Sure, a bad imitation is distinguishable. But a good one?
A really good simulacrum of a highly formulaic or stylistically mannered performance could be convincing. Because the "real thing" is already sort of artificial.
wj - ...actors can play parts, with authentic appearing emotions, even about experiences they have never personally had — all it takes is having seen someone else experiencing it.
Agreed, but look at what I said: What you do need, however, is some life experience to connect it with. Note that I did not say that they need to have that precise experience, just enough to act as a bridge between their own experience and others'. As russell says, it takes empathy, or as they used to say "fellow feelings." An AI has no experiences, and isn't a person, so can have no personal perspective and cannot reflect. It has to be trained to extrapolate within very narrow ranges and cannot imagine or improvise or project. Even a sociopath has a better perspective for understanding. At least the sociopath is embodied and sensate and conscious. An AI is a database with a good costuming department.
What is required is empathy. Which machines do not have. They can imitate. They cannot empathize. Those are different things.
Certainly they are different. The question is, are they distinguishable? I'm not sure that they necessarily are? Sure, a bad imitation is distinguishable. But a good one?
Put another way, is real empathy required? Or can it be simulated convincingly?
What you do need, however, is some life experience to connect it with.
I’m not so sure about that. Certainly it can help. But actors can play parts, with authentic appearing emotions, even about experiences they have never personally had
What is required is empathy. Which machines do not have.
They can imitate. They cannot empathize. Those are different things.
And I don’t think that one has to have written a song in order to understand and serve the emotions of the song. What you do need, however, is some life experience to connect it with.
I'm not so sure about that. Certainly it can help. But actors can play parts, with authentic appearing emotions, even about experiences they have never personally had -- all it takes is having seen someone else experiencing it. Or showing how it looked when a third party did. Great actors do it most convincingly, but even journeyman level actors can do a pretty convincing job.
Are singers any different from actors in that regard? I'm willing to be convinced, but it may take some doing.
And I don’t think that one has to have written a song in order to understand and serve the emotions of the song. What you do need, however, is some life experience to connect it with.
Perfect example: Johnny Cash singing Hurt. The video's pretty amazing too.
And I don't think that one has to have written a song in order to understand and serve the emotions of the song. What you do need, however, is some life experience to connect it with.
And I'm not talking about a CGI/animated puppet for real performers (a la Gorillaz or Dethklok). That's just human musicians cosplaying something else. What Spotify, Sony, Warner, etc. are after is on-trend content generated by trained expert systems in response to prompts or to the other content listened to by users of their services.
lj - ...and I would be surprised if this week’s is on anyone else’s.
Incorrect, sir.
The Hu - Yuve Yuve Yu (Mongolian Folk Metal)
https://www.youtube.com/watch?v=v4xZUr0BEfE
Heilung - In Maidjan (Danish Shamanistic Neofolk)
https://www.youtube.com/watch?v=kmWTZ3KfnXE
Both of which have, with other songs from these bands, had some heavy rotation in my playlists. Both bands are pretty awesome, and have dedicated fanbases. I find that Heilung live video astounding and powerful every damn time I watch it.
"Throat singing" is also practiced in Alaskan native societies. Got a CD of it, somewhere.
Takes some getting used to, but I always thought that it would be a musical style that could fit into a rap/hip-hop work, if someone was sufficiently motivated to try.
Shorter me - show me the AI music generator that will come up with a line like "Looks a lot like Che Guevara / he drove a diesel van".
It's simultaneously tongue-in-cheek hip and hilarious and ironic and allusive in about 10 different directions. Totally obvious and common place chord changes, but the snarkiest lyric ever.
I don't think AI is capable of that. In fact, I'm curious to know if AI can make a good joke, at all.
the Monkees were a made-for-TV group but the songs were written by some of the best pop song writers of the day, and the music was performed by real live A list studio cats.
plus, at least one of the guys (Mike Nesmith) was actually a competent musician and songwriter.
net/net, not at all like AI generated music.
