I have also started using markers for differentiating what is and isn't AI generated in my visual art. I place them where I would otherwise place a signature.
I have developed somewhat of a hybrid AI workflow, and I think it's important (for me) to be transparent about it - in case I ever receive any degree of artistic recognition - because I don't want to spend my time lying about what I do.
That’s the flip side of what I’ve been thinking, which is that we need a federal law requiring some kind of digital watermark for AI produced material. Maybe your idea is more workable - some kind of non-governmental certification, like Good Housekeeping Seal.
It already has; the person you're replying to doesn't know what they are talking about. AI watermarking schemes work by introducing subtle statistical patterns into the generated text. The only way to remove it is to rewrite the text, which defeats the purpose of using generative AI in the first place.
Interesting set of comments here. Unfortunately, the market will win out. Just think of the Made in China (by slave labor) labelling requirements we currently have in place. People will go to WalMart and vote with their wallets. So mediocre AI created products will slowly dominate the market because people tend to purchase the cheapest thing they can to fill their perceived needs. I'm not sure how to put the genii back into the bottle...
I think this will also kill giant monopolistic bookstores and global market websites. We'll go back so smaller, boutique establishments, whether online or offline, where the product selection is curated by people who know the material. Like the old days.
The onus should be on the imitator. It’s a crime to live in an upside down society where artificial is normalized and what is whole, good, and true has to be labeled. 🙃
I've been thinking about this issue a lot, and I agree with your solution. Asking those who manage AI content to stamp themselves is ideal, but it relies on regulation, enforcement, and the actions of others.
A far easier solution that any human creator can implement themselves, right now, is to publicise and market their work as solely human-created (or whatever descriptor matches your process, if you did use some AI assistance).
Then watch the lawsuits begin as soon as any AI content tries to get marketed as human-created. I guarantee this will speed up the process of getting regulations and legal protection, rather than simply asking for regulation to be invented.
This is truly disturbing. Being robbed and impersonated in one fell swoop does indeed seem accurate. Everything that has troubled me about the numbers-based metrics dominating higher ed, and now publishing, finds its logical conclusion in this hellscape for artists and writers.
This world… I tell ya… man oh man… I enjoy your writings and love when you share thoughts and info about music, albums I’ve never heard of and in general, your thoughts on life. Keep it up… if you are who you say you are… LOL
I worked for Publix SuperMarkets for 35 years. My last 10 or so years we had a computer ordering system called AR… Automated Replenishment… I called it “Absolutely Ridiculous” to some of my superiors. :-)
It became a wonderful program, but there was a learning curve… :-)
I call AI… Absolutely Idiotic, Almost Insanity, Amazingly Infantile… or any other combinations I come up with. In time though, I assume it may be a part of my life. As of yet… I know nothing of it and have no use for it… maybe that makes me Awfully Ingenious… :-)
Thanks for what you do… if you’re actually doing it LOL :-)
Look at your Substack settings. There is a setting to "Block AI training." Which is scary enough as it is. See Settings-->Publications details:
"This setting indicates to AI tools like ChatGPT and Google Bard that their models should not be trained on your published content. This will only apply to AI tools which respect this setting, and blocking training may limit your publication's discoverability in tools and search engines that return AI-generated results."
I still don't see how AI is useful? Certainly not for artists of every kind. It only helps to make consumers and audiences lazy and stupid and devoid of thinking critically for themselves. Why come up with your own ideas when you can ask a machine to invent something for you??? I read University of Penn is the first to offer a degree in AI. So now you can graduate in pirating, plagiarizing encouraged. Maybe they'll include a course in AI forensics for when it murders our individual identities.
There's certainly uses for it. I know of a few projects in the medical space, communicating with animals, etc.
Problems are now arising because people abuse it to make money, to harm others, to see if they can get away with it etc. And of course they would. This risk should have been factored in and the release should have been done more responsibly. I recommend looking for videos put up by the center for humane technology on YouTube if curious
Ask AI to write you a book on the history of jazz (or anything else) and you'll get an utterly banal collection of cliches with perfect spelling, very good grammar and a semblance of professional writing style. There's no way to learn anything from it because all the facts and opinions will be either trivial or false. Its only useful for writing things that you can be sure nobody will read. It is a real problem that interesting and useful books are obscured under piles of fraudulent pseudothought at Amazon and elsewhere. I'm starting to think that AI may be incapable of producing anything of real value. I think a lot of investors are coming to the same conclusion.
