|
Post by geoff738 on Jan 6, 2024 16:10:06 GMT -6
Aye Drake. I’m in Toronto and, um, yeah. Don’t get it. But I’m old, so there’s that.
Cheers, Geoff
|
|
|
Post by unit7 on Jan 6, 2024 17:13:45 GMT -6
I believe AI will have a big impact and I'm scared and not very optimistic. At least from a musician's and audio engineer's point of view. A Donald Fagen line comes to mind.. "A just machine to make big decisions/Programmed by fellows with compassion and vision/We'll be clean when their work is done/We'll be eternally free yes and eternally young"
Another aspect is the regrowth of new musicians. Especially in classical there has been a bad trend for some years. It's becoming harder and harder for the symphonic orchestras to find/employ musicians that are up to given standards. Many say that in this digital age it's hard to motivate kids to practice for years to learn an instrument. I don't believe AI will change that, most probably the opposite.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 6, 2024 17:22:20 GMT -6
There's some very nuanced and insightful / informed thoughts here, and some quite naive (under informed) ones as well. One thing is certain. The near future is going to reveal the real truth.... AI is growing exponentially fast. That much is fact and undeniable. It is scary that some of the most informed tech gurus are shaking in their boots. It already is but it's not doing anything really for creativity in mixing or mastering. It's just trained recursive algorithms, statistical prediction, bayesian networks, and not skynet. Mainly tonal balancing, nuking resonances (TDR smart ops), nuking the high end (Soothe), brightening the high end (Gulfoss), suggesting tonal balancing and ham fisted dynamics processing based on genre like Ozone, and all the program dependent processors going back to the optimod, compellor, and drawmer auto comps. The limiters have always had "AI" and now stuff like the DSM and Elevate take broadcast like tools and give it them to everyone with a computer cleanly and dirtily respectively with way more program dependency than something like DynOne or multiband garbage. The broadcast tools have insane program dependencies now that dwarf anything else even if some limiters are cleaner or can do something special.
The problem is the music industry rejected innovation and well-built equipment sometime in the 90s to present in favor of ersatz of dark age processors and abusing the "program dependency" or "ai" of digital dynamics processing to ruin recordings. Look at the clone wars. Just go to the recent Aphex Compellor mod thread and hear what a modern processor does and doesn't do to a drum kit versus the "classic" and the "modern classic" and the Compellor only has a fraction of the control and program dependency of modern tools.
The biggest danger is AI writing music, synthesizing performances, and productions. Making a computer voice sing. Playing sample backs and synthetic instruments. More than AI tools driving us out of a job or some kid with Ozone or TDR Limiter 6 not being able to pay a mastering engineer anything. Not even 20 a track but he got Ozone Elements for free and Limiter 6 for 10 bucks. The horribleness of hackjob EDM productions shows us what these sound like when used by those incapable of harnessing them. Elevate is also being abused to really push stuff through the speakers even though it can easily be worse than L2 should the final clipper module be abused.
Dan
|
|
|
Post by noob on Jan 6, 2024 17:37:06 GMT -6
I'm not really worried about the average person using AI. I'm worried about how it will get capitalized on by big corporations and popular music will become even more bland and formulaic. Mastering engineers will probably be the first ones to lose out imo, but the great ones will always be needed. I'm not one to prophesize though, I just hope it makes people more creative and smart, not the opposite.
|
|
|
Post by thehightenor on Jan 6, 2024 17:47:14 GMT -6
Imho - we are way more advanced technologically than we are in our capabilities to manage the technological advancements in a good way. If Covid is a lab accident, that is just one recent example that killed millions. And if it was not a lab accident, the point is that it could have been. I do not want to argue whether in fact it was a lab accident or not as what the heck do I know, but that is not the point. We have that capability, and that is the point. And another point is that we are doing little to stop experimenting with technologies that have consequences beyond our comprehension. How many of us really thought, let's say ten years ago, that there would be a worldwide pandemic that killed millions because of something so microscoptic? Human history shows that any negative side of a technology will eventually be used. So, we do need to consider the negatives of AI. And the time to do that is now. Look at all the computer viruses, spybots, authorizations, ilok, passwords, etc. - all signs that we have too many dishonest and nefarious people in the world. We all suffer because of these sad character traits. How many more advanced technologies do we want to give to these people? The telephone was a level down from in-person communication. And text chats are a level down from the telephone. What is the next level down? Pretend people? And what are the consquences for human relationships? Will people be even capable of looking at each other in the eyes anymore? Urban dwellers lost the night sky with artifical light. What do we lose with each so-called technological advancement? Imho, that must be considered. Now come up with a technology that improves human character development - that would be something. But that does not seem to be the focus. Someday it might have to be. Good post. And more to the point we keep stacking bullets into an already loaded barrel. Every time I watch the news recently, I'm reminded of the fact - grave danger remains ever present from technology developed in 1945 !! When will we escape the "Monster from the Id" as Dr. Morbius reminded us in Forbidden Planet. People worry so much about climate change .... I worry about people induced change! I know it's ideological, but it's 2024 and yet we have never seemed further away from coming together collectively to solve our collective problems as we do today. If Aliens parked up near the Earth to watch us - globally as a species - they would scratch, whichever part of their body is their head, and say "what the heck are these creatures doing to each other and their environment - how odd"
|
|
|
Post by Tbone81 on Jan 6, 2024 17:53:27 GMT -6
I wish I could be as optimistic as some of you here, I really do. But after listening to a few podcasts with top AI Engineers saying that they literally don’t know how the AI’s work…that they just program the conditions of the AI to operate and then wait to see what the output is…well, it’s rather frightening to me.
A few years ago AI couldn’t write any songs at all, now we have some mediocre songs that are still better than half the bedroom “producers” out there. Maybe that’s a low bar…but exponential growth is exponential growth. In my mind the question is not “if” AI will be to replace us all (AE’s, songwriters, musicians etc) it’s simple a matter of “when”.
The only thing that really keeps me from worrying about it is the realization that impact on society as a whole will be orders of magnitude greater than anything that happens to our small industry. That’s what really keeps me up at night.
|
|
ericn
Temp
Balance Engineer
Posts: 16,098
|
Post by ericn on Jan 6, 2024 18:39:01 GMT -6
The first major AI effect on music creatives Apple Music’s Karaoke, lots of venues that used hosts who paid royalties are going cheap and using Apple Music subscriptions for background and Karaoke.
We can all hope hope that the next thing we see is AI applied to OS and better multi core utilization, but that’s a pipe dream.
AI will probably show its ugly side first, a “ Taylor”, “ Madonna “ and “ Beatles “ style button in Garage Band followed by an auto vocal comp feature, why just put down a couple of vocal passes on the same time line hit a button and there is your comp. Eventually a more pro version with some control will appear in logic. Because of the $$ in AI development I suspect Apple will be the first in our arena.
This is a completely off the wall interpretation but it really makes sense when you think about it, UA has heavily invested in Luna, Waves, who have found a nice niche in live via Soundgrid being used with 3rd party consoles have been developing their own mixing platform that is a native equivalent to a digital console, why? Again this is a guess, but as a plugin developer if you wanted to develop AI mixing tools you would need to move beyond the structure of a modern DAW, you are going to have to develop an architecture that gives you feed back on every input. Sure you could wait for the DAW vendors to do it, but what are the chances of Avid, Apple, Steinburg / Yamaha etc to all follow the same path? Probably better to develop in house where you control everything.
|
|
|
Post by ab101 on Jan 6, 2024 20:07:51 GMT -6
I was listening to a broadcast about how kids are now growing up with "full service" cell phones with the internet, social media, etc. at young ages. The prior generation usually only had phones for phone calls and texting. But the young ones now use the cell phone for everything - instagram, tik tok - news, etc. They are bonding to non-humans. Instead of playtime with kids outside, it is about more insular high tech. And the "likes" are used for addicition and ego catering. This is leaving to a vast emptiness as ego chasing does not lead to "filling up." There is no more patience. So, many kids today are having mental health issues.
So, what does this mean about AI and music? I suppose these kids will grow up not caring about whether the music is made by human beings!
And yes, Eric - I see that modeling expanding. Just like modeling different hardware, now the modeling will be for iconic people. By the way, there are already choir libraries with different consonants, etc. so one can make a choir sing lines without recording an actual choir. The idea this will lead to solists as well is likely.
I live close to the Southwest, where one can see ancient "ruins" showing past people that appear to have a high degree of spiritual awareness, but not very technologically advanced - as the stone age ended here last on the planet I believe. How much porno or violence between humans are on those petroglyphs? Almost none.
But the ancient ruins of today's world will show humans with great technological advancement and demented spirituality if any. Violence is celebrated today in video games, etc. Even wolves generally do not kill for sport.
Where is the Bob Dylan of today? (Who is shaking these windows and rattling these walls?) Who is raging against this commercial demise of our internal core?
I strongly am in the camp that "Art is a hammer," not just a mirror. It is time for some real hammering.
Addendum: I am not really pessimistic. I just think it is wise to see the benefits and dangers of new technologies. New activists will arise. And I do believe there is a spiritual revolution is lying ahead. Perhaps we just need to "bottom out" first.
|
|
|
Post by bossanova on Jan 6, 2024 20:26:41 GMT -6
I’m 38, I have a kid, my wife works in tech, and I think y’all are being real pessimistic. As a songwriter (not a great one, but someone who has put a lot of study into what works and what doesn’t, what makes someone else’s work great), AI isn’t even close. Band in the Box is still the best melodist I’ve heard and that’s 20+ years old. The best language models write songs that rhyme and match a certain topic, but they don’t surprise. They don’t show insight. They don’t coin new metaphors. Most can’t even vary form. 2nd rate Jingle generators at best.
It might be coming eventually, but I think true dominance in artistic fields is farther away than some people think.
|
|
|
Post by viciousbliss on Jan 6, 2024 21:13:34 GMT -6
Check out these headlines from the latest articles:
AI GETS BUSY HELPING RESEARCHERS CREATE THE FIRST SYNTHETIC BLOOD PLASMA
WALMART’S SUPPLIERS WOULD RATHER NEGOTIATE WITH AI THAN A HUMAN
THIS NEW AI TOOL FROM HEYGEN DUBS YOUR VOICE IN SEVEN LANGUAGES
IN BIZZARE TWIST SCAMMERS ARE SCAMMING SCAMMERS WITH SCAM GPT’S
THE FIRST 247 AI NEWS CHANNEL FINALLY GOES LIVE
IN A FIRST A HACKER IS DEMANDING FULL CONTROL OF A MAJOR COMPANY
There was more stuff. An Ultrasound printer printing organs inside a body. Researchers at Indiana U connected a sort of artificial cortex made from brain cells to a chip packed with electrodes and it could learn. Even beating AI at solving certain math problems. It's called Brainoware. It's looking like we're getting closer to being able to augment ourselves with cybernetics and other things to vastly increase our own abilities. There's just so much going on with emerging exponential tech. It's all connected, so it's tough to look at something like AI and think about it in solely in terms of music. Just because that's assuming music itself won't be completely reinvented with things like the DAW and stereo sound becoming archaic. Very few people could begin to really give a qualified answer as to what happens with all these emerging technologies.
I included the last headline after I read the article. There will be a need to go back and use older technologies or even methods that don't rely on any tech at all in order to keep things secure from hackers. There's really not a huge point in thinking about AI in music aside from how it integrates at the moment and whether you'd rather sell your stuff and do something else while your gear and plugins have some value. The future is really beyond our control. By the time a natural human can be considered obsolete for audio work, a big percentage of current jobs will have been automated, resources won't be as scarce, and prices should have fallen a ton on most things. Eventually AI and augmented humans will be running everything, but that's obvious enough. The average person I know outside of the Internet always tries to forecast the future based on the past. Imagine trying to forecast the future in 1924 without having any real knowledge of the emerging technologies of the time. Television technology had a lot of advances throughout that decade, but the average person of 1924 probably had no idea what was around the corner. Eventually we'll see a radical redesign of human society, much as our society today is radically different from say 1850, but this will be even bigger. I'm not really worried about AI doing something horrendous. AI isn't going to have random robots shooting up buildings or anything.
|
|
|
Post by drbill on Jan 6, 2024 21:28:34 GMT -6
My only real positive thoughts about the AI dilemma are that hackers will probably find ways to take it down.
|
|
ericn
Temp
Balance Engineer
Posts: 16,098
|
Post by ericn on Jan 6, 2024 22:40:17 GMT -6
My only real positive thoughts about the AI dilemma are that hackers will probably find ways to take it down. Nah the hackers will use it to crack your bank account and personal info without fishing😁
|
|
|
Post by Johnkenn on Jan 6, 2024 23:47:20 GMT -6
The Robots will kill us all.
|
|
|
Post by christopher on Jan 7, 2024 0:20:57 GMT -6
I haven’t found a use case for anything. It’s a fun toy! For example.. I asked an image generator to make an image of a sexy woman singing into a mic in the studio. I expected a typical looking singer and LDC.. it gave me a bunch of pics of a supermodel with D cups, topless 😳 holding a blurry SM58 in various positions. Not exactly what I wanted, so I asked for more revisions. No matter how many revisions and multiple sexy singers I ask for, the only mic was a 58 they shared. Just a waste of time
|
|
|
Post by rowmat on Jan 7, 2024 2:59:16 GMT -6
|
|
|
Post by thehightenor on Jan 7, 2024 3:21:51 GMT -6
I’m 38, I have a kid, my wife works in tech, and I think y’all are being real pessimistic. As a songwriter (not a great one, but someone who has put a lot of study into what works and what doesn’t, what makes someone else’s work great), AI isn’t even close. Band in the Box is still the best melodist I’ve heard and that’s 20+ years old. The best language models write songs that rhyme and match a certain topic, but they don’t surprise. They don’t show insight. They don’t coin new metaphors. Most can’t even vary form. 2nd rate Jingle generators at best. It might be coming eventually, but I think true dominance in artistic fields is farther away than some people think. +1 I keep saying it. It’s called Artificial Intelligence. Songwriter, creativity doesn’t come from raw brute force “intelligence” It comes from your mind being connected to your heart and soul. Not from a Neural Network connected to a farm of Nvidia GPU’s.
|
|
|
Post by viciousbliss on Jan 7, 2024 5:24:05 GMT -6
Hypothetically speaking, would you guys trade your audio career for infinite or near infinite resources, the ability to have any medical issue solved, the ability to become augmented, a cyborg, or something better, full immersion VR, and to experience things that would be more stimulating than anything we can imagine? I think most wars are over resources. They're done with a scarcity mindset. Crime is done to obtain resources or a thrill. Both crime and war will be a thing of the past at some point. We already have some basic gene editing today. Eventually that will evolve into something better than Gattaca and we can weed out antisocial disorders. Movies like that are just an exercise in thought because they only accelerate one aspect of technology and leave everything else locked into the time period it was made in. Even something like Star Wars feels like its world has limitations of 20th Century America. They've got cybernetics and light speed but people still age and drink blue milk? They don't even have cell phones..hah
Let's say AI does cause demand for audio services to drop and rates to plummet. Do we need a contingency for that? Audio won't be the only field where automation happens. Everyone can't go become a plumber, there will have to be a UBI amount that is a livable income. I think the more complex engineering and song-writing will take a lot longer to replicate than these formulaic modern pop songs. But we've already lost so much of that complex stuff because musicians either can't find band members or don't have the resources to make it. We could see a renaissance of Master of Puppets quality work at some point if there are people out there with a desire to create still. So much of our modern DAW music sounds fake and robotic as it is. AI may kill the demand for that and create demand for more complex things. That MOP kind of quality writing and engineering does not exist in modern music really. If I recall, Kurzweil forecasted that by 2070 or 2080 that currency would have no value anymore.
There's really no harm in pushing your skills and knowledge as much as possible. I think we're seeing the worst of the dog eat dog race to the bottom in audio right now. The people who are really dedicated and keep going will be the ones who stay relevant the longest. And there will probably be communities with different levels of technology. So maybe if you want to live in 1995 indefinitely, you could do that and there will just be some provision for people to get access to 2040s medical care. This isn't gonna be something where everyone's jobs are automated and we're all in breadlines begging some terminator for a weekly ration.
|
|
|
Post by seawell on Jan 7, 2024 5:33:57 GMT -6
Hypothetically speaking, would you guys trade your audio career for infinite or near infinite resources, the ability to have any medical issue solved, the ability to become augmented, a cyborg, or something better, full immersion VR, and to experience things that would be more stimulating than anything we can imagine? I think most wars are over resources. They're done with a scarcity mindset. Crime is done to obtain resources or a thrill. Both crime and war will be a thing of the past at some point. We already have some basic gene editing today. Eventually that will evolve into something better than Gattaca and we can weed out antisocial disorders. Movies like that are just an exercise in thought because they only accelerate one aspect of technology and leave everything else locked into the time period it was made in. Even something like Star Wars feels like its world has limitations of 20th Century America. They've got cybernetics and light speed but people still age and drink blue milk? They don't even have cell phones..hah Let's say AI does cause demand for audio services to drop and rates to plummet. Do we need a contingency for that? Audio won't be the only field where automation happens. Everyone can't go become a plumber, there will have to be a UBI amount that is a livable income. I think the more complex engineering and song-writing will take a lot longer to replicate than these formulaic modern pop songs. But we've already lost so much of that complex stuff because musicians either can't find band members or don't have the resources to make it. We could see a renaissance of Master of Puppets quality work at some point if there are people out there with a desire to create still. So much of our modern DAW music sounds fake and robotic as it is. AI may kill the demand for that and create demand for more complex things. That MOP kind of quality writing and engineering does not exist in modern music really. If I recall, Kurzweil forecasted that by 2070 or 2080 that currency would have no value anymore. There's really no harm in pushing your skills and knowledge as much as possible. I think we're seeing the worst of the dog eat dog race to the bottom in audio right now. The people who are really dedicated and keep going will be the ones who stay relevant the longest. And there will probably be communities with different levels of technology. So maybe if you want to live in 1995 indefinitely, you could do that and there will just be some provision for people to get access to 2040s medical care. This isn't gonna be something where everyone's jobs are automated and we're all in breadlines begging some terminator for a weekly ration. No thanks 😁
|
|
|
Post by thehightenor on Jan 7, 2024 6:23:29 GMT -6
Hypothetically speaking, would you guys trade your audio career for infinite or near infinite resources, the ability to have any medical issue solved, the ability to become augmented, a cyborg, or something better, full immersion VR, and to experience things that would be more stimulating than anything we can imagine? .... .... what and give up my Retro STA Level .... no way - are you crazy!
|
|
|
Post by drbill on Jan 7, 2024 9:47:27 GMT -6
Hypothetically speaking, would you guys trade your audio career for infinite or near infinite resources, the ability to have any medical issue solved, the ability to become augmented, a cyborg, or something better, full immersion VR, and to experience things that would be more stimulating than anything we can imagine? That would be a hard no. I know the essence of what will come along with that.
|
|
|
Post by chessparov on Jan 7, 2024 11:54:43 GMT -6
Artificial Intelligence was developed because they gave up looking for genuine intelligence here otherwise. Great Thread. Thanks guys. Will ketchup tonight. If my Robotic Babysitter allows me. She's my Transistor Sister. Chrus
|
|
|
Post by ml on Jan 7, 2024 13:18:57 GMT -6
We will be fine as long as Ai doesn’t grow human ears.
|
|
|
Post by the other mark williams on Jan 7, 2024 15:31:01 GMT -6
We are nowhere close to being prepared for how quickly things may change. Someone earlier mentioned image generators getting the number of fingers wrong and that kind of thing. Dall-e 2 (image generator by the folks at ChatGPT) was released with a public API in September 2022. Wildlife photographers began experimenting with it to see how close it could get to a photorealistic image of a specific bird, for instance. It wasn't very good. I mean, the fact that it could do anything was interesting, but yes, too many toes or an extra claw, a head that didn't match the body, that kind of thing. You could look at it and tell immediately it was fake. Dall-e 3 was released in April 2023 (7 months later), and it was astonishing how much better it was at the same tasks. To the point that it could fool professional photographers who just didn't happen to be familiar with that particular bird. That is an insane amount of growth and refinement in just seven months: Fooling professional photographers into thinking an image was a photo of a specific bird, when it wasn't actually even a photo at all, much less a photo of that specific bird. The music composition AIs are not good enough to replace human composers right now. But you are badly underestimating this technology if you think AIs will NEVER be good enough to replace human composers. This is one of the primary reasons the Writers Guild and SAG-AFTRA went on strike and held out until they could get some protections for screenwriters and actors: the threat of being replaced by AI. As musicians and producers and recording engineers, we have never had as solid a union as those guys. We do not have the same protections. One (albeit small) thing we could pass legislation about is forcing AI companies to disclose where they get their material to train their AIs in their machine learning processes. If rock composer AIs are being trained by "Don't Fear the Reaper," I personally believe donr should be compensated somehow. The tech companies are enriching themselves further by using material that they did not write or create. They should have to pay some type of licensing fee that currently does not exist.
|
|
|
Post by notneeson on Jan 7, 2024 15:47:28 GMT -6
We are nowhere close to being prepared for how quickly things may change. Someone earlier mentioned image generators getting the number of fingers wrong and that kind of thing. Dall-e 2 (image generator by the folks at ChatGPT) was released with a public API in September 2022. Wildlife photographers began experimenting with it to see how close it could get to a photorealistic image of a specific bird, for instance. It wasn't very good. I mean, the fact that it could do anything was interesting, but yes, too many toes or an extra claw, a head that didn't match the body, that kind of thing. You could look at it and tell immediately it was fake. Dall-e 3 was released in April 2023 (7 months later), and it was astonishing how much better it was at the same tasks. To the point that it could fool professional photographers who just didn't happen to be familiar with that particular bird. That is an insane amount of growth and refinement in just seven months: Fooling professional photographers into thinking an image was a photo of a specific bird, when it wasn't actually even a photo at all, much less a photo of that specific bird. The music composition AIs are not good enough to replace human composers right now. But you are badly underestimating this technology if you think AIs will NEVER be good enough to replace human composers. This is one of the primary reasons the Writers Guild and SAG-AFTRA went on strike and held out until they could get some protections for screenwriters and actors: the threat of being replaced by AI. As musicians and producers and recording engineers, we have never had as solid a union as those guys. We do not have the same protections. One (albeit small) thing we could pass legislation about is forcing AI companies to disclose where they get their material to train their AIs in their machine learning processes. If rock composer AIs are being trained by "Don't Fear the Reaper," I personally believe donr should be compensated somehow. The tech companies are enriching themselves further by using material that they did not write or create. They should have to pay some type of licensing fee that currently does not exist. It will be interesting to see what happens with the NYT suit against Open AI, in regards to compensation for training sources.
|
|
|
Post by rowmat on Jan 7, 2024 16:10:43 GMT -6
We will be fine as long as Ai doesn’t grow human ears. It will learn to lip read.
|
|