A lot of musical styles, especially commercial pop styles, are highly formulaic, so it wouldn't be that hard to have AI crank it out.
And what you would get would be highly formulaic pop music. Which a lot of people really like, and would be a perfectly fine commodity and lifestyle accessory. It may sound like I'm being dismissive when I say that, but I'm not - that is what a lot of music is made for, and how a lot of music is used.
It's like the art prints at your doctor's office waiting room. They aren't Rembrandt, or even Andy Warhol. But they are pleasant to look at, and don't clash with the color scheme.
Again, not being dismissive. It's nice to have pleasant, undemanding stuff to look at (and even ignore) when you're waiting for an appointment.
What you will not get from AI is a Leonard Cohen, or a Tom Waits, or a David Bowie, or a Paul Simon. To cite some better-known examples. You might get a Beatles of the quality of "I Want To Hold Your Hand", but not an "Eleanor Rigby".
If you fed an AI music generator a diet of any or all of those guys, you might get a simulacrum of their work. But it will be missing the special ingredient that actually makes you sit up and take notice when you hear their stuff - the human insight, the unusual chord change, the frisson that comes from the unexpected use of language in the lyric.
AI is inherently derivative. Derivative work can be useful, and has its place, but it isn't going to tell you anything you don't already know.
What makes the folks I named here artists, rather than simply entertainers, is the way in which they subvert the stylistic formulas they work in to discover meaning beneath the surface of the style.
Maybe someday some kind of AGI gizmo will be capable of that. If so, it probably will not be in a direction that resonates with humans.
But I am skeptical that AGI will ever actually be a thing.
My next thought was that lots (most?) pop stars are performers, and their songs are generally written by someone else.
Linda Ronstadt always maintained that she had no talent as a songwriter. Lots of very good songwriters wrote material with her specifically in mind, though.
wj - you can have AI generate the name of your star and generate genre appropriate "photos" of them. You can then use Claude to generate all the lyrics for your song from a simple, one sentence prompt. You can then feed those lyrics into Suno and have it generate a genre-appropriate song based on the lyrics complete with a vocalist.
It's all just a stew of algorithmically generated near-plagiarism.
Can an AI generated pop star understand your broken heart?
I read that, and my first thought was The Monkeys. A totally made-up-for-television group. In other words, about as authentic as an AI generated pop star.
My next thought was that lots (most?) pop stars are performers, and their songs are generally written by someone else.** If one person writes the music, another person writes the lyrics, and a third performs the song? Which, if any, have to understand your broken heart?
** There are exceptions. People who write and perform their own stuff, at least mostly. But they are just that: exceptions.
I'm prepping and making changes to my syllabus for the writing class I teach that centers around college music - writing reviews of "college radio" albums from the '80s and '90s, and then taking what they have learned from doing that to comment on the music that is a part of their college lives today.
I had been having them write an essay exploring the question "Does College Music Still Exist?," and digging into the social side of what defines the music of that moment and that community. Sadly, I've never been satisfied with the depth of their engagement with the topic and have been wanting to change it up for a few terms now in the hopes of finding something that gets them thinking more deeply and feeling like they have something that the want to say.
My current idea is to get them thinking more deeply about the ecology of their music media. If the music you listen to is chosen for you by an algorithm, what makes it yours? Should Spotify allow the uploading and monetizing of music produced by AI? Can an AI generated pop star understand your broken heart? Do you really listen to songs you don't buy and music you don't own, or is it just something to consume like fast food?
I keep making stabs at how to turn all that into a philosophical question that can provoke reflection and inspire many different responses - something like "What's Wrong With Listening To Spotify?" or the like, but I haven't found one with the right mojo and moxie. Any ideas?
In reference to the Georgia PSC election, I commented on the previous thread about it, but a little more detail: The 5 PSC seats are titled District 1-5, but the elections are statewide. There have not been PSC elections since 2020 due to lawsuit(s) claiming that the statewide elections are discriminatory, as a result the two seats the Democrats just won will be up for election in one and five years, respectively, the normal term is 6 years. The last Democrat to win election to the PSC left office at end of term in 2007, and no Democrat has won a statewide election to any state office in over 20 years.
So to sort of answer Michael, while there are definite local issues regarding rate hikes, this is a big effing deal, especially given the 63-37 margins. Since the state turnover to Republican control in the aughts, it has been the general presumption that Democrats best shot at winning a statewide race is in presidential election years (which puts governor and other executive offices out of reach since they are scheduled in off years). So winning not just an off-year but odd-year election, where there were no other statewide offices on the ballot to drive turnout (and some/many jurisdictions may not have had any local races; locally we had unusually low profile mayor's race, Atlanta City Council President race and several other Council races, as well as judicial retention elections and one ballot initiative).
There were no personality/scandal issues weighing down the Republican office holders, it's understood that any Republican on that body will be in Southern Company's pocket, the only question is how deep.
The exit polling is broken down in what may be too many different ways at the link. The long and short of it is that being white, older, male, and less educated made someone more likely to have voted for Ciattarelli.
One interesting point is that they asked people if they were concerned about political violence. Not enough people weren't concerned to be statistically meaningful, so the vote breakdown on the "No" side was "N/A."
There's a drop-down where you can get exit polling for the elections in other states.
The thing is, Trump is an asshole and he's making a lot of people's lives more difficult than they need to be. And (R)'s are basically in thrall to the guy.
Last November, enough people were pissed at Biden for any of a variety of reasons, some of them legitimate, and so were willing to give Trump another shot at it.
And now they see Trump without the moderating influences of the people who more or less kept him in the ballpark of legitimate governance.
And they think it sucks.
All of that, plus it's not uncommon for off year and mid-term elections to favor the party not in power, with exceptions for wartime and cases where the party in power is actually nailing it.
On a lighter note, while thinking of examples of rightwingers only showing compassion to suffering encountered in their own circles (remembering that Dick Cheney's support for gay marriage was undoubtedly to do with having a lesbian daughter), I was reading various pieces about DC and smiled to see this:
Former US vice-president Dick Cheney, who has died, had intimidating power. For instance, when Cheney shot a friend while hunting, an apology was made by the friend to Cheney. His fearful aura made it all the more amusing when CNN accidentally published an obituary of Cheney in 2003, but it was unfinished and had been based on a template used for the Queen Mother. Cheney was described as “the UK’s favourite grandmother”.
The following is according to exit polling for the NJ gubernatorial election. In parentheses are the percentages of the overall votes for each group, followed by the percent of the group that voted for Sherrill, followed by the percent for Ciattarelli.
White (70%) 47% 52%
Black (10%) 94% 5%
Hispanic/Latino (10%) 68% 31%
Asian (5%) 82% 17%
Other (4%) 54% 43%
As mentioned at GftNC's NYT link, groups who shifted right in the last presidential election have moved back left after seeing what they really voted for. I'm curious how age groups voted we well and will probably dig something up later.
I mean, 94% of the Black vote. Wow!
*Comment archive for non-registered commenters assembled by email address as provided.
On “Weekend Music Thread #04 John Mackey”
Sure, a bad imitation is distinguishable. But a good one?
A really good simulacrum of a highly formulaic or stylistically mannered performance could be convincing. Because the "real thing" is already sort of artificial.
Beyond that, I don't think so.
"
wj - ...actors can play parts, with authentic appearing emotions, even about experiences they have never personally had — all it takes is having seen someone else experiencing it.
Agreed, but look at what I said: What you do need, however, is some life experience to connect it with. Note that I did not say that they need to have that precise experience, just enough to act as a bridge between their own experience and others'. As russell says, it takes empathy, or as they used to say "fellow feelings." An AI has no experiences, and isn't a person, so can have no personal perspective and cannot reflect. It has to be trained to extrapolate within very narrow ranges and cannot imagine or improvise or project. Even a sociopath has a better perspective for understanding. At least the sociopath is embodied and sensate and conscious. An AI is a database with a good costuming department.
