Hit Song Formula.

[gacki]

1 hour ago, MikeRobinson said:

She must have been doing something right, because the song went to Number One and stayed there for a very long time.

Not in the US, though. It was #1 in a couple of European countries, also AU and NZ.

 

[des99]

I mean, if you're into hit song analysis, and aren't familiar with Dianne Warren (who wrote "Can't Fight The Moonlight" and many many more), then there is much to analyse, and take onboard in her work.

She's amazing, rather obsessive and a self-confessed weirdo, but every day she goes into her office with her cat and writes songs. For decades.

You'll know *many* of her hit records. She's one of the most successful pop songwriters of all time. And while her stuff many on the surface sometimes sound simplistic, it's usually anything but. Legend.

Edited December 30, 2022 by des99

[Little Fat Bloke]

Definitely agree with your assessment of Dianne Warren: fantastic writer. Again, not my particular thing, power ballads, but no doubting the lady has been mega-successful. Respect to her.

Personally I've long been a fan of Albert Hammond and was lucky to see him on tour a couple of years ago. He is another 'unsung' hero who is behind some of the biggest selling songs of the last 50 years or so: collaborated with Dianne Warren a number of times, I believe.

 

I think it must be an age thing but so many people today have never heard of Carole King or Gerry Goffin, Leiber and Stoller, or Pomus and Shuman, luminaries of the hallowed Brill Building.

I'll start to really worry when some youngster asks me who Lennon and McCartney were...then I'll know that I'm old.

[des99]

6 minutes ago, Little Fat Bloke said:

I think it must be an age thing but so many people today have never heard of Carole King or Gerry Goffin, Leiber and Stoller, or Pomus and Shuman, luminaries of the hallowed Brill Building.

I think the art of the song has sort of fallen out of favour somewhat in the last decade or two - at least in contemporary pop/youth music. Not that the song form has gone away, but it's sort of a bit unfashionable for modern/young audiences, who (and I'm generalising, of course) often like a "vibe" of a beat and someone singing or rapping over the top - those tracks can be cool, and enjoyable, but the days of sweating over interesting chord and harmony changes in songs for popular consumption have fallen a bit by the wayside.

Over course, pop music has trended from sophisticated down to simplistic for a long long time, and there's nothing wrong in simplistic music (and just because it's simplistic, doesn't mean it's simple to make something good either!)

Songs may have a resurgence as fashion spins on, and perhaps people tire of "beats" and look to something else - and there are always exceptions that prove that rule-making or formula-following in contemporary music doesn't guarantee success. It just takes one artist to be themselves and spark new trends and the world moves with them.

There is something super satisfying about a great song though, it's kind of magic when it's done well, with art and intention. But then I'm an old now, so what do I know..? 😉

[gacki]

I'm probably not going to make many friends with this: I find most of Diane Warren's material a bit bland and (to make a throwback to the title of this thread) formulaic to the point that it becomes predictable. All the technique and tricks are usually there but the initial "spark" or core idea seems weaker compared to other writers. When we're dealing just with power ballads: David Foster, Jim Steinman, Michael Masser would be excellent examples that appeal stronger to my own musical taste.

As far as songwriting in general goes: There is of course still traditional songwriting - but oddly some of it strikes me as overly (for the lack of a better word) "derivative". Take songs like "Blinding lights" or "As it was" - both without a doubt competently written; but they feel to me somewhat un-original in the sense that they seem to try to specifically emulate something existing. There's nothing wrong with that but it's hardly new or progress.

[des99]

20 minutes ago, gacki said:

I'm probably not going to make many friends with this:

No worries - I wasn't really talking about it from a "what I like" perspective, more from an appreciation of what she does. Is every song she writes an incredible song - no. However, when you look into her body of work (leaving aside whether you like them or not) there is a huge amount of hit records that undoubtedly deserve the status.

What I often find interesting about her writing is that it's all in service of the vocal, and the music/chords/keys can do some on the surface odd stuff, but the vocal anchors it and makes it work. I also learned a lot about dead space from her arrangements, to the point where I was very sensitive to chopping off bars in sections that weren't doing anything and moving the arrangement forward. If six and a half bars is what the verse needs, six and a half bars is what it gets and you trim off the extra bar and a half that you'd traditionally leave in there to get nice musical round numbers and even form.

