers.

I've spent a lot of time looking for good music. When I do find some I haven't run into before, it's almost inevitably tracks made 30+ years ago, some new tracks from legacy artists (who may be rich enough to keep doing much the same thing they did 40 years ago, nevermind commercial viability) or some very occasional niche stuff. The last time I found an entire new album worth of good material was when Loreena McKennitt released Lost Souls in 2018 (and she was in her 60s by then, so not exactly a "modern" artist). Finding new indie releases on the level of say Depeche Mode's Violator, Aphex Twin's Selected Ambient Works 1 or Roxy Music's Avalon just isn't going to happen.

So, how many hundreds or thousands of hours would you say is it acceptable to use to find more than a tiny handful of such gems? Please give a serious answer with actual numbers.

You'd want it to be somewhere around a 1:1 ratio of [Time spent searching:Time spent enjoying] if you ask me, although the search can be rewarding in its own way, since you stumble upon curiosities and learn new things in the process, often.

And the most efficient way to find stuff is to to connect with people who have already done the searching and have dredged up gems, and are happy to share those findings. There's communities out there that like the things you like, and have more free time than you do, and thus there's gains from cooperation to be had, rather than trying to search everything up solo.

So think of it less in terms of the time spent finding the music you like, and more in terms of finding communities that spent time finding music you like, and can save you a lot of time and effort via combining efforts.

If I spend 10 hours to find a single album that I will then add to my collection and listen to sporadically going forward, I do think I'd consider that time well spent. Especially if I spread that 10 hours out over weeks or months.

I don't have an account that can generate stuff but if you do can you make Suno generate some Romantic period Chopin-esque classical music.

All I could find with a cursory search was: https://suno.com/song/8a864816-4b39-4913-acd4-debb9beab873 which is Baroque but I must say pretty damn good.

EDIT: We now have the modern day "Vatican Rag": https://suno.com/song/ebe85375-dc99-4368-8bc1-f9fe24ad21f3

Will do that later today.

Thx

Okay, went ahead and put a selection of them in a playlist.

https://suno.com/playlist/d8348313-aa62-4474-b272-75d4dbabf641

Note these took me <20 minutes to generate, counting time setting up prompts (I used ChatGPT to fill out the prompt descriptions a bit).

And less than $2 worth of credits, I'd guess.

I don't have an ear for music but these pass as the sort of classical songs I listened to on CD growing up.

The playlist has an "inspire" button that would let one use these songs as the basis for future generations.

Lest I pretend it does it perfectly every time, it did ALSO produce THIS monstrosity on the same prompts:

https://suno.com/s/5Gc5DUCoji1hG2kN

Clip the last generation to about 0:00-1:04 and it's a brilliant mindfuck. I love watching it just spiral out of control. Couldn't stop giggling.

AIs of all kinds really do seem to have a knack for creating nightmare-fuel content that is just barely, barely comprehensible but deeply unnerving for hard-to-articulate reasons. Like just under the surface, there's a psyche made of pure chaos. Really gets at the "Shoggoth with a smiley face mask" nature.

It's decent, not amazing and none of those would pass muster for me as a Chopin fan unfortunately, not least the second to last one which somehow has English vocals (Chopin did actually write a small amount of songs with words but they were all in Polish and not published during his lifetime)...

Still I guess if you showed these to somebody back in 2015 they'd still be rightfully amazed. Still significant work to be done, at least in replicating Chopin properly.

I'm starting to get Gell-Mann Amnesia with Suno, it does really badly according to my ears for the genres where I sort of know what "good music" is supposed to sound like and it does really well according to my ears for the genres where I don't know much about them beyond vaguely knowing what they should sound like.

Also WTF is that last generation?

That last generation was the sort of thing that it used to produce more often circa V3.

Some weird distortion that spins off into eldritch horror territory as it compounds.

Which emphasizes my point that it has advanced a lot over just the last 6 months.

I really would like to try a double blind experiment on this forum to see who can still spot the AI songs.

I want it to be a good and useful personal assistant; that means not being obtrusive or people-pleasing or fishing for further engagement or trying to be my friend, but having just enough volition to occasionally nag me about stuff I should be doing.

If ask.com still owns the Ask Jeeves trademark, they can sell it for big bucks right now to someone trying to build a good digital personal assistant.

I'm not sure how you can make one that actually prods you to do things unless you give it the ability to physically interact with you somehow.

My phone has a lot of notification and reminder systems, but I ignore those all the time.

