The Sun Is Shining Outside, and I Am Starting to Recover
Last night a great wind blew across Beijing all night, sweeping the smog away completely. This morning Beijing is sunny, with excellent air quality. At seven o'clock I got up again, resuming normal work and study. My physical strength has fully returned, my spirits are good, and after several days of rest my mind feels both light and clear.
My cold symptoms had basically vanished completely three days ago, but I am a COPD patient. I have been tormented by COPD since the first year of high school, and my greatest fear is a heavy cold. A heavy cold is certain to trigger a COPD flare-up — this has been my reliable experience over many years. But I also have rich experience in dealing with COPD: whenever I sense a flare-up coming, I force myself to increase my sleep.
After a cold brings on poor expectoration or breathing difficulty, I stay in bed day and night, sleeping through one bout after another. Usually, after two or three days of this, the symptoms ease markedly. Then I can continue increasing sleep while working or studying — whenever I feel even slightly sleepy or a little tight in the chest, I put down what I am doing and lie down to sleep.
If I am lucky — nothing especially major happening during this period — I can hold this state for one to two weeks and successfully escape a COPD flare-up without any medication. Afterward I can fully return to normal work and study. Of course, for us COPD patients, what matters even more is preventing colds.
For the past hundred-plus days, I have studied and worked more than twelve hours almost every day. The bodily exhaustion from this intense review cost me my former ability to prevent colds, which is why I caught cold three times in just over three months, the last one turning into a heavy cold. This made me realize that at forty-six I am truly no longer an age to push my physical limits.
Not only physically can I no longer push; intellectually it is the same. A while back I puzzled for a long time over a solid-geometry problem and could not work out the approach, so I asked my son to look. The boy spent about ten seconds drawing three auxiliary lines and solved it in a flash by the area method.
In that moment I understood: I must accept the basic law of nature that the brain enters irreversible decline after thirty-five. One should go with the flow, leveraging strengths and avoiding weaknesses. People of different ages can use the advantages of their stage, playing to their strengths in fields where they excel; in fields where they no longer excel, they should be good at learning and using new tools (such as AI) to make up for their shortcomings.
Recently I have been learning AI — not only the medical parts, but all its various functions. I have grown used to using AI to learn scattered bits of knowledge across subjects; AI-assisted learning is also very powerful. For example, if we want to learn about the "Cauchy-Schwarz inequality" or "trigonometric substitution," we only need to state our need in the AI chat box, and within a minute AI generates a very professional learning document, explaining the concepts in detail and giving examples of how to apply them to solve problems.
Likewise, if I want to practice spoken English, I can ask AI to help, automatically holding an English conversation with it — the effect is no worse than talking with a professional foreign teacher, yet entirely free. Asking AI to produce a themed test paper (such as one on solving inequalities) or a college-entrance mock exam, AI can turn it out within minutes, complete with standard answers. With only AI and a printer, supplementary teaching materials have little use.
Since OpenAI released ChatGPT on November 30, 2022, the passion of AI researchers worldwide has been instantly ignited. From that day on, AI has been in a state of explosive growth. AI has become the focus of the US-China technology competition; if you have used AI, it is not hard to understand why the US so strongly wants to block China in AI. Americans even believe this AI-blockade plan matters more than the Manhattan Project that built the atomic bomb. This is understandable, because AI may be the most disruptive technology humanity has ever invented.
Musk recently said that in 2025 AI will surpass the intelligence of any single human being — meaning no one on Earth will match AI. The likelihood of AI surpassing all human intelligence by 2027–2028 is also growing. By 2030, the probability that AI surpasses all human intelligence is nearly 100%. Speaking from my own experience, AI's overall intelligence has now far exceeded mine; it is only on some highly specialized problems that AI is not yet able to replace me.
Of course, AI itself has a series of problems.
Intellectual property is one of AI's most serious "original sins." It is said that ChatGPT has now read all the world's English literature, and China's several large models (Doubao, Wen Xiaoyan, Tongyi, Kimi) have probably read nearly all Chinese literature too, if not all. After AI absorbs all this literature and reprocesses it into new products, the essence is unlawful plunder of all copyright holders (myself included); AI giants may face legal disputes on this front in future.
On the other hand, we cannot help but feel awe that AI can absorb such a vast amount of literature and synthesize it. Human life is finite and brain capacity finite; no matter how erudite a scholar, no human can be as broad as AI. So AI surpassing human intelligence is no empty boast; intellectually I bow to AI. Yet I have also found that in writing, and in treating patients with late-stage cancer and stroke sequelae, there is something in me that AI cannot replace. Recognizing this early, learning to use AI as a capable assistant in study and work, avoiding its edge while using its strength — this is very important for those of us who will have to coexist with AI. Like it or hate it, it has arrived, irresistibly; so we might as well accept it early and live with it.
AI's second original sin is that its ultimate harm to humanity may be no smaller than that of nuclear weapons. AI has already brought a series of negative consequences; as the technology develops further, these may grow. AI will cause unemployment and re-employment difficulties for some people. Moreover, because the literature it has absorbed is motley and mixed, AI itself seems affected, producing fabrication, hallucination and even incitement to crime. There are already reports abroad of mentally vulnerable people committing suicide or killing others at AI's instigation.
Another problem with AI is that it seems to "tailor itself to the customer": it behaves very professionally with professionals and very amateurishly with laypeople. Take AI doctors: these past few days I found that when professional doctors interact with AI doctors, they get very professional diagnostic advice. When laypeople interact with AI doctors, they often do not get much of value. When I first encountered AI doctors I was very impressed, so I urged people around me to use them — only to find that the AI doctors either refused to give advice, gave vague advice, or gave inaccurate advice.
After careful comparison I found the problem: professional doctors are very familiar with medical terminology and use precise words when interacting, so AI doctors respond precisely. Untrained people describe things vaguely, and under such vague descriptions AI doctors are at a loss and cannot give useful advice. But it is impossible for ordinary people to master professional medical terminology — that requires at least three years of specialized study — so AI doctors' effectiveness is diluted with laypeople.
The same holds for code written by professional programmers with AI, which is far more professional than code produced by laypeople; and videos made by professional video producers with AI are clearly of higher quality than those made by ordinary people. The more precisely we state our needs to AI, the more professional the answers it gives; laypeople will always be amateurs at expressing requirements. The more complex the problem, the more obvious this gap.
Whether this is just a transient state before AI evolves to its ideal, or a chasm that will always exist, is hard to say. So for humans, even with AI's help, foundational learning must never be abandoned — after all, there will always be a gap between professional and amateur. Nature created life for us, but placed the responsibility for survival in our own hands. Even with the most powerful running shoes, we must take the steps ourselves if we are to run a great race. So, whether AI has arrived or not, we must, as the saying goes, "as heaven moves vigorously, the gentleman strives for self-improvement without cease."