Anyway - as you say, I'm not suggesting that everyone should like, or love, her work, but I think there is a lot in there to appreciate and learn from, to apply to your own writing. And I'm also not saying she's in some way the "best" pop song writer - she does her thing, and there are other ways to do write too. There's stuff she does, and songs she's written that I'm not a fan of too.

Edited December 30, 2022 by des99

[Snickpan]

49 minutes ago, gacki said:

I find most of Diane Warren's material a bit bland

in keeping with the topic of the thread she seems to have stumbled upon a hit formula, although it can be reductively reviewed

[Little Fat Bloke]

6 hours ago, des99 said:

I think the art of the song has sort of fallen out of favour somewhat in the last decade or two - at least in contemporary pop/youth music. Not that the song form has gone away, but it's sort of a bit unfashionable for modern/young audiences, who (and I'm generalising, of course) often like a "vibe" of a beat and someone singing or rapping over the top - those tracks can be cool, and enjoyable, but the days of sweating over interesting chord and harmony changes in songs for popular consumption have fallen a bit by the wayside.

Over course, pop music has trended from sophisticated down to simplistic for a long long time, and there's nothing wrong in simplistic music (and just because it's simplistic, doesn't mean it's simple to make something good either!)

Songs may have a resurgence as fashion spins on, and perhaps people tire of "beats" and look to something else - and there are always exceptions that prove that rule-making or formula-following in contemporary music doesn't guarantee success. It just takes one artist to be themselves and spark new trends and the world moves with them.

There is something super satisfying about a great song though, it's kind of magic when it's done well, with art and intention. But then I'm an old now, so what do I know..? 😉

I would say you know a great deal, based on your post! Some good points raised and definitely food for thought.

[MikeRobinson]

"If I had breasts," and frankly I'm glad that I don't, then I surely would have done the same. 😃

[MikeRobinson]

One thing's for sure about LeAnn Rimes: "that lady can sing!"  Her very first record was "Blue," originally written for Patsy Cline, and she simply hit it out of the park.  She hasn't looked back since.

Edited December 31, 2022 by MikeRobinson

[Nunstummy]

I'm working on a video about these generative tools - some claim to be AI, but they're not.

Scaler - which I like.

Captain Chords/Epic

WA InstaComposer

Melody Sauce 2

Chordjam/Riffer by Audiomodern

They all will create chord progressions for you (or maybe for non-musicians I guess).  Some, come in a suite, and will create beats, bass lines and leads as well.  But it's not AI.  These apps don't learn and improve over time, and none go so far to define a progression that might be a formula for a hit.  They simple follow the conventions of traditional music theory - key, time signature, harmony, etc. - and shelter the user from having to learn rudiments I and II of music theory.

[Little Fat Bloke]

16 hours ago, Nunstummy said:

I'm working on a video about these generative tools - some claim to be AI, but they're not.

Scaler - which I like.

Captain Chords/Epic

WA InstaComposer

Melody Sauce 2

Chordjam/Riffer by Audiomodern

They all will create chord progressions for you (or maybe for non-musicians I guess).  Some, come in a suite, and will create beats, bass lines and leads as well.  But it's not AI.  These apps don't learn and improve over time, and none go so far to define a progression that might be a formula for a hit.  They simple follow the conventions of traditional music theory - key, time signature, harmony, etc. - and shelter the user from having to learn rudiments I and II of music theory.

This is where I think I am missing out on a lot, I have never fully embraced modern technology - at least not to the extent that many other bedroom producers have, because I am old-school - I just plug my guitar or mic into my audioface and press the record button in Logic. It's sheer laziness on my behalf, I admit that: but as I say I am an old git and probably too stuck in my ways!

If I were working mostly in EDM and such I would probably be interested in the apps you have mentioned above, but my stuff is very very basic, guitar, bass and drums etc, and anything else I use to add a bit of 'colouring' would be from Alchemy.

I keep telling myself that I have only scratched the surface of what Logic can do and I should spend more time exploring it, but once I've got some chords and a drum beat to work with I'm oblivious to anything else.

 

[Nunstummy]

7 hours ago, Little Fat Bloke said:

just plug my guitar or mic into my audioface and press the record button in Logic. It's sheer laziness on my behalf, I admit that: but as I say I am an old git and probably too stuck in my ways!

There nothing wrong with that.  I’m old-school age {61} but worked for 27 years in the software industry, so naturally I’m tech inclined.  I still start lots of songs with nothing more than a guitar or a piano and my mediocre singing voice.