"
What is required is empathy. Which machines do not have.
They can imitate. They cannot empathize. Those are different things.
Certainly they are different. The question is, are they distinguishable? I'm not sure that they necessarily are? Sure, a bad imitation is distinguishable. But a good one?
Put another way, is real empathy required? Or can it be simulated convincingly?
"
What you do need, however, is some life experience to connect it with.
I’m not so sure about that. Certainly it can help. But actors can play parts, with authentic appearing emotions, even about experiences they have never personally had
What is required is empathy. Which machines do not have.
They can imitate. They cannot empathize. Those are different things.
"
And I don’t think that one has to have written a song in order to understand and serve the emotions of the song. What you do need, however, is some life experience to connect it with.
I'm not so sure about that. Certainly it can help. But actors can play parts, with authentic appearing emotions, even about experiences they have never personally had -- all it takes is having seen someone else experiencing it. Or showing how it looked when a third party did. Great actors do it most convincingly, but even journeyman level actors can do a pretty convincing job.
Are singers any different from actors in that regard? I'm willing to be convinced, but it may take some doing.
On “Weekend music thread #05 Tuvan throat singing”
nous, I should have known...
On “Weekend Music Thread #04 John Mackey”
And I don’t think that one has to have written a song in order to understand and serve the emotions of the song. What you do need, however, is some life experience to connect it with.
Perfect example: Johnny Cash singing Hurt. The video's pretty amazing too.
"
russell is well on his way with this prompt.
And I don't think that one has to have written a song in order to understand and serve the emotions of the song. What you do need, however, is some life experience to connect it with.
And I'm not talking about a CGI/animated puppet for real performers (a la Gorillaz or Dethklok). That's just human musicians cosplaying something else. What Spotify, Sony, Warner, etc. are after is on-trend content generated by trained expert systems in response to prompts or to the other content listened to by users of their services.
On “Weekend music thread #05 Tuvan throat singing”
lj - ...and I would be surprised if this week’s is on anyone else’s.
Incorrect, sir.
The Hu - Yuve Yuve Yu (Mongolian Folk Metal)
https://www.youtube.com/watch?v=v4xZUr0BEfE
Heilung - In Maidjan (Danish Shamanistic Neofolk)
https://www.youtube.com/watch?v=kmWTZ3KfnXE
Both of which have, with other songs from these bands, had some heavy rotation in my playlists. Both bands are pretty awesome, and have dedicated fanbases. I find that Heilung live video astounding and powerful every damn time I watch it.
"
"Throat singing" is also practiced in Alaskan native societies. Got a CD of it, somewhere.
Takes some getting used to, but I always thought that it would be a musical style that could fit into a rap/hip-hop work, if someone was sufficiently motivated to try.
On “Weekend Music Thread #04 John Mackey”
Shorter me - show me the AI music generator that will come up with a line like "Looks a lot like Che Guevara / he drove a diesel van".
It's simultaneously tongue-in-cheek hip and hilarious and ironic and allusive in about 10 different directions. Totally obvious and common place chord changes, but the snarkiest lyric ever.
I don't think AI is capable of that. In fact, I'm curious to know if AI can make a good joke, at all.
"
the Monkeys
the Monkees were a made-for-TV group but the songs were written by some of the best pop song writers of the day, and the music was performed by real live A list studio cats.
plus, at least one of the guys (Mike Nesmith) was actually a competent musician and songwriter.
net/net, not at all like AI generated music.
A lot of musical styles, especially commercial pop styles, are highly formulaic, so it wouldn't be that hard to have AI crank it out.
And what you would get would be highly formulaic pop music. Which a lot of people really like, and would be a perfectly fine commodity and lifestyle accessory. It may sound like I'm being dismissive when I say that, but I'm not - that is what a lot of music is made for, and how a lot of music is used.
It's like the art prints at your doctor's office waiting room. They aren't Rembrandt, or even Andy Warhol. But they are pleasant to look at, and don't clash with the color scheme.
Again, not being dismissive. It's nice to have pleasant, undemanding stuff to look at (and even ignore) when you're waiting for an appointment.
What you will not get from AI is a Leonard Cohen, or a Tom Waits, or a David Bowie, or a Paul Simon. To cite some better-known examples. You might get a Beatles of the quality of "I Want To Hold Your Hand", but not an "Eleanor Rigby".
If you fed an AI music generator a diet of any or all of those guys, you might get a simulacrum of their work. But it will be missing the special ingredient that actually makes you sit up and take notice when you hear their stuff - the human insight, the unusual chord change, the frisson that comes from the unexpected use of language in the lyric.
AI is inherently derivative. Derivative work can be useful, and has its place, but it isn't going to tell you anything you don't already know.
What makes the folks I named here artists, rather than simply entertainers, is the way in which they subvert the stylistic formulas they work in to discover meaning beneath the surface of the style.
Maybe someday some kind of AGI gizmo will be capable of that. If so, it probably will not be in a direction that resonates with humans.
But I am skeptical that AGI will ever actually be a thing.
"
If either of you want to brainstorm a thesis, I've got office hours on Zoom.
"
My next thought was that lots (most?) pop stars are performers, and their songs are generally written by someone else.
Linda Ronstadt always maintained that she had no talent as a songwriter. Lots of very good songwriters wrote material with her specifically in mind, though.
On “Still I Rise”
Slight exaggeration, last GA Democrat state office winner was 2006 election.
On “Weekend Music Thread #04 John Mackey”
wj - you can have AI generate the name of your star and generate genre appropriate "photos" of them. You can then use Claude to generate all the lyrics for your song from a simple, one sentence prompt. You can then feed those lyrics into Suno and have it generate a genre-appropriate song based on the lyrics complete with a vocalist.
It's all just a stew of algorithmically generated near-plagiarism.
You can watch Rick Beato do just that here.
https://www.youtube.com/watch?v=eKxNGFjyRv0
"
Can an AI generated pop star understand your broken heart?
I read that, and my first thought was The Monkeys. A totally made-up-for-television group. In other words, about as authentic as an AI generated pop star.
My next thought was that lots (most?) pop stars are performers, and their songs are generally written by someone else.** If one person writes the music, another person writes the lyrics, and a third performs the song? Which, if any, have to understand your broken heart?
** There are exceptions. People who write and perform their own stuff, at least mostly. But they are just that: exceptions.
"
I'm prepping and making changes to my syllabus for the writing class I teach that centers around college music - writing reviews of "college radio" albums from the '80s and '90s, and then taking what they have learned from doing that to comment on the music that is a part of their college lives today.
I had been having them write an essay exploring the question "Does College Music Still Exist?," and digging into the social side of what defines the music of that moment and that community. Sadly, I've never been satisfied with the depth of their engagement with the topic and have been wanting to change it up for a few terms now in the hopes of finding something that gets them thinking more deeply and feeling like they have something that the want to say.
My current idea is to get them thinking more deeply about the ecology of their music media. If the music you listen to is chosen for you by an algorithm, what makes it yours? Should Spotify allow the uploading and monetizing of music produced by AI? Can an AI generated pop star understand your broken heart? Do you really listen to songs you don't buy and music you don't own, or is it just something to consume like fast food?
I keep making stabs at how to turn all that into a philosophical question that can provoke reflection and inspire many different responses - something like "What's Wrong With Listening To Spotify?" or the like, but I haven't found one with the right mojo and moxie. Any ideas?
