Trusting Technology Blindly Is Worse Than No Technology

One morning I saw a video of a young entrepreneur — who studied AI in Cambridge and now runs an AI startup — holding forth on AI's impact on humanity. He believes jobs like lawyers, doctors and office white-collar workers may all be replaced by artificial intelligence. On a whim, I decided to test an AI lawyer.

I tested the AI lawyer using a real traffic-accident insurance claim case I encountered in 2024. In the accident I was riding a bicycle; the other party was driving a motor vehicle, and we collided. The traffic police on the scene issued a "Road Traffic Accident Liability Determination" holding me mainly responsible and the other party secondarily responsible. The other party's insurer later sued me, claiming repair costs. This is a real case that has already been judged at first instance by the Haidian District People's Court of Beijing and finally by the Beijing Financial Court.

I hoped the AI lawyer would help me achieve a result of paying not a cent in repair costs. Regrettably, the AI lawyer told me that in Beijing the chance of fully avoiding compensation for the other party's repairs and winning was very low, and it suggested I change my goal to reducing the compensation proportion. I said I still hoped to pay nothing and asked it for a plan. In the end, the AI lawyer said it could not produce such a plan. If you test an AI lawyer the same way, I believe the answer will be the same.

In fact, the final judgment in my case came down this August. I won, and paid not a cent — a result that thoroughly changed the trial rules of Beijing's courts for similar cases. I wrote a separate article about how I won, relying on statutory provisions and rigorous logic. Since that older article exists, I will not repeat it here; interested readers can search for it themselves.

If I had believed the AI lawyer back then, I might not have had the courage to fight the case. Of course, what this startup star said also has a point. Why? Because at the time I consulted Beijing friends who are lawyers, judges and traffic police about my own case, and their opinions matched the AI lawyer's. In that sense, the AI lawyer can indeed do the daily work of an ordinary legal worker, and its judgment was not off base. But creative legal work that ordinary lawyers cannot do, AI lawyers cannot do either.

I am not a lawyer; this lawsuit was the first of my life. Before it, I had no legal experience at all. I won by combining scraps of legal knowledge from friends with everything I dug up on the internet using my best "associative learning method." I synthesized this legal knowledge, grasped the core, and asked the court, under the principle of "like cases judged alike," to rule for me. In the end, both the first- and second-instance courts did rule for me on exactly that basis.

Likewise, a friend urged me a year ago to use AI for writing, to save my own energy and time. After trying AI writing, I did not take his advice. This friend had previously hired several editors to help maintain his websites; once he mastered AI writing, he dismissed them all. He is not himself good at writing, and the editors he hired were only average, so AI's output satisfied him.

But honestly, I feel AI-written articles do not meet my standards. I have never been fully satisfied with my own writing and am always striving to improve it; but I am even more dissatisfied with AI's writing, and cannot accept its quality. Although most people are satisfied with AI writing, I am the exception.

I have had some writing talent since primary school, and from seventh grade on I have written almost every day, only occasionally stopping when too sick to get out of bed. Not to be immodest, my writing is indeed above average, so I personally cannot accept the quality of AI writing. AI articles may have no grammar or logic problems, but it is hard for them to infuse a piece with masculine vigor or feminine grace. Some combative prose by real authors carries, between the lines, an implicit wind and thunder.

Take Chen Lin's denunciation of Cao Cao, Luo Binwang's denunciation of Wu Zetian, or Liang Qichao's denunciation of Yuan Shikai: such writings are both literary and sonorous in a way AI could never produce. I do not reach the level of those masters, but I believe there is something in my writing that AI cannot match, so I will not lower my standards and let AI do it for me.

In medicine I have similarly found that much original work is beyond AI. Many breakthrough medical insights must be gropped out by doctors in practice; the flexibility and adaptability involved are something AI cannot master to perfection, though human doctors can. However clever AI is, it has inherent limitations and blind spots; in some respects it may even be worse than a cat.

My basic attitude toward new technology is: do not resist it, do not belittle it, but also do not exaggerate it or believe it blindly. Neither treat it as a frivolous trick nor as a monster. It is necessary to actively adopt it for one's own use, but equally necessary to retain one's own initiative. Knowledge and technology are important tools for progress; but if we over-rely on tools and neglect our own innate potential, then indeed, trusting technology blindly is worse than no technology at all.

Finally, a joke: "great strength produces miracles" — when facing a thorny problem, AI cannot work miracles; miracles require great strength. But here by "great strength" I mean creative thinking, not the "strength potion" the "Great Strength Brother" talked about. In an age of technological innovation, creative thinking is precious. I think a great benefit of AI's arrival is that it will change our education. If our education remains centered on pouring in knowledge, we will indeed be eliminated by AI — how can the human brain match a chip for memorization? But if our learning and education are centered on creative thinking, we will not be eliminated by the times. The world has no shortage of talent that goes from 1 to 100, but it will always lack talent that goes from 0 to 1.