Originally Posted by John Lawrence Schick
One thing that AI Users fail to address is that Suno is based on stolen music (lyrics & music), sounds, videos, voices, and probably more that I can't think of at the moment.

John, that concern comes up a lot, and it’s a fair question to raise. But the way these systems work is often misunderstood.

AI models like Suno aren’t databases of songs that get copied and rearranged. During training they analyze huge amounts of material to learn statistical patterns — things like rhythm structure, melodic movement, chord relationships, and how lyrics tend to flow with music.

After training, the system doesn’t have access to those original songs the way a sampler or hard drive would. What it learned are patterns, not recordings.

It’s actually closer to how humans learn music.

A songwriter might grow up listening to Beatles records, Motown, Nashville radio, hip-hop, classical music — thousands of influences. When they write something new, they’re drawing on patterns they absorbed over time. But that doesn’t mean they’re copying those songs.

AI is doing something similar mathematically.

Now, the legal and ethical questions around training data are still being debated — courts are sorting that out right now — but the technology itself isn’t pulling pieces of existing songs and stitching them together.

It’s predicting patterns and generating something new from those patterns.

And just like with any instrument or production tool, what comes out depends heavily on the human using it — their ideas, their prompts, their taste, and their direction.

At the end of the day it’s still a tool.
And tools have always changed how music gets made.

Greg