Search engines have started doing something that looks almost too modest to matter: when someone types a phrase suggesting they may be considering suicide, the page interrupts the search with a message, a hotline number, a plea to pause. The version that became familiar through Twitter in the United States is not unique. Douban had a similar intervention early on: search for “suicide” and a box appears, offering advice and numbers to call. Later, Chinese search engines picked up the same feature.
For a modern society, this is a fairly ordinary social function. The web cannot know what a person is actually going through, and a pop-up may feel powerless. But it can be the final handhold someone grabs. On its own, though, the feature is not especially interesting. It becomes interesting beside two other stories.
A test, then an apology
A university arranged psychological testing for students. Some results suggested a tendency toward depression. The school then demanded that those students write an 800-word self-criticism. At first hearing, the story is absurd, but several angles are worth pulling apart. In China, self-criticism is usually a tool used after someone has done something wrong—a way to make the offender obedient, or at least absolutely compliant with arrangements. How does a mental health test become grounds for a self-criticism?
I have written enough self-criticisms over the years to know how I would have ended up in that situation. If a school gave me a psychological test, I would have deliberately chosen the most alarming options and produced a result showing that I was profoundly unwell. When the result came back, a teacher would suspect I had not followed the rules, and I would be made to write a self-criticism for treating the test like a joke.
That is the best-case scenario I can imagine. The worse one is exactly what the story describes: a student’s test is deemed unacceptable, it suggests depression, and the school responds by requiring a self-criticism as a way to improve mental health. That is absurd. It is the logic of a place that believes collective willpower can defeat science—as if depression could be treated, or at least improved, by writing an essay admitting you were wrong.
A search history as evidence
Then there was a “major police alert.” I could not find any news report about it in Chinese search engines. What I saw was a photograph of a computer screen, with a fragment of text saying it concerned a “premeditated murder.” So let us assume for now that the report is fake news.
The alert said Ningbo had successfully warned and handled a premeditated murder involving a college student. According to the text, Wang, an 18-year-old female student at Ningbo Health Vocational and Technical College, had frequently searched on her phone for things like “how to kill a roommate and get away with it.” She was summoned, and she allegedly admitted that she was unhappy with the styles and lifestyles of two roommates, Wang and Xu, and had thought about killing both. On September 12, she bought a fruit knife, a utility knife, and a pair of gloves at a school supermarket and planned to kill them while they slept. The photograph ended there.
I have written before about why people distrust artificial intelligence, and about AI and criminal prediction. Complex computation can infer a person’s criminal intent, at least in a rough way. It cannot reach into someone’s subjective thoughts, but objective signals can still serve as reference points. In the elements of intentional homicide, for example, courts look not only at the defendant’s subjective intent but also at the tools used.
Take a hypothetical offender, Zhang. He wants to argue with a neighbor and picks up a wooden stick to threaten him. During the argument, Zhang swings the stick, intending to scare the neighbor, but the neighbor fails to dodge, is struck on the back of the head, and dies. Is Zhang guilty of intentional homicide? Under ordinary criminal law reasoning, he is more likely to be convicted of negligent homicide. He did not have the subjective malice of wanting to kill, and a wooden stick is not an obviously lethal weapon.
But if Zhang takes a cleaver from his home before going to argue, the conviction starts to lean toward intentional homicide. The question becomes whether, at the moment he swung the blade, his subjective intent was to kill. Of course, this is an idealized legal analysis.
Back to the female student. Searching “how to kill a roommate and get away with it” already suggests some intent to kill. Buying “tools of the crime” moves her closer to the boundary of attempted crime; if nothing intervenes, she could move from preparation to attempt and then to a completed offense. The question is whether this kind of “warning and handling” is reasonable. To answer that, we first have to answer a prior question: is it reasonable that our search histories are monitored in real time?
If the story ended with the school warning Wang and requiring an 800-word self-criticism, would the problem be solved? I suspect it might get worse. Her intention and thoughts have been fully exposed. A self-criticism cannot function as a shackle against crime.
Search for “kill roommate + Zhihu” and you can see that Wang is not the only person with such thoughts. People who are impulsive, psychologically fragile, and easily influenced may at that very moment take the wrong step. So look at the two stories together. When a school arranges psychological testing, what if part of its purpose is to infer which students might pose a risk of killing a roommate? That would be a form of prediction. Then requiring unstable students—those marked as possible roommate-killers—to write self-criticisms almost begins to make sense. The logic is familiar in China: the question is right versus wrong. If you have entertained the thought of killing a roommate, then no matter how wrong the roommate was, no matter how foolish or unbearable, you cannot actually kill them. Therefore the thought itself is wrong.
A Zhihu user asked whether university mental health testing is prevention or fuel. The question described the usual situation: most universities conduct psychological tests and surveys to prevent mental health problems among undergraduates and graduate students. The intention is good, but does it work? The user said students generally react in one of two ways. One: this is a waste of time; fill it in randomly and move on. Two: I know I have mental health problems, but if I am found out, I will be called in for a talk, subjected to ideological education, have points deducted from my comprehensive evaluation, or even be ordered to leave school. So I must pretend I am fine.
If a student is indeed predicted by a psychological test to have a “criminal mind,” does the self-criticism prevent trouble or add fuel? The answer is not comforting. In the end, the fastest way to solve a problem is often to solve the person who raises it or makes it.