If I recall their AI was licensing an actors voice for the game alone.
But the actor got paid still for their voice. They’d train the AI on it and then the devs could have the AI say your name or say whatever daily thing they’re doing.
I’m personally less worried about it done this way, especially if it requires the human to remain involved to retrain / replace.
But what about all the other people’s work that trained the ai to do the mimicry? That’s what upsets me about AI. What you’re focused on is basically a likeness contract, which existed long before AI.
That’s what I’m saying, likeness contracts with AI seem ok.
Idk if the RAG requires training on other data first? I have only worked with LLM/AI software dev on a very minimal level so am unsure tbh.
Generally believe clankers bad, but AI models trained on consenting humans (and their data) while being paid for it seems ok? Minus the whole climate destroying bits.
My understanding is that before it’s trained on the actors voice it still requires massive amounts of data for the neural net to understand the rules of human speech. Those people aren’t being compensated or even asked.
If I recall their AI was licensing an actors voice for the game alone.
But the actor got paid still for their voice. They’d train the AI on it and then the devs could have the AI say your name or say whatever daily thing they’re doing.
I’m personally less worried about it done this way, especially if it requires the human to remain involved to retrain / replace.
But what about all the other people’s work that trained the ai to do the mimicry? That’s what upsets me about AI. What you’re focused on is basically a likeness contract, which existed long before AI.
That’s what I’m saying, likeness contracts with AI seem ok.
Idk if the RAG requires training on other data first? I have only worked with LLM/AI software dev on a very minimal level so am unsure tbh.
Generally believe clankers bad, but AI models trained on consenting humans (and their data) while being paid for it seems ok? Minus the whole climate destroying bits.
My understanding is that before it’s trained on the actors voice it still requires massive amounts of data for the neural net to understand the rules of human speech. Those people aren’t being compensated or even asked.