Nonsense yes, because the one thing algorithmic processing of big data is unlikely ever to be capable of doing, with all its vast stores of coded and mathematically manipulated information streams, is to reflect on any of it in the context of human experience. Initial amazement at LLM-generated “product” derives largely from its uncanny resemblance to nonsense already generated in huge quantities by humans.
Sorry to hear this. I hope people have the sense not to buy it. I feel like the simplest solution is to make copyright material opt-in for machine learning - by law. The fair use argument seems bogus, it’s quite clear that a lot of AI - perhaps all - amounts to a sophisticated form of plagiarism. I think opt-in is where we are heading; I hope sooner rather than later to avoid a lot of damage along the way.
Yes, how is it not copyright infringement? But even if we can sue on that basis, who do we sue? And where's the money for the lawsuit coming from? Once again, those who lack creativity steal ours.
I wonder if something akin to global unions are coming. Pay a monthly fee and they cover you legally in certain countries. In addition, proper legislation, which I imagine is most likely to come from the EU, which by the end of this year might be the last overall democracy with a 'superpower' status. If the EU can hit these companies with massive fines, or even prosecute their workers with criminal charges, it may cease to be a practice except at the margins.
Imogene Heap started something called Creative Passport a few years ago, with the intent of artists being able to keep all their references in one place without having to upload to multiple platforms. I haven't checked on it lately but maybe something like that could be in charge of deploying your idea.
Thanks for highlighting this. As you say: “There ought to be a law to stop this.” Yes, and there ought to a broad legal framework for the advanced technology that thrives on theft and exploitation of everyone’s privacy and intellectual property, but sadly no such thing has been developed. It is past time to address these issues. A functioning Congress would help a lot. Register. And vote.
I think this is actually one of the most useful functions of democracy: It is VERY hard to change something, at the same time as you provide the illusion that you are making a difference.
Which is FANTASTIC in a functioning society. Unfortunately, society is becoming increasingly in NEED of change, so...
This is why many people are demanding "strong leadership to drain the swamp". Unfortunately most of these people have no real understanding of what strength is, and confuses it with bravado, overconfidence, wealth and self-proclaimed success.
They also don't know that draining swamps releases a lot of gases, which eventually becomes problematic down the line :P
It’s never pretty, but bare functionality is a low bar. Very occasionally even Congress can accomplish things when the people insist. The best politicians lead but most follow, and I agree, it’s money mostly they follow. Still, it’s better than tyranny or anarchy.
Senators Amy Klobuchar and Bill Cassidy (others too) have tried to shine a light on the need for a better regulatory framework, but we are way “behind the curve” in the US. Too many real issues government should address are drowning in all the noisy nonsense.
I have no doubt that Congress, as a whole, is incapable of understanding the underlying issues of this problem and will take their cue from Amazon and OpenAI lobbyists.
Maybe, but somehow the whole society must deal with technology like this, which is neither perfect nor omnipotent - it won’t even meet all the claims hawked for it. Eventually “gravity,” which never rests, will assert itself in ways that even smart people must face.
This and related issues are getting worse quickly. In the last few days I've seen the following: Facebook posting direct competitors ads on a friend's business page. Google promoting product "reviews" from major media sites that aren't actual reviews over those from legitimate review sites, driving down their business and promoting misinformation. To use Cory Doctorow's term, the enshitification of the Internet is happening at an ever-increasing rate.
Prime tiny example... Auto correct has altered my its to it's as I fast typed. Scale that shit up and ... But it will get better at what it does. We need strategies and policies.
I like where you're going with this. We need to find ways to outsmart the AI purveyors at their own game. Robert Greene, in his book The 48 Laws of Power, gives some insight into this process with anecdotes from real life historical situations that may be considered similar to what we're now dealing with.
We label movies, TV shows, tobacco, and foods already, with differing results. Producers avoid an X rating for their movies, since it's box office poison. For foods: they already tell you how much sugar and fat they contain, although that hasn't proven very effective so far.
If Amazon, Barnes & Noble, and Goodreads were forced to put "AN AI-GENERATED BOOK" at the very top of every book description, that would be a good start, at least.