My experience also. Easy to drown in notification ocean.

Actually that right there is a use for AI.

Read the notifications pouring in, mute the unimportant, queue up the actionable but not urgent, and only let you see ones that read as actually important (not just "flagged" as important).

Or just do stuff on its own. If I could have Gemini monitor websites XYZ for sales on some consumable and order ten when there's a sale, that would be pretty convenient.

I suspect that's just result in there being no more sales on such items, demand would probably smooth out substantially.

Flip side, AI monitoring how much of a given consumable I have left and then ordering it for me (keeping me in the loop, I'd hope) before it runs out entirely, and doing at least some due diligence to ensure its a decent price.

I suspect that's just result in there being no more sales on such items, demand would probably smooth out substantially.

There's probably some variance in people's willingness to buy in bulk. Costco exists but that hasn't resulted in everyone matching Costco's prices. Maybe more websites will start offering bulk discounts (kind of like subscribe and save).

You don't notice the very clear artifacts in the sound?

I don't really understand why these exist and would have assumed that would have been easy to fix but seeing as they haven't been perhaps it's much harder than I imagine.

Beyond that the music itself is bland and the audiomixing poor, but perhaps these are prompting issues with these particular tracks.

If the technical issues can be ironed out this seems fine for stuff like background music in shops and receptions. Given that a Spotify subscription costs as much why would anyone switch though?

What is the use case here? I'm not being facetious, I think this is cool as hell but what is the use case? Composing and recording music is already dirt cheap. Extremely quick prototyping? Perhaps as a way to create temp music?

Composing and recording music is already dirt cheap.

There's at least, like, one semester of instruction required to learn a DAW. I can teach myself C++ and how to sys admin and use a DJ mixer. But a DAW just feels too hard without hand holding, same as Photoshop for me.

Being able to prompt Suno and get something fun to listen to (albeit flawed) feels liberating.

Is this going to be akin to that study about how good programmers working on mature projects did not save time using AI tools even though they thought they did?

I meant in the sense of hiring someone to do it for you. There is such an insane overproduction of these kinds of competences and it's been going on for like 50 years.

Oh, I see.

I would guess Suno's use case is the long tail of people who want bangin' music for their YouTube channel/podcast/home videos but have no idea how to license music or find someone to compose it for them.

As someone that occasionally DJs I have wanted music that hits a certain vibe but finding it on Beatport can take hours upon hours of listening. "Deep house, breathy female humming, no lyrics, sexy" is a really hard search but Suno will make you a couple in less time than it takes to listen. I may or may not use them because of quality issues but it does seem to be improving pretty quickly.

What is the use case here? I'm not being facetious, I think this is cool as hell but what is the use case? Composing and recording music is already dirt cheap.

You get 3000 credits a month for $10, and each song gen costs 10 credits. So you can get a couple hundred songs (assuming you're putting in a little bit of effort to tweak each one) each month for less than the cost of a CD back in the day. Whether that's a sustainable price point I don't know.

So we're talking below dirt cheap at this point.

MY use case has been making new high-energy songs to slip into the playlist at the gym, which is a fun process.

But this is also a step towards what I suggest is my preferred use-case: to make bespoke TV episodes, it'd need some ability to compose soundtracks and theme songs and such.

You don't notice the very clear artifacts in the sound?

I've been using Suno for about a year and a half, and its gone from "Constant artifacts that instantly betray its AI" to "If it played on the radio I wouldn't peg it unless I was paying close attention, and even then I wouldn't be sure." It even adds in respiratory sounds for Pete's sake!

Or more to the point, I think that if we did a double-blind test with randomly chosen people listening to AI songs vs. decently skilled indie artists, 80+% of them wouldn't reliably catch which were AI and which weren't, if we curated the AI stuff just a bit.

Given that a Spotify subscription costs as much why would anyone switch though?

Funny you should mention that.

Owing to the absolute dirt-cheapness mentioned above, its a 'viable' (if you cheat) business model to mass produce barely passable songs and upload them en masse to every streaming service under the sun.

And its actually debatable if this really makes the services worse given the fact that most users don't seem to notice or care much.

It certainly makes it harder for new, undiscovered artists to stand out. And that's the one thing AI has going against it. You can't yet attend a concert for an artist who only exists digitally.

But mark my words now, the first large music festival showcasing ONLY AI-produced music will be happening inside of 5 years.

Also, like two years back I talked about how I was still collecting music to my local devices through force of habit. It seems even more laughably futile now in the face of tech which can keep producing songs faster than I can even listen to them.