On “Still I Rise”
In reference to the Georgia PSC election, I commented on the previous thread about it, but a little more detail: The 5 PSC seats are titled District 1-5, but the elections are statewide. There have not been PSC elections since 2020 due to lawsuit(s) claiming that the statewide elections are discriminatory, as a result the two seats the Democrats just won will be up for election in one and five years, respectively, the normal term is 6 years. The last Democrat to win election to the PSC left office at end of term in 2007, and no Democrat has won a statewide election to any state office in over 20 years.
So to sort of answer Michael, while there are definite local issues regarding rate hikes, this is a big effing deal, especially given the 63-37 margins. Since the state turnover to Republican control in the aughts, it has been the general presumption that Democrats best shot at winning a statewide race is in presidential election years (which puts governor and other executive offices out of reach since they are scheduled in off years). So winning not just an off-year but odd-year election, where there were no other statewide offices on the ballot to drive turnout (and some/many jurisdictions may not have had any local races; locally we had unusually low profile mayor's race, Atlanta City Council President race and several other Council races, as well as judicial retention elections and one ballot initiative).
There were no personality/scandal issues weighing down the Republican office holders, it's understood that any Republican on that body will be in Southern Company's pocket, the only question is how deep.
"
From here:
https://www.cnn.com/election/2025/exit-polls/new-jersey/general/governor/0
18-44 31% 45 or older 69%
Sherrill 67% 51%
Ciattarelli 32% 48%
The exit polling is broken down in what may be too many different ways at the link. The long and short of it is that being white, older, male, and less educated made someone more likely to have voted for Ciattarelli.
One interesting point is that they asked people if they were concerned about political violence. Not enough people weren't concerned to be statistically meaningful, so the vote breakdown on the "No" side was "N/A."
There's a drop-down where you can get exit polling for the elections in other states.
"
In regards to this
https://www.theguardian.com/us-news/2025/nov/05/zohran-mamdani-transition-team
I'm wondering if Mamdani said 'I've got an all female transition team' or if this is something a Graniuad writer noticed and made it the lede.
"
The thing is, Trump is an asshole and he's making a lot of people's lives more difficult than they need to be. And (R)'s are basically in thrall to the guy.
Last November, enough people were pissed at Biden for any of a variety of reasons, some of them legitimate, and so were willing to give Trump another shot at it.
And now they see Trump without the moderating influences of the people who more or less kept him in the ballpark of legitimate governance.
And they think it sucks.
All of that, plus it's not uncommon for off year and mid-term elections to favor the party not in power, with exceptions for wartime and cases where the party in power is actually nailing it.
That's my analysis, anyway.
On “People and poliltics”
I've been trying to be the UK's favourite grandmother for a long time. No dice.
"
On a lighter note, while thinking of examples of rightwingers only showing compassion to suffering encountered in their own circles (remembering that Dick Cheney's support for gay marriage was undoubtedly to do with having a lesbian daughter), I was reading various pieces about DC and smiled to see this:
Former US vice-president Dick Cheney, who has died, had intimidating power. For instance, when Cheney shot a friend while hunting, an apology was made by the friend to Cheney. His fearful aura made it all the more amusing when CNN accidentally published an obituary of Cheney in 2003, but it was unfinished and had been based on a template used for the Queen Mother. Cheney was described as “the UK’s favourite grandmother”.
On “Still I Rise”
The following is according to exit polling for the NJ gubernatorial election. In parentheses are the percentages of the overall votes for each group, followed by the percent of the group that voted for Sherrill, followed by the percent for Ciattarelli.
White (70%) 47% 52%
Black (10%) 94% 5%
Hispanic/Latino (10%) 68% 31%
Asian (5%) 82% 17%
Other (4%) 54% 43%
As mentioned at GftNC's NYT link, groups who shifted right in the last presidential election have moved back left after seeing what they really voted for. I'm curious how age groups voted we well and will probably dig something up later.
I mean, 94% of the Black vote. Wow!
*Comment archive for non-registered commenters assembled by email address as provided.