Amazon demonstrates Alexa mimicking the voice of a deceased relative
cnbc.com
cnbc.com
With the data that companies have already sucked up on lots of people, there's technically little in the way of them creating such "mimics". Is there anything legally preventing them from doing so without a family's consent once the person is deceased? Given they've already got legal access to the deceased's data (presumably e.g. through their social media subscriptions), who (if anyone) "owns" the person's "tone of voice", "style of speaking", "visual likeness", and "bundle of personality traits"? If such things are somehow copyrightable, then surely every comedian doing an impersonation routine is in breach.
I can imagine someone salivating at the thought of all those rich veins of personalities you could harvest on the internet e.g. your uncle that did that cringeworthy tik-tok that went viral (god rest his soul), who has now been turned in an immortal AI "bumbling fool" for e.g. advertising purposes.
I don't like this timeline.
I also can’t imagine speaking to my Grandma/pa the way I speak to Alexa either, so that’s food for thought.
Jests aside, this is already possible with ML and a large enough data set. There’s nothing state of the art to see here, right? Just implementing existing tech at Enterprise scale/telling the masses?
For examples, see 15.ai or https://github.com/CorentinJ/Real-Time-Voice-Cloning. There’s another commercialized service similar to the latter example here I saw recently but I can’t recall the name. I wish the article had some more technical details.
I tried to soak in all the feels while looking through automagically-animated photos of my own deceased grandparents on a genealogy site a couple years back, but eventually lost it when an animation of one photo turned my grandma's accidental blink into something that appeared to be a looping Stevie Wonder impersonation.
I'm gonna be honest, I scrambled to save a copy of that one to send to my sister.
So, speaking for myself as a living person, I am not going to hang a lot of hopes on those treasured solemnities. I admit I mostly hope my kids can get some humor out of this type of thing when I'm gone.
(Plus with the high demand for things that can loiter above battlefields these days, it may be a bit tricky to celebrate my dead body by turning it into a drone like the cat guy did. So naturally, these other options are all welcome.)
My biggest sense of loss is for someone who is still alive, and I think about them every day. I don't think I'd want to risk introducing synthetic samples of them into the already-probably-divergent (and sometimes painful, and embarassing) memories I have of the past.
At the same time: technology and the world around us are changing, and perhaps it's worth considering that future generations may have different norms and preferences.
Currently I don't think I'd want any of my personal or intimate communications to be used in generative models, and certainly not without clear and informed consent (potentially multiple rounds thereof). Workplace and public communications seem more reasonable - and match my bias towards open software and culture.
Where does this path lead? I don't know. It's worth discussing (and if necessary, regulating) though, rather than being surprised when corresponding technology products arrive.