And more recently we discussed the art of cover songs.

This is also a machine that, if legal restrictions were not an object, lets you translate any given song to any given genre, instantly.

mark my words now, the first large music festival showcasing ONLY AI-produced music will be happening inside of 5 years.

Maybe, and maybe there will be sufficiently large cohort of people who want to go that it becomes a viable vehicle for the entertainment industry.

Much more likely, I think, is the move away from AI / digital art because people realise they need something human. That something human will be delivered not on screen or via any digital mechanism, thereby creating high demand for real life events (everything from spoken word poetry to pop-up tiny stage theatre to large concerts). The price inflation of tickets for gigs and live sports events is a symptom that shows this need for a real experience is already happening. It’s growing and not yet been adequately catered for.

Also, like two years back I talked about how I was still collecting music to my local devices through force of habit. It seems even more laughably futile now in the face of tech which can keep producing songs faster than I can even listen to them.

Why would it be futile if you actually cared about those songs? Unless you mean you collect music in the same way warez people collect releases, ie. without any real care about the actual content.

Unless AI music generators achieve near true sentience level understanding of music and prompts and can use that understanding to analyze a database of my preferences, I just can't see AI music in any way competing with the music I have collected (and slowly keep adding). If anything, the problem with collecting more music is that it's so hard to find something I'd like that I didn't already know of (and Spotify's piss poor recommendation system certainly doesn't help there *).

*: Would it really be that difficult for them to add options to "never suggest this artist / album / any variant of this song for this playlist / ever"?

And that's the one thing AI has going against it. You can't yet attend a concert for an artist who only exists digitally.

Ahem...cultured people have been doing this for a while now.

Yeah, its a very solid point.

THAT SAID... I'm wondering if people would actually be willing to buy tickets to sit and have the AI's songs just played at them over the speakers, even if there was an video accompaniment.

Or they can commission an actual band to play the songs, but at that point... just become a fan of said band?

Well, guess we'll see if I'm right:

But mark my words now, the first large music festival showcasing ONLY AI-produced music will be happening inside of 5 years.

Judging from YouTube, I think it's primarily useful for creating infinite quantities of music in genres that contain a lot of fuzz and vagueness already.

I suppose background music for YouTube channels that are very small or even solo operations could be a good use case. These people want to pay nothing (and licenses for use in videos is not cheap), might not want to use the same audio library as everyone else and the music itself isn't the focus.

And perhaps more importantly, using Suno or other service might be quite a bit faster than trying to find a suitable backing track from collections of free tracks.

Maybe, but people seem to reuse the same tracks from every video so the actual time to find a track might not matter much if at all. I guess the question becomes, what do you care more about? The $10 for a month of suno + whatever time it takes to generate a song you're satisfied with, or spending ~1 hour in the free audio library?

Suno, the AI song generator, just released version 5. And now I think we are 100% past the uncanny audio valley.

At least to the extent that the examples trigger my ctrl-w reflex in exactly the same way and for the same reasons as modern human created music does. I suppose that could be considered "progress".

I feel this. My immediate thought on listening to one of those Suno tracks: “We’re going to get a thousand Sabrina Carpenters now”

It can emulate just about any time period, and REALLY traditional stuff too.

I'm not even trying to argue that its not AI slop at the end of the day, just like I think most pop music is human-made slop, but its a leap in capabilities.

Can it do The 80s?

Not just completely forgettable fourth rate slop or faux-80s but actually good stuff that resembles songs like this, this or this, without forgetting this absolute classic.

My working theory: All those are great songs, but what makes them great is not JUST the notes and lyrics and performance. These things didn’t exist in a vacuum. What made them great was how people experienced them together in time and/or space, and responded to them together in time and/or space. This is why I’m hopeful that human art still has a bright future, but that bright future must take place offline more and more. (I appreciate this sounds ludicrous to anyone who’s lived their entire life with and on the internet.)

The Song of Roland example is not very good and sticks out instantly for anyone who actually understands that era of music. It sounds like the modern 21st century musician's idea of Medieval music and has very little to do with actual Medieval music.

I HAVE TO IMAGINE that there aren't many recordings of actual Medieval music around to train the AI on.