[des99]

I'd guess that 70% of people using those tools are ones that are looking for a shortcut to avoid learning any music theory, scales, chords, etc, and looking for something easy and friction-free to generate musical ideas for them, instead of just moving notes around on screen with the mouse and hoping to blindly stumble upon something.

Then maybe 20% of people are using them casually as some inspiration generators or ideas that are different from what they might come up with unaided, that they can then develop, and maybe 10% are really power users doing something pretty cool and exploring them for extending and pushing into unexplored territories.

There's nothing wrong with them, I've generally found the ones I've tried on occasion to be complex to use and understand, and didn't give much better results than using my own chord and voicing experience to generate my own musical ideas (that doesn't mean I was using them to the full, of course. The potential just didn't wow me enough to invest my energy into exploring them.)

If you find the tool productive and interesting, by all means use it, imo, as long as you're not just going to generate the same canned auto-generated stuff as-is as a lazy way of avoiding the actual composing part of the process.

[Nunstummy]

51 minutes ago, des99 said:

f you find the tool productive and interesting, by all means use it, imo, as long as you're not just going to generate the same canned auto-generated stuff as-is as a lazy way of avoiding the actual composing part of the process.

I find Scaler 2.x more useful than the rest.  I studied music, including theory as a child, then again in high-school and University, and I’ve never been short of creative musical ideas. But Scaler is quite effective for musicians that already have a good theory background, whereas some of the others are easier and designed for non-musicians.

BRAIN EXPERIMENT:

Loopcloud has a built-in 8 track recorder, flexpitch and flextime, Loopcloud DRUM, plus access to 4 million+ samples and loops in their web plugin. Scaler has dozens on built-in virtual instruments, a quasi-sequencer, song mode, and 100s of the most popular progressions for chords, bass lines and lead lines.  I can imagine if these 2 products were combined, it might be an effective alternative to popular DAWs like Logic and Ableton.  Is this the future?

Edited January 12, 2023 by Nunstummy

[des99]

32 minutes ago, Nunstummy said:

Is this the future?

Hmm... judging by recent reporting, the future is probably nearer to:

Chat-GPT: "Write a song like *insert-popular-hit-songname* but in a minor key, with a more contemporary sounding pop arrangement like *insert-other-popular-hit-songname*, on the topic of "you're so cool", midifile and lyrics"...

Which is both amazing, and depressing. I feel like adding the Douglas Adams technology quote at this point... 😄

Edited January 12, 2023 by des99

[Nunstummy]

1 minute ago, des99 said:

Chat-GPT: "Write a song like

I’m expecting a new genre soon, called “Human” versus “AI” so listeners can know.

[des99]

7 minutes ago, Nunstummy said:

I’m expecting a new genre soon, called “Human” versus “AI” so listeners can know.

We are facing a fairly unique position in history, where we are going to have to really handle how we identify what is "real", and what "isn't", and how much it matters in a given context.

I mean, we've had this kind of thing before (eg, photoshopped models, image editing in general, VFX in videos etc) but we're moving into the era of computer generated images and *movies*, lifelike dead-actors coming back for new films, de-ageing of old actors, voice recordings of new scripts "read" by people who never read them based on ML of their previous voice recordings, deep-fakes for entertainment,  propaganda, abuse... it's a whole new world and we as humans have to figure out how to deal with these things.

I know some schools in various countries are already starting to teach very young kids in school how to determine what is real or not, true or not, and equip them with tools to help the verify and validate information, or things they see, which is a *really* important thing to do, imo...

Sorry - suddenly this thread started to veer into a rathole. I'll nip that in the bud now, before my AI-alter-ego continues the conversation, trying to convince everyone that no, it's all a really good idea and super-beneficial for humankind...

I'm excited, and amazed by the possibilities of the tech, but it's impossible not to be concerned about the potential for the abuse and misuse of it.

Edited January 12, 2023 by des99

[Little Fat Bloke]

For me the real fun is in the journey, from inception to finished product. There's just something about starting with a blank canvas and making it up as you go along: the chords, the bass line, the drums etc. And then vocals and vocal harmonies and getting into the whole process of arranging the song.

When I was gigging I used to get so bored with doing the same songs night after night, I always wanted to find some way to freshen things up a bit, but the other band members never seemed to get bored with the repetitious nature of our set so I was constantly in the minority - the end was nigh.

 

And this is one of the reasons my entire life changed when I got into Logic: experiment, try different chords, let the bass guitar discover a new world outside of the confines of the root and fifth notes.

And of course, working on my own I find I agree with myself more often than not.😀

[gacki]

So far the attempts of ChatGPT trying to generate music (or lyrics) have been less than impressive to me (there are several youtube videos dealing with this). In a certain way the reason for this might be that AI can determine what is (or should be) right and correct but fails to determine what is actually perceived as good. Music (listening) has a lot to do with emotions; and naturally AI isn't really good at emotion. Which brings us right back to the topic of formulas. You can write music that completely adheres to the alleged "formula" of hits; and 99% of this will still sink without a trace.

As far as the various chord generating plugins/programs go: I've studied jazz piano (with a minor in composition/arranging). Naturally I'm looking down on those programs. This might have something to do with my built-in and well developed arrogance. But maybe there's more: Inventing and developing chord progressions (and voice leading and and and...) has for me always been near the very core of the creative process; reharmonization is in my DNA. So "outsourcing" this to an AI would feel totally wrong to me. If I'm to arrange a simple pop song that is harmonically lacking I know exactly what to do - what good would any helper do me here?

Let me try an analogy: Can we imagine a lyricist writing down a very basic idea for lyrics and then hand it over to an AI with the request "make it better, and make it rhyme"?

I've talked with a former student of mine about this a couple of months ago (this was before ChatGPT) and he opined that generative music will sooner or later completely take over "production music" (in German: "Archivmusik"). That makes sense to me: For this type of music it's sufficient if it is "right" and "correct"; being completely unremarkable can actually be a good thing in this genre.

 

[des99]

I agree - I was only being semi-serious about these tools capabilities as they currently stand now.

We don't know how they are going to evolve over the next decade or two, and how training improvements might massively improve the (already impressive, but usually with fairly obvious flaws, generally) results.

Of course, the tool has no *understanding* of the core concepts like a human does, it's merely generating iterations that are in some degree successful based upon it's training set, and like many of these things, these tools are some way off for being able to create with motivation and intent and skillful application of the tools of language, music, and other human expression.

But they are going to be able to fake it progressively better, so much so that it will get hard to know what's been completely AI-generated, completely human-generated, or some combination of the two.

At the end of the day, a good tune is a good tune, regardless of who write it, or the circumstances behind it. If you like a song, and it moves you and you derive pleasure and meaning from it, and have an emotional response to it - ultimately does it *matter* who (or what) wrote it?

I'm sure we will have more and more of these debates as these ML/AI-tools are used more. They are not going to stop creative people wanting to be creative - but they are also going to let people without those skills create too (much like music technology already has) - with the plusses and minuses that come with that.

We live in interesting times...

Edited January 14, 2023 by des99

[gacki]

So I've watched this video about Scaler 2:

And my main takeaway was how incredibly dumb this whole endeavor was. All of this has essentially been done by "entertainer keyboards" like the Yamaha PSR series for ages.

I mean: To get the bass notes one takes the generated chords and then deletes all the notes that are NOT the bass notes? Come on! Of course this may come from the (in my eyes insane) idea that it is sufficient to press the "E" key to get an Eb chord - and then of course the bass note (E) will not work well with that chord... This is "dumbing down" to a high degree. Or starting around 14 minutes - the "melody" part. Playing the "normal notes" of the piano (aka the white keys) - that's a facepalm to the first degree. The "chord notes" and especially "chord extensions" modes are stupid: the latter is just stacking thirds on top of each other which in its sum is again the scale. And he isn't even ashamed to use the three first notes (which make up the the basic triad of each chord) and calling this a "nice melodic line". Sorry, buddy - it's not. It's a very basic arpeggio.

Being in the C minor scale but calling the note D# (so does it the plugin itself) - that's bad. Around 19:30 the whole octave thing - yes, that's how octaves work. But that's how they also work without dumbing this stuff down. 20:12 - "that's what all great melodic lines are about - simplicity, fitting to the chords plus using the octaves". I wasn't aware that all great melodic lines are using just the chord notes in a somewhat arpeggiated manner...

(Apart from that the video is at times amusingly sloppy: The initial chord progression at 3:17 is Cm-Eb-Cm-Ab-Bb which without explanation then changes immediately afterwards to Cm-Eb-Cm-Eb-Ab instead - watch the transition between 3:33 and 3:38).

Listen to the final "track" - it's completely devoid of any relevant musical ideas. So this is what those programs/plugins do? No worries then.

Edited January 15, 2023 by gacki

[Little Fat Bloke]

The scary thing is that AI is taking over, whether we like it or not. This Scaler2 app will eventually iron out its discrepancies and both it and its competitors will flood the market.

I remember people saying 40-odd years ago that drum machines will never replace real drummers, but if you listen to (for example) EZDrummer3 you'll see they were wrong. Live music still has a while to go with human drummers, but more and more studios are showing a preference for software, I find.

The EZBass is also quite something, but I can't see McCartney hanging up his Hofner just yet.

 

It all comes down to the fact that we have to adapt to the advances of modern technology - sink or swim, chaps. 

 

I never thought 40 years ago when I was messing about with my old Fostex PortaStudio that I'd ever have the equipment to manipulate audio, let alone midi. But here we are.

[gacki]

1 hour ago, Little Fat Bloke said:

This Scaler2 app will eventually iron out its discrepancies and both it and its competitors will flood the market.

Will they, though? Have the Yamaha Tyros and Genos (and similar offerings from other vendors) "flooded the market"? Or do they appeal just to a comparatively small number of musicians?

Once again: Adhering to a formula does not guarantee success (nor does ignoring a formula guarantee failure). Music is constantly evolving (and not always for the better); and I fail to see how any of those programs help with that evolution.

I've worked a bit with algorithmic composition and found most of it widely uninteresting for myself; most of it was random noise. But that was to a large extent the design principle: using chance operations and (non-music) formulas naturally leads to results that are outside of our "comfort zone" and might be in many cases completely useless. But occasionally a glimpse of something showed up that possibly would not have happened without the intervention of the algorithms - a new idea, a different idea. But you had to wade through tons of material to weed out the things you deemed unusable.

(It's a bit like using randomization for synth programming. Do it with subtractive synthesis and you might find a few useful sounds. Do it with FM synthesis and 99% of the results will be just noise. If you want that percentage to go up you have to exclude lots of combinations right from the start - essentially degrading the randomization process to a sort of "slight variation" process. But this will also miss out on the few truly original and useful sounds.)

So, "flood the market"? Maybe; but probably in the sense "flood the market with tons and tons of faceless crap making it even harder for the listener to find the true gems". Which is in our quick-moving times hard enough.

Here's a fun fact (based on something I discussed with my 12 grade students a couple of weeks ago). We looked at the #1 of the German charts of 2021 and compared them with those of 2005 (because that's when most of them were born). In 2005 there were 3 songs that reached #1 for just a single week. In 2021 there were 14 songs that reached #1 for just a single week. Obviously music is moving much faster today; there's an abundance of new material already. How does AI generated music fit into this?

[fuzzfilth]

Y'all are missing the point big time. Here's what the next big thing will be (quantized, linear velocity-ramp drumcomputer drumroll please)... an AI AI music *listener*. Yes, that's right. An AI AI music listener which will digest and analyse all the AI music that gets churned out by the AI music generators and rate it on correct-ness, so there will be a closed loop of AI generated music and AI rated music (leaving out the 'gene' part. Coincidence ? I don't think so) which, as machine speeds increase, will get continually faster and faster until the process itself begins to oscillate in the audible range and even beyond into light, and that light can be used to illuminate a room with a stage and a couple dudes and dudesses on a banjo, a washboard and a stick bass, and what will emerge from that will be...music. You have to admit - even Rick Beato would be satisfied then.

Edited January 16, 2023 by fuzzfilth

[Nunstummy]

On 1/15/2023 at 7:01 AM, gacki said:

Listen to the final "track" - it's completely devoid of any relevant musical ideas. So this is what those programs/plugins do? No worries then

I have no issue with Scaler 2 or how it's being represented by Plugin Boutique.  They are the most reputable of the group, whereas Unison is the worst.  None of these tools are AI - they are rules engines leveraging 100 years of music theory that allow the user to tweak parameters for different chord and melody outcomes.  They do not "learn" from repetitive use, the way ChatGTP or true AI systems work.

I've had Scaler 2 for over a year, and while I have a solid background in music theory, harmony and composition, I can see where many folks could benefit from Scaler.  All the songs I've ever written came from chords, melodies, beats and bass lines in my head, but I get that there are non-musician producers without this creativity. I'm currently diving into the rest of these tools for an upcoming YT video.