The text irresistibly reminds me of Schickele's parody biography of PDQ Bach, satirizing the writing style of academic music critics. "Unlike the many lesser-known but nevertheless competent composers who dotted the musical landscape of the Age Of Enlightenment..." Schickele was doing the same thing AI does now, but he made it funny.
Sorry to hear about this Mr Gioia. I think it has as much to do with the state of the culture as well as AI ! I hope you can get this sorted, because your writing and point of view are an invaluable contribution to music, culture, and life in general! I for one, am rooting for you! Steady on sir!
I’m convinced we will one day have a Certified Human stamp on creative work. The uprising is just beginning.
I suspect something similar. We have "hand made" labels. A "Brain made" label or "human made" is becoming more and more needed.
I've already started doing that with my writing.
I have also started using markers for differentiating what is and isn't AI generated in my visual art. I place them where I would otherwise place a signature.
I have developed somewhat of a hybrid AI workflow, and I think it's important (for me) to be transparent about it - in case I ever receive any degree of artistic recognition - because I don't want to spend my time lying about what I do.
Further details here:
https://backtobasic.substack.com/p/sketch-1-cosmic-horror
I particularly favor the aversion to lying. A bias for truth, in art and life, is a powerful force, I think.
That’s the flip side of what I’ve been thinking, which is that we need a federal law requiring some kind of digital watermark for AI produced material. Maybe your idea is more workable - some kind of non-governmental certification, like Good Housekeeping Seal.
Any watermark can be removed or duplicated with the technology available today.
Well, if there’s a need maybe that will change.
It already has; the person you're replying to doesn't know what they are talking about. AI watermarking schemes work by introducing subtle statistical patterns into the generated text. The only way to remove it is to rewrite the text, which defeats the purpose of using generative AI in the first place.
Interesting set of comments here. Unfortunately, the market will win out. Just think of the Made in China (by slave labor) labelling requirements we currently have in place. People will go to WalMart and vote with their wallets. So mediocre AI created products will slowly dominate the market because people tend to purchase the cheapest thing they can to fill their perceived needs. I'm not sure how to put the genii back into the bottle...
How about. 'This product is 99% AI free' or 'This product may contain traces of AI'
I created one for my music videos
Absolutely. Human made will probably come to be niche market but will be able to charge more. A bit like organic V non-organic.
Wow ! Amazing but yes - it has come to that. WHat you say will totally come to pass
I think this will also kill giant monopolistic bookstores and global market websites. We'll go back so smaller, boutique establishments, whether online or offline, where the product selection is curated by people who know the material. Like the old days.
The onus should be on the imitator. It’s a crime to live in an upside down society where artificial is normalized and what is whole, good, and true has to be labeled. 🙃
I've been thinking about this issue a lot, and I agree with your solution. Asking those who manage AI content to stamp themselves is ideal, but it relies on regulation, enforcement, and the actions of others.
A far easier solution that any human creator can implement themselves, right now, is to publicise and market their work as solely human-created (or whatever descriptor matches your process, if you did use some AI assistance).
Then watch the lawsuits begin as soon as any AI content tries to get marketed as human-created. I guarantee this will speed up the process of getting regulations and legal protection, rather than simply asking for regulation to be invented.
This is truly disturbing. Being robbed and impersonated in one fell swoop does indeed seem accurate. Everything that has troubled me about the numbers-based metrics dominating higher ed, and now publishing, finds its logical conclusion in this hellscape for artists and writers.
That's a shame that book was pulled - I was just about to order a copy!
As for Luke Ellington, he took the B train.
And the famous Charles Mingus once wore a mince pie hat - I read that somewhere...
Thank you for the laugh. Much needed after reading this.
This world… I tell ya… man oh man… I enjoy your writings and love when you share thoughts and info about music, albums I’ve never heard of and in general, your thoughts on life. Keep it up… if you are who you say you are… LOL
I worked for Publix SuperMarkets for 35 years. My last 10 or so years we had a computer ordering system called AR… Automated Replenishment… I called it “Absolutely Ridiculous” to some of my superiors. :-)
It became a wonderful program, but there was a learning curve… :-)
I call AI… Absolutely Idiotic, Almost Insanity, Amazingly Infantile… or any other combinations I come up with. In time though, I assume it may be a part of my life. As of yet… I know nothing of it and have no use for it… maybe that makes me Awfully Ingenious… :-)
Thanks for what you do… if you’re actually doing it LOL :-)
Look at your Substack settings. There is a setting to "Block AI training." Which is scary enough as it is. See Settings-->Publications details:
"This setting indicates to AI tools like ChatGPT and Google Bard that their models should not be trained on your published content. This will only apply to AI tools which respect this setting, and blocking training may limit your publication's discoverability in tools and search engines that return AI-generated results."
