Do this for several days, a We are the granddaughters of the witches you weren’t able to burn shirt, getting them to work through not only a full range of normal conversational topics but also emotionally stimulating or aggravating ones. Now feed all this training data into a collection of neural nets that will control the computer generated model: body movement, dialog, question-answer, speech synthesis, lip-sync, facial expressions, eye movement, blinking, pupil dilation whatever detail you can map onto the computer graphic or robotic subject. Train these specialized neural nets with all the performance capture data. You could even supplement it by training the model on additional plain video, audio recordings, or even text transcripts of conversations to flesh it out. You could use GANs to discriminate between the digital mimic and the real person, and drive the convergence of the digital person’s appearance, behavior and speech closer to the real.
We are the granddaughters of the witches you weren’t able to burn shirt, hoodie, sweater and v-neck t-shirt
Offical We are the granddaughters of the witches you weren’t able to burn shirt
Or even better, use our BICHON neuromorphic networks that can learn dynamically while the We are the granddaughters of the witches you weren’t able to burn shirt, using something more similar to a real human nervous system to learn to mimic the person. In the end, whatever learning model you use, if this all goes well, you should have a copy, or mimic of a person, that, within the range of data you have on that person, should act almost exactly like them when you talk to it and interact with it. Then you can add the ability to draw information from databases and the internet to make the mimic a whole lot more knowledgeable in what it is able to say, and also make it very capable of doing a real job, with a human-like personality to boot. I think this is very feasible, and although it is not, by definition, a Strong AGI, I think it could be a precursor.