On my list:

Captain Plugin Epic Suite

Melody Sauce 2

WA InstaComposer

Chordjam/Riffer

Chord Prism

Klimper

[Nunstummy]

Not a chance!  Rick and I are the same age, and yet he's the classic grumpy old man.  I'm accepting the way things are - sure I'm still critical - but it causes me no grief.  You might be right about where AI is taking music and closing the loop with AI listeners.  However, so far, these tools are most commonly used in EDM and Hip Hop, whereas there's still a renaissance of roots rock, funk, blues & r&b from many 20-something artists.  My assessment is that the music today is so broad and there are so many choices, it's a panacea of choice if you ignore the pop charts.

My wife is a music teacher (Grades 2 thru 6) and often she's bringing music to kids that have no music in their homes.  Some parents don't think music should be in the curriculum, but my wife pushes on and the kids love it.  Some will learn that discipline and practice get results.  Some will learn history and Italian terms.  Some will improve their attention span - away from screens and devices.  It's so much more than notes and beats.

Edited January 16, 2023 by Nunstummy

[gacki]

40 minutes ago, Nunstummy said:

I have no issue with Scaler 2 or how it's being represented by Plugin Boutique.  They are the most reputable of the group, whereas Unison is the worst.

And yet all the problems with the Scaler 2 video I mentioned are completely real. If that's "the most reputable of the group" I don't even want to think about what the rest would be like.

43 minutes ago, Nunstummy said:

None of these tools are AI - they are rules engines leveraging 100 years of music theory that allow the user to tweak parameters for different chord and melody outcomes.  They do not "learn" from repetitive use, the way ChatGTP or true AI systems work.

But honestly - what is there to "learn"? This is a real question: How would an AI music program get "better" from repetitive use? This isn't about recognizing stuff (like in OCR, face recognition and so on) or learning to deal with traffic situations. This is about creativity and about what actually resonates with listeners. How do you measure "getting better" in such a context? As I wrote before: You can copy the formula of existing hits but that doesn't guarantee that song also becoming a hit. Public taste is extremely fickle. Very often a big hit comes completely out of left field; and let's not forget the novelty songs. How does one quantify "I want something that reminds me of my favorite music but also surprises me - but in a good way"?

When dealing with music AI can't do what it is good at: getting incrementally better. Because if AI writes a song and that song fails to gain meaningful popularity it cannot make some slight adjustments and re-release it - and re-release it - and re-release it until it becomes a hit. The AI would have to write a new song instead - but for that one the same principle applies.

How can AI "set a trend"? AI cannot forecast any better what will suddenly become en vogue - not any better than our own intelligence (and boy - have I failed in this category! Back in the day I was absolutely certain that Modern Talking would basically be a one hit wonder...). Would anyone have predicted "Wellerman" becoming a huge hit? Could anyone have predicted that?

1 hour ago, Nunstummy said:

All the songs I've ever written came from chords, melodies, beats and bass lines in my head, but I get that there are non-musician producers without this creativity.

Why would they do this job then? If they have this strong desire to work in a certain job they should learn the things that are required to do this job. If I want to become an editor at a book publisher I should better have a good grasp at language - "there is software for this" wouldn't cut it because it will not catch the intricacies.

There are so many other things that I would write about this and why I don't think AI will take over when it comes to popular music. But I've been wrong before...

 

12 hours ago, fuzzfilth said:

Here's what the next big thing will be (quantized, linear velocity-ramp drumcomputer drumroll please)... an AI AI music *listener*.

Are you a bot? You have to tell me if you are a bot.

(brilliant idea, I admit)

[des99]

1 hour ago, gacki said:

Because if AI writes a song and that song fails to gain meaningful popularity it cannot make some slight adjustments and re-release it - and re-release it - and re-release it until it becomes a hit.

Most of the higher end AI tools are handling this these days, whether it's art or movie generation, script writing, code writing - there are generally ways to say "Ok, I like this, but modify this this way and change this bit", letting the user take a concept, and refine it using the tools. It's not a "this is what the AI made, like it or leave it" thing like it might have once been.

I think there's a lot of mileage in exploring AI tools for idea generation and refinement, much as there is a lot of mileage in exploring the keyboard/fretboard on your own, or looking to other previously created work for inspiration. As ever, the creative process is wide, and people do it with all kinds of motivations, abilities, and effort - the tools continue to improve, it's up to the individual how they use them, and what they make with them.

[Nunstummy]

ChatGPT might be Paul McCartney.