When autumn’s edge grew sharper, I spent two full days locked in a sprawling conversation with Gemini 3 Pro—an artificial intelligence model. It wasn’t a simple exchange of questions and answers, but more like charting an unlit map within my own body.
At eighty-one kilos, and having endured over a decade of torment from lumbar disc herniation, I wanted one thing: to understand how a man like me should move.
My wish was modest. Jogging in Zone 2—just enough to melt a few stubborn kilos and soothe the dull, desk-born ache that haunted my back.
It began on a Monday, after my usual run. I uploaded the Apple Watch data, expecting encouragement—perhaps even praise.
Instead, the model returned a verdict as cold as a court order:
“Your foot lingers on the ground for 285 milliseconds,” it said.
“You’re muscling through. Inefficient to the core.”
Worse: asymmetry at ten percent. My hips and back, it explained, moved like rusted gears welded together from years of sitting. To keep pace, I was forever shuffling my feet.
Its tone was blunt, stripped of mercy: “Stop running. Fix the machine before rejoining the race. Otherwise, in ten years, your only ride will have four wheels—and a hospital logo.”
The words hollowed me out. Swimming was too cumbersome, cycling impossible without the space. Running was my single refuge—and now it was gone.
I fought the verdict. Sought second opinions from other models. All delivered the same pronouncement, as though reading from an agreed script: stop.
Climb hills. Take stairs. Awaken the glutes you’ve long abandoned.
That afternoon steeped in frustration, I shouldered a backpack and coaxed my wife—still sniffling from a lingering cold—up the slopes of Nanshan.
The autumn light was generous, yet I walked as if crossing thin ice, every step tentative, convinced my body was near the brink.
Desperation pushed me toward numbers and chemistry. The AI prescribed vitamin D3 and K2 for bone strength, deep-sea fish oil against inflammation, magnesium glycinate before bed to loosen my muscles.
In the mornings, I began farmer’s walks with a ten-kilo plate in my pocket while walking the dog—monk-like in my devotion.
Absurdity crowned itself the next morning. Years of faithful exercise, and yet to the AI, I was a Ferrari engine strapped to a chassis ready to crumble.
Then came the turn. While sorting last year’s medical report, the model suddenly requested my neck, waist, and hip measurements.
I froze. I had been letting a blind man diagnose me—feeding it scraps of data, never the story of who I was.
Frustration surged, tangled with a strange relief. I uploaded everything: medical history, full body measurements, the latest exam results.
This time, its tone shifted entirely. The “frail patient” morphed into a “warrior waiting on orders.”
I took the risk. Laced up. Warmed my limbs. Ran thirty minutes in Zone 2 at a steady cadence of 180, closing with a fierce walk.
I sent the new results—zero percent asymmetry, cadence locked at 180—back to Gemini.
Its reply: the ban is lifted.
Keep your cadence. Hold your vertical oscillation. You’re safe.
A hundred thoughts pressed at once.
Two days had thrown me through a rollercoaster, leaving behind a clear truth: AI is a mirror. Show it only a shard, and it will return a distorted face.
When you ask an expert—even a machine—about your pain, background is not optional. Context is not decoration. Data alone is dead; only when embedded in your full humanity does it breathe, and become a key.
And life? Much the same.
Most of the time we are stuck not because the problem is insoluble, but because we’ve forgotten to lift it into a wider sky.
秋意渐浓的时候，我花了两天时间，和 Gemini 3 Pro 这个人工智能模型进行了一场漫长的对话。这不像是一次简单的问答，更像是一次对我身体内部疆域的勘探。我想搞清楚，作为一个体重 81 公斤、被腰椎间盘突出折磨了十几年的中年人，到底该怎么运动。
起初，我的愿望很朴素：通过 Zone 2 的有氧慢跑，既能减掉几公斤赘肉，又能缓解那种长期伏案带来的、隐隐作痛的背疾。
一切始于那个周一。我照例跑完步，将 Apple Watch 里的数据喂给了模型。原本期待的是几句鼓励，没料到，它给出的结论像是一张冰冷的判决书。
“你的脚在地面黏滞了 285 毫秒，”它说，“你在靠肌肉硬撑，效率极低。”
更糟糕的是不对称性，达到了 10%。这意味我的髋部和背部像生锈的齿轮，已经由于长期久坐而粘连。为了维持配速，我不得不频繁地倒脚。模型的话语没有温度：“必须立刻停止跑步。先把车修好，再上赛道。否则十年后，你唯一的交通工具就是轮椅。”
看到屏幕上这几行字，我感到一种深刻的绝望。对于一个腰椎不好的人，游泳太麻烦，骑车没场地，跑步几乎是我唯一的慰藉。现在，这个慰藉被剥夺了。
我不甘心，又找了几个不同的模型交叉验证，试图找出破绽。但它们像是商量好了一样，众口一词：停止跑步。它们建议我去爬坡，去走台阶，去激活那个已经“死掉”的臀部。
这是一个充满挫败感的下午。我不得不放弃跑步，背着包，拉着感冒还没痊愈的妻子去爬南山。秋日的阳光很好，我却小心翼翼，每一步都走得如履薄冰，仿佛我的身体真的已经处于崩溃边缘。
为了自救，我开始变得迷信数据与化学。模型建议我补充维生素 D3 和 K2 保护骨骼，吃深海鱼油抗炎，睡前服用甘氨酸镁松解肌肉。我甚至在早晨遛狗的二十分钟里，像个苦行僧一样，在兜里揣上 10 公斤的杠铃片做农夫行走。
这种荒谬感在第二天早上达到了顶峰。虽然我坚持锻炼了十几年，但在 AI 眼里，我就像给一台法拉利引擎配了一副快散架的底盘。
直到我想起去预约体检，事情才迎来了转机。
在整理去年的体检报告时，模型突然向我索要颈围、腰围和臀围的数据。那一瞬间，我意识到自己犯了一个巨大的错误——我一直在让一个盲人给我看病。
此前，我只给了它破碎的运动数据，却从未告诉它我是谁，我的身体基础如何。
我感到一阵由衷的懊恼，但也伴随着一丝狂喜。我立刻补全了所有信息：最新的体检报告、身体围度、既往病史。这一回，AI 的态度发生了战略性的逆转。原本那个“虚弱的病人”，在数据的重新组合下，变成了一个“待命的战士”。
我决定赌一把。换上跑鞋，下楼，充分热身。 我在 Zone 2 的心率区间里，用 180 的步频跑了 30 分钟，最后加了一段大幅度快走。回来后，我把这次全新的数据——不对称性降为 0%，步频稳定在 180——再次发给它。
这一回，结论变了：全面解除禁跑令。它告诉我，只要维持这个步频和垂直振幅，跑步是安全的。
看着屏幕上的字，我心里五味杂陈。
这两天像是一场过山车，让我明白了一个朴素的道理：人工智能像是一面镜子，如果你只给它看局部，它只能反射出扭曲的景象。
当我们向专家——哪怕是 AI 专家——求助时，仅仅描述“疼”是不够的。你需要交代背景，交代上下文，交代你这个人的全部。数据是死的，只有放在具体的人身上，也就是有了“语境”，它才能活过来，变成解决问题的钥匙。
生活也是如此。很多时候我们陷入困境，不是因为问题无法解决，而是因为我们忘记了把问题放在更广阔的背景里去打量。