Well, that's a problem, because there are some, but they are utterly drowned out by so-called "bardcore" or "neo-medieval" music that has very little to do with what actual medieval music sounded like other than that is has a thin veneer of what modern audiences think it sounded like. The funniest part is that, in addition to medieval, it's tagged as modal and Gregorian Chant, when it's neither of those things. We have a pretty good idea of what actual medieval music sounded like by virtue of it having been written down, and we know what the theory behind it was and what the performance practices were. Almost everything in that example is anachronistic. Actual medieval folk music would have been monophonic in texture (every voice and instrument is playing the same melody line; modern concepts of accompaniment didn't exist yet) and modal in harmony (tonality i.e. chords had yet to be invented). The prodromes of modern harmony were present beginning with organum around 1200 (where the vocal lines would occasionally sing different complementary notes), but that would have been Latin church music, not any king and queen larping. You'd get polyphony in the 14th century but again, it would take folk music a long time to catch up to what composers were writing for the church. It wasn't until the 16th century that what we would call modern harmony and performance practice was fully developed and widespread in Europe. Before then, music would have sounded more like this, especially in England, which was far away from the locus of culture at the time.

The fact that authentic music is all but drowned out by bastardized modern versions is only further proof of the limitations of AI training—garbage in; garbage out. What you posted has more in common with a Taylor Swift song than with actual medieval music.

I can't really put my finger on what's wrong with it but I wouldn't want to listen to any of this again despite liking some of the genres it's aping.

I guess I agree that it's succeeded in producing replacement-level slop and to the extent that people thought this was impossible they should probably eat crow. But I see no indications of SOVL.

Yea but in three years it will?

two weeks three years to kiev sovl

but I wouldn't want to listen to any of this again despite liking some of the genres it's aping.

The question of the hour: Is that really different than most songs produced by human artists?

I admit that I keep falling back to the same ~1000 songs that I enjoy listening to, very few of which are less than 5 years old, many of which are older than I am. And most 'new' songs I'll play like a dozen times and then they sit in an unused playlist for months or years.

I don't think I've heard a SINGLE pop song in the last year that I consider 'memorable' (not entirely true: Chappel Roan's "HOT TO GO!" sometimes pops into my brain unbidden).

I truly do enjoy Kendrick Lamar's music, but after listening to GNX on repeat for a couple weeks I've not felt any desire to add it to my main playlist. Humble is on there though.

And I lamented before that there's really no such thing as a new 'genre' anymore. So the AI does have the advantage of letting me play around with combining genres to see if anything neat falls out or is worth pursuing.

I am going to agree there's no actual replacement for having a talented live performer in front of you.

I don't think I've heard a SINGLE pop song in the last year that I consider 'memorable' (not entirely true: Chappel Roan's "HOT TO GO!" sometimes pops into my brain unbidden).

Similar vibe (sassy red-haired female with socio-culturally relevant songs), you should look up CMAT.

The question of the hour: Is that really different than most songs produced by human artists?

Does it even matter? The world is already full of slop and having one more way to produce slop isn't helpful for anyone except some SEO spammers.

The question of the hour: Is that really different than most songs produced by human artists?

Possible, hence why I called it replacement level slop.

I do find Spotify's algorithm shows me recent songs that I like with some regularity (and no, these aren't from the Spotify ghost artists/ais), so I'm not as negative as you about today's music.

Music is perhaps not the point of pop music, it's the admiration and parasocial relationship with the artist. It's not clear to me if people will be willing to do this with an AI, but perhaps.

Music is perhaps not the point of pop music

Certainly not today, but it used to be, at least for the better tier pop. Just take the Beatles. Are they pop? Inarguably. Were they musically good? Without even the tiniest shadow of a doubt.

There are gobs of excellent pop music all the way up to the 90s. Then it went to shit for reasons I haven't been able to fully articulate yet but involves the concentration of labels, rise of solo artist & built groups and of course modern production methods (and a bunch of other things).

Edit: The music being a major point of pop music goes back centuries: For example Mozart and Beethoven were "pop" artist in their time.

Then it went to shit for reasons I haven't been able to fully articulate yet but involves the concentration of labels, rise of solo artist & built groups and of course modern production methods (and a bunch of other things).

To add: it went to shit also because pop music industry’s big data data-sciencing and machine-learning was able to identify what the masses wanted, and gave it to them. Manufacturing of, e.g. Sabrina Carpenter and Sabrina Carpenter types and the disappearance of auteurs and any sense of authorship to the margins.

The music being a major point of pop music goes back centuries: For example Mozart and Beethoven were "pop" artist in their time.

The top answer suggests that a much larger fraction of the population has heard e.g. Michael Jackson than Beethoven or Mozart in their time. Beethoven and Mozart were "pop" for the upper crust of society. Has music popular with the 1% gotten worse? I don't know, but certainly mass media has transformed who you have to appeal to in order to have mass success.

Beethoven and Mozart were "pop" for the upper crust of society.

Sure, but that was an economic divide, not a divide based on artistic qualities. It doesn't change the fact that 1) they were widely popular and 2) they have inarguable artistic merit. If anything, they were more pop music than what general folk listened to as folk music was anonymous and lacked any of modern pop music's parasocial relationship that Mozart and Beethoven had among parts of their audience.

More comments

it's the admiration and parasocial relationship with the artist.

I do think there's an issue where with AI you can't really make the AI itself the 'object' of your obsessions because there's nothing 'there.' And making the person prompting the AI the object seems extremely odd.

But I feel pretty similar about people who seem obsessed with certain DJs, when there is certainly an argument that all they're doing is pressing "Play" on the computer and then fiddling with some knobs. They still have 'fans.'

But, uh, we see that (some) people are readily accepting AI boyfriends and girlfriends.

And Hatsune Miku has a large following, even does concerts. Granted, the Japanese can anthropomorphize ANYTHING.

I'd want absolutely perfect health tracking, the kind of thing where the AI can tell me exactly what to eat, what supplements to take, how much and what type of exercise to do etc. Might require a bunch of implants, but I'd love to always feel well-rested and optimised.

Me, I think I want the ability to produce bespoke episodes of older TV shows that I enjoyed but were cancelled or went off the rails and/or had horrible conclusions. GOT and Firefly are obvious examples there. But I have several others in mind.

This would also be good. The petty part of me desperately wants to re-make all the BBC costume dramas without all the ahistorical diversity casting, but I'll settle for Six Seasons and a Movie of Firefly.

https://suno.com/s/voPPxtsXxRjFRF93

This could genuinely be a Paramore B-side. Crazy.

I'd want absolutely perfect health tracking, the kind of thing where the AI can tell me exactly what to eat, what supplements to take, how much and what type of exercise to do etc. Might require a bunch of implants, but I'd love to always feel well-rested and optimised.

I agree... except I'd be pretty nervous on the nefarious uses that particular setup could create.

My piecemeal solution is that I can feed it the information about myself I want it to have rather than have it tracking me full time.

I do like the thought of being able to ask how to optimize for particular goals. "I'm training for a Marathon in 4 months, create a training and diet regimen tailored to my current work schedule."

This could genuinely be a Paramore B-side.

I know, right? I've gotten it to near-perfectly emulate Rammstein, Motley Crue, and Drowning Pool, and decent-quality ripoffs of Rage Against the Machine and Linkin Park.

System of a Down has proven trickier.

The Holy Grail of AI privacy and security is one that's both powerful and efficient enough to run on your own hardware so you have full control over it and your information. In some imaginary world where your implants can run their own computations internally and don't require internet access (except occasionally if you want to download verified updates) then they're safe and secure from external threats.

I personally will never feel comfortable unless the entire system is airgapped, and maybe not even then. Think, for example, of how you have to keep a pacemaker away from electromagnets.

But life demands acceptance of certain risks, and there's a strong argument that health benefits from AI monitoring your body and suggesting adjustments would outweigh most risks from having that data exposed.

Of course, follow the logic and its just straight up transhumanism. Discard the weakness of the flesh entirely.

There's also some level of difference in invasiveness and permanence.

Removing your arm to replace with a bionic one, is categorically different from implanting an electrical muscle-growth stimulator underneath your skin, is categorically different from putting on strength-boosting mech armor.

Following the logic only goes as far as your axioms allow, and different axioms will lead to different endpoints.

I would absolutely argue we're talking a matter of degree, but yeah, if your goal is "live as long as a carbon-based multi-cellular organism possibly can in ideal health" that's different than if the goal is "avoid literally every risk of actual brain death possible, with zero regard for preserving organic parts."

But we're already blurring lines by strapping sensors and stimulators and other mechanical parts in.

Now, to put a final spin on the question, what would you DO with your now-optimized health thanks to your AI patron keeping you in tip-top shape?

My preference for the AI generating personalized entertainment options somewhat answers that question. I could watch Firefly Season 50 and then go to the theaters and enjoy "Mission: Impossible 16: Mercury Poisoning" where AI Tom Cruise sprints across the surface of Mercury while holding his breath or something.

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats