A media company is considering a project to produce content using the voices and likenesses of real people with generative AI. When reviewing the project from a responsible AI perspective before release, which concern should receive particular attention?
Select the concern to watch for when using voices and likenesses.
The increased consumption of compute resources during training and generation
Compute resource consumption is a consideration from a sustainability (environmental) perspective.
However, in this question, which involves the voices and likenesses of real people, the concern that requires 'particular attention' is the matter of the individuals' consent and rights, so this is incorrect.
Using voices and likenesses without the consent or rights of the individuals
This is correct. Using voices and likenesses without the individuals' consent or rights clearance leads to infringement of portrait/publicity rights and impersonation. Responsible use requires verifying consent and rights.
The output changing every time even with the same prompt
The non-determinism of output is a general characteristic and challenge of generative AI.
It is considered for quality control, but it is not the ethical or legal concern that is specifically at issue when handling people's voices and likenesses, so this is incorrect.
The quality of the generated content being lower than the filmed footage
A difference in quality is a matter of the production's quality and is not an ethical concern of responsible AI.
What is at issue in this question is the risk involving the individuals' rights, so this is incorrect.
When using people's voices and likenesses with generative AI, 'using them without the individuals' consent or rights' is a serious concern. Because it leads to infringement of portrait/publicity rights and impersonation, verify consent and rights before use.