Behind the headlines: doubts about China’s “robot judges”
Spanish version: Detrás de los titulares: dudas sobre los “jueces robot” en China
This isn’t an article against the use of artificial intelligence (AI) in judicial systems. Rather, it’s a reminder of the importance of doing some research before repeating what we saw in some article or video, or — even worse — in its headline or description.
The idea I want to convey, that goes way beyond the debate about the use of AI in judicial systems, is the following: someone shouldn’t automatically become a model to follow just because he/she did something first. Especially, if we don’t have enough reliable information about that something. And regardless of our intent, every time we promote the use of AI in judicial systems putting China and its “robot judges” as an example, we contradict this idea.
But, what do we — really — know about China’s “robot judges”?
China’s “robot judges”
If you Google search for “robot judges”, the first results will probably relate to China and Estonia. This article focuses on China’s case.
The following video of a “robot judge” from the Hangzhou Internet Court asking a defendant if she had an objection to the evidence submitted by the plaintiff was uploaded by CGTN, a news channel owned by the Chinese government, on December 6 of last year:
A few days earlier, CGTN had uploaded this video titled “An exclusive tour of China’s first Internet Court in Hangzhou”:
Little over 20 minutes of flat screens, graphs, a remote-controlled robot, and even a drone (in a courthouse?), of which, for the purposes of this article, I’d only like to highlight one thing: nothing is said about “robot judges”. Furthermore, on minute 13:27 a person with the exact same background as the “robot judge” in the first video appears and is presented as the “main judge” of the case. This caught my attention because many clips of the first video were taken from this tour.
It’s worth mentioning that, to date, China has two other Internet Courts in Beijing and Guangzhou.
Doubts about China’s “robot judges”
Headlines like “China and Estonia develop ‘virtual judges’ based on Artificial Intelligence to decide claims (…)” motivated me to investigate a little more about China’s “robot judges”.
Because the message people (let’s forget about lawyers for a minute) receive when they hear about “robot judges” is that, in some place in the world, there are machines replacing human beings in performing the task of deciding claims.
Is this what’s happening in China? Let’s see.
1. Do China’s “robot judges” exist?
No. Several sources confirm that China’s “robot judges” don’t decide claims. Not even small claims. Actually, they’re algorithms that help (human) judges with repetitive tasks. Apparently, this was even admitted by the Beijing Internet Court in a press release. What’s more, in China they’re aware that these “robot judges” act as assistants.
If that’s the case, why call them “robot judges”? In any event, to avoid confusion, we should start calling them “robot assistants” because that’s what they are.
Estonia is planning to use AI to decide small claims (up to €7,000.00 or little over US$7,900.00); however, in Richard Susskind’s words, “this project is little more than an idea at this stage”(1).
2. Do China’s “robot assistants” work?
The little information that I’ve been able to find on the different ways that AI is supposed to be making the work of Chinese judges easier (i.e., what type of repetitive tasks is it handling?) and the results of its use comes from the Chinese government.
For obvious reasons, it looks like there are no impartial organizations producing studies that would allow us to objectively evaluate the (positive or negative) consequences of these reforms.
Something that caught my attention was the extremely positive tone of all of the comments from the China Justice Observer (CJO), whose representatives have — in some occasions — been quoted as the voice of China’s civil society.
3. Would a Chinese “robot judge” be “fair”?
I refer to the statement of the Vice President of the Hangzhou Internet Court, Ni Defeng, in the first video (starting on minute 0:45):
“What we are doing now, you can’t understand it as merely improving efficiency. It also speaks to the issue of legal justice. The faster (justice) — is (a) kind of justice on its own because justice delayed is justice denied.”
I don’t know about you, but this statement leaves me with the sense that efficiency is the supreme value. Claims that are decided faster, at a lower cost. And what about the quality of the decisions and people’s needs?
I’m afraid that any judicial system (with or without AI) designed with that mindset is not going to give us the results we want, if what we want is justice by and for all. People-centered justice.
Let’s not forget that injustice will not be less unjust if we happen to get it in less than 5 minutes through a full HD curved screen.
Fact-checking and transformation of judicial systems
If we want to promote the transformation of judicial systems we have to be very careful with the messages we send, not to groups of “experts”, but to the general public.
By speaking lightly about AI, machine learning, big data, etc., without even glancing behind the headlines, we hurt our cause.
The key words here are responsibility and restraint.
(1) SUSSKIND, Richard. Online Courts and the Future of Justice. New York: Oxford University Press, 2019, p. 166.
