In 2024, I started generating music with AI models. I've generally found that the latest models are now capable of outputting any sound that is possible. The limitations of AI music are now generally caused by my selecting the wrong inferences (it usually takes 600-1000 runs plus editing), not understanding song structure, or not understanding what listeners like and dislike.
Every time I publish a new release, which I've been doing about once per month, I will post it here and create a poll scaled from 1 to 7. The polls will remain open for one week after release. If the average rating on any song is greater than or equal to 5, this market will resolve to YES. Otherwise, if January 1, 2026 arrives before that occurs, the market will resolve to NO.
In 2024, the highest average rating for these polls was 4.33.
I intend to pay for and use the latest models and software tools as they are released, even if they cost money.
Current 2025 catalog:
1,000
3.00@SteveSokolowski Manifold is unfairly biased againts AI generated content and Steve Sokolowski. I think this market will resolve NO.
@jim The whole world is biased against AI generated content. Listen to the first song and tell me whether you think the average human would have composed it better manually. Society still will be biased even when AIs are creating stuff better than Bach (which will probably be later this year at this rate.)
That's fine with me. While everyone else is biased, I'm using AI to become superhuman (as you will see tomorrow on my X feed.)
I'm curious as to how much slippage was lost in that trade. Did you just type an extra zero?
@SteveSokolowski No slippage I think, luckily! There weren't any limit orders and no one was quick enough to trade at between me buying and selling.
I have just listened to Pretend to Feel. I am not a generous reviewer but I was impressed. Lyrics feel a bit too GPTese at places. But the range on vocal techniques etc. is incredible.
2025 will be an interesting year.
I'll try to get into a YES position at 30%.
Fill me @Bayesian?
In 2024, I started generating music with AI models. I've generally found that the latest models are now capable of outputting any sound that is possible. The limitations of AI music are now generally caused by my selecting the wrong inferences (it usually takes 600-1000 runs plus editing), not understanding song structure, or not understanding what listeners like and dislike.
Interesting thesis. Music-gen capabilities overhang.
@jim You saw this with the vocals in that song. By default, AI music vocals are generally poor, and the reason most AI vocals are poor is because people think you can do it without a lot of work
There is a lot of AI-generated garbage music out there because people click "create" and go with the first song. You might be interested to know that of the entire song, only six seconds - the first part of the second verse - came from the original generation. Every single other part was generated through diffusion inpainting and extension prediction.
@jim Oh no, you're up against @KevinBlaw ! He'll be glad to fill your orders, all the way down to zero.
Society still will be biased even when AIs are creating stuff better than Bach (which will probably be later this year at this rate.)
lmfao...no wonder your slop sucks