🧵 AI-generated soul music by "Eddie Dalton" is taking over iTunes, YouTube, and Spotify--raising the question of where humans think beauty and art lives.

Software-made music has gotten so good now that 97 percent of people cannot differentiate it from human-made music.
https://plus.flux.community/p/eddie-dalton-isnt-real-but-what-does

Eddie Dalton isn’t real, but what does that mean?

Computer-generated soul music is taking over the internet, raising questions about where humans think art lives

Flux

Eddie Dalton had 3 of the top 5 Apple Music songs, and all of the songs released on YouTube have hundreds of thousands of views within just days.

People absolutely love it.

"This song has touched the depth of my soul," one listener wrote on YouTube.

You can listen here:
https://www.youtube.com/watch?v=az5FSZzm-k8

Another Day Old (Official Lyric Video)

YouTube

@mattsheffield

Sorry, sounds awful. No soul.

@mastodonmigration Is that because you know it's AI though? Given that 97% of people cannot tell the difference, what if you didn't know the context?

There's so much formulaic music made by assembly lines of people who stitch the components together and then the singer is auto-tuned to hell while singing it.

@mattsheffield

But, even though people listen to this processed music, it doesn't have any power. A good song has something special that grabs you, shakes you, and doesn't let go.

As an example consider popular artist Lorde from her first album White Teeth Teens https://youtu.be/lDwOa57FdBI. Whether you like it or not, you have to acknowledge it is brilliant and raw.

She then came under the influence of producer Jack Antonoff, who homogenized and purified her sound and the magic was lost.

White Teeth Teens

YouTube

@mastodonmigration @mattsheffield one _alleged_ listener said that.

Speaking as a musician, someone who has also been a radio professional, and who has worked in-studio on other people's albums: it sounds and feels like music for a commercial. Like it's music for an advertisement promoting an active seniors community living facility, maybe assisted living in particular.

I feel like people who like it are reacting to it punching nostalgia buttons. This works well for older people in particular.

All that makes sense given that the software is trained on pre-existing music. People will connect to parts of what it's copying, and associate the music they remember with this. But anyone looking for a new sound won't find any hint of one here.

@mastodonmigration @mattsheffield To be clear, I'm not saying that people aren't enjoying it. I have to think they are.

But a disturbing number of people - _particularly older people_ - watch commercials _on purpose_, too.

A friend of mine talks about his (elderly, now) mum protesting him muting commercials if it's one, in her words, "she hasn't seen yet."

This strikes me as very much in the same bucket.

@mastodonmigration @mattsheffield _Most people_ latch onto music in their teens, lock onto it, and don't move past it. They like new music less and less as they age. Charts showing this effect go around occasionally. GenX is a little atypical in that the percentage of GenX who aggressively seek out new music despite being long past the typical era for that is meaningfully higher than usual, but even there, that's not most people in the cohort. Just more than usual.

Is there a future of "AI"-generated nostalgia-triggering variations of old music for older people?

Maybe.

Is that what's going on here?

Also maybe.

@mastodonmigration @mattsheffield But is it art?

_Definitely, absolutely not_.

It is product. It may be a product people enjoy, that they want. But it is product.

Not art.

@moira @mattsheffield

Interesting point. As an oldster, most friends are indeed stuck back in the seventies or eighties for the music they listened to in high school and college. But, in most cases, the songs they still play are pretty good tunes that were hits at the time because they had that something special and endure the test of time for the same reason. Each era produces some great music, and AI that makes music that mimics a certain genre will still not have the right stuff.

@mastodonmigration @mattsheffield Evidence suggests it might, at least for some.

I want to be clear that my musical analysis is independent of ethics. The first thing I thought when I saw the its so-close-but-that's-not-a-real-microphone title card was "what're the odds the people behind this _aren't_ white? 50:1? 100:1?" tho' the phrase 'digital blackface' escaped me. It feels icky, and _at very best_, appropriative.

Unfortunately, where there's a way, there's a will, and nostalgia is a monstrous creature with tendrils that reach deep. And _that_ is why the AI version doesn't _have to_ have the right stuff on its own, it just needs to _invoke_ the right stuff _in the listener's memories_. The _listener_ will then fill in the rest.