2 ms·
I kept thinking it should be possible to make something like it back in the day, and now with LLMs and deepfake, it's even closer to reality!
by exe34 3y ago
I kept thinking it should be possible to make something like it back in the day, and now with LLMs and deepfake, it's even closer to reality!
- sigilis 3y agoI’ve seriously toyed with the concept. It may be possible now to get an application like that on a tablet, offloading computation to OpenAI and its ilk, but it wouldn’t be reliable. We’re not good enough at text and image generation to do it autonomously yet. We’re a little better at speech synthesis than Stephenson predicted, but it might still be nice to have a ractor too.