That is cooked. "training"
Holy crap, really?
I still don't see how AI is useful? Certainly not for artists of every kind. It only helps to make consumers and audiences lazy and stupid and devoid of thinking critically for themselves. Why come up with your own ideas when you can ask a machine to invent something for you??? I read University of Penn is the first to offer a degree in AI. So now you can graduate in pirating, plagiarizing encouraged. Maybe they'll include a course in AI forensics for when it murders our individual identities.
There's certainly uses for it. I know of a few projects in the medical space, communicating with animals, etc.
Problems are now arising because people abuse it to make money, to harm others, to see if they can get away with it etc. And of course they would. This risk should have been factored in and the release should have been done more responsibly. I recommend looking for videos put up by the center for humane technology on YouTube if curious
Ask AI to write you a book on the history of jazz (or anything else) and you'll get an utterly banal collection of cliches with perfect spelling, very good grammar and a semblance of professional writing style. There's no way to learn anything from it because all the facts and opinions will be either trivial or false. Its only useful for writing things that you can be sure nobody will read. It is a real problem that interesting and useful books are obscured under piles of fraudulent pseudothought at Amazon and elsewhere. I'm starting to think that AI may be incapable of producing anything of real value. I think a lot of investors are coming to the same conclusion.
Your assessment will be vindicated further, I suspect, as AI-LLM-generated “writing” increasingly regurgitates its own vomit.
A vast feedback loop of nonsense.
Nonsense yes, because the one thing algorithmic processing of big data is unlikely ever to be capable of doing, with all its vast stores of coded and mathematically manipulated information streams, is to reflect on any of it in the context of human experience. Initial amazement at LLM-generated “product” derives largely from its uncanny resemblance to nonsense already generated in huge quantities by humans.
Seems to me that AI is just a tool. It doesn’t create things, but it does rearrange things in a sometimes useful way.
https://www.cam.ac.uk/research/news/scientists-begin-building-ai-for-scientific-discovery-using-tech-behind-chatgpt. It'll be useful to do certain things. Again, we need to remember the revolts when the printing press was developed. Also the opposition to calculators. We're in a similar moment.
Sorry to hear this. I hope people have the sense not to buy it. I feel like the simplest solution is to make copyright material opt-in for machine learning - by law. The fair use argument seems bogus, it’s quite clear that a lot of AI - perhaps all - amounts to a sophisticated form of plagiarism. I think opt-in is where we are heading; I hope sooner rather than later to avoid a lot of damage along the way.
Yes, how is it not copyright infringement? But even if we can sue on that basis, who do we sue? And where's the money for the lawsuit coming from? Once again, those who lack creativity steal ours.
I wonder if something akin to global unions are coming. Pay a monthly fee and they cover you legally in certain countries. In addition, proper legislation, which I imagine is most likely to come from the EU, which by the end of this year might be the last overall democracy with a 'superpower' status. If the EU can hit these companies with massive fines, or even prosecute their workers with criminal charges, it may cease to be a practice except at the margins.
Imogene Heap started something called Creative Passport a few years ago, with the intent of artists being able to keep all their references in one place without having to upload to multiple platforms. I haven't checked on it lately but maybe something like that could be in charge of deploying your idea.
Surprised? Baylor Swift has a new album about to drop.
Can we plug the leaks? A digital rights law with serious teeth has been needed since the 80’s, or at least that was my rant then.
The revolution is in billing. Make em pay.
Also: Brian Roemmele suggests personal AI protection against this. Promising? Perhaps.
Is there some precedent for getting scammers like the ones Ted writes about to pay?
Thanks for highlighting this. As you say: “There ought to be a law to stop this.” Yes, and there ought to a broad legal framework for the advanced technology that thrives on theft and exploitation of everyone’s privacy and intellectual property, but sadly no such thing has been developed. It is past time to address these issues. A functioning Congress would help a lot. Register. And vote.
The legal system is not going to do anything substantial to protect copyright holders. It is up to Congress.
And guess what? Congress is owned lock, stock and barrel by the tech oligarchy!
Bend over, here it comes again
OR, change the Congress. It is not impossible, just difficult.
In election year it's a good time to hold their feet to the fire.
That will last until the flame dies down, and it always dies down.
You can vote till all the notes fall off the page and there still won't be a functioning congress, unless they are bribed to function.
I think this is actually one of the most useful functions of democracy: It is VERY hard to change something, at the same time as you provide the illusion that you are making a difference.
Which is FANTASTIC in a functioning society. Unfortunately, society is becoming increasingly in NEED of change, so...
This is why many people are demanding "strong leadership to drain the swamp". Unfortunately most of these people have no real understanding of what strength is, and confuses it with bravado, overconfidence, wealth and self-proclaimed success.
They also don't know that draining swamps releases a lot of gases, which eventually becomes problematic down the line :P
I agree. Well said. And “strongmen” almost always are in it for themselves exclusively, as are the greedy sycophants who serve them.
You nailed it. I have no confidence in govt., any govt., anywhere.
It’s never pretty, but bare functionality is a low bar. Very occasionally even Congress can accomplish things when the people insist. The best politicians lead but most follow, and I agree, it’s money mostly they follow. Still, it’s better than tyranny or anarchy.
The EU is doing surprisingly good work on this. I think it's closer than we might think
Senators Amy Klobuchar and Bill Cassidy (others too) have tried to shine a light on the need for a better regulatory framework, but we are way “behind the curve” in the US. Too many real issues government should address are drowning in all the noisy nonsense.
I have no doubt that Congress, as a whole, is incapable of understanding the underlying issues of this problem and will take their cue from Amazon and OpenAI lobbyists.
Maybe, but somehow the whole society must deal with technology like this, which is neither perfect nor omnipotent - it won’t even meet all the claims hawked for it. Eventually “gravity,” which never rests, will assert itself in ways that even smart people must face.
It’s awful too for readers to have to wade through fake listings to get to real writing.
This and related issues are getting worse quickly. In the last few days I've seen the following: Facebook posting direct competitors ads on a friend's business page. Google promoting product "reviews" from major media sites that aren't actual reviews over those from legitimate review sites, driving down their business and promoting misinformation. To use Cory Doctorow's term, the enshitification of the Internet is happening at an ever-increasing rate.
I put my name into an internet search site and found that I had a facebook page. I haven't had a facebook page in 10 yrs.
I asked chatgpt for my bio and it got both BA and PhD alma maters wrong. And spewed repetitive nonsense about the content of my academic work.
Prime tiny example... Auto correct has altered my its to it's as I fast typed. Scale that shit up and ... But it will get better at what it does. We need strategies and policies.
If ever, I read a good argument to return to ink on paper, this is one.
Does it help to know that creative works generated by AI cannot be copyrighted? Steal that book and resell under a different name.
I like where you're going with this. We need to find ways to outsmart the AI purveyors at their own game. Robert Greene, in his book The 48 Laws of Power, gives some insight into this process with anecdotes from real life historical situations that may be considered similar to what we're now dealing with.
Mandatory labelling is a very promising approach, and I support it.I independently came up with the same approach to kids' addiction to social media:
https://albertcory50.substack.com/p/enforcing-a-social-app-ban-for-kids
We label movies, TV shows, tobacco, and foods already, with differing results. Producers avoid an X rating for their movies, since it's box office poison. For foods: they already tell you how much sugar and fat they contain, although that hasn't proven very effective so far.
If Amazon, Barnes & Noble, and Goodreads were forced to put "AN AI-GENERATED BOOK" at the very top of every book description, that would be a good start, at least.
The text irresistibly reminds me of Schickele's parody biography of PDQ Bach, satirizing the writing style of academic music critics. "Unlike the many lesser-known but nevertheless competent composers who dotted the musical landscape of the Age Of Enlightenment..." Schickele was doing the same thing AI does now, but he made it funny.
And made it obvious that he was parodying Bach.
Sorry to hear about this Mr Gioia. I think it has as much to do with the state of the culture as well as AI ! I hope you can get this sorted, because your writing and point of view are an invaluable contribution to music, culture, and life in general! I for one, am rooting for you! Steady on sir!