The Collapse of Code: When Liberal Arