The "virtual AI judge" has a collection of 120 frequently asked questions. Based on keywords in user input, it picks one of the predetermined answers and uses speech synthesis to have an avatar read it out. So, a chat bot.
More useful is probably the "mobile court", which is supposed to enable filing a suit online and participating in hearings via video conference. The avatar is only a small part of that.
They've also opened a shop on Taobao, which seems absurd to me, but they seem to be very enthusiastic about it.
Before we have developed a machine learning algorithm which can decide what contexts are involved in a corpus by checking which words are included, and in what order and within what subcontext?
At the official website, they also refer to it as AI judge.
Besides that, it is using facial recognition and voice recognition through WeChat running on mobile devices.
On another note, why is there always a push to "humanise" the system with some bullshit robot? It's completely unnecessary. What it boils down to is some weighted matrices doing mathematical work. I would argue that if it was used by a human judge to aid their judgement, mixed in with their prior experience. I would argue that in a lot of cases it is going to make them question their own biases if the system is designed to be race blind and only using the salient facts.
It says "... to be used for the completion of “repetitive basic work” only..."
That is what most software automation attempts to do!