I have seen this question pop up time and again.
“AI has come for video and music – next will be sound”!
But will it?
It’s clear that AI might eventually be capable of generating complex textures and sounds, but will it be able to synchronise with video? Will it be able to capture the pacing, mood, pacing, or fine nuances of a scene?
What happens when someone cannot describe exactly what they want? How will AI interpret that?
My theory is that some of the basic sounds will easily be replicable – foley – footsteps. A bottle opening, but then again AI is built on copying – and as critics have already pointed out AI is trained by scraping vast amounts of human-created data, and companies are effectively packaging that collective intelligence to sell back to the public. Every LLM lacks context and true creativity. Replicating things in their data set is what they are amazing at. So, can you be 100% sure that what you requested is fresh and original or just a copy of something else? And if so does that leave you open to being sued?
The other day I was perusing a site with AI created videos – I clicked on one and sure the video was ok, but what shocked me was the music it had “created”. It was actually a verbatim copy of a piece of music I love and have bought a licence for to use myself in the past. Here was a rip off – with no safety net. Had someone used that they would surely open themselves to a copyright notice. What’s the say the same doesn’t happen with your sound effects?
How will you even know? Till the lawsuit lands…
Can it? Maybe. But I don’t think it will have the same creativity – and I very much doubt it will be safe to use…
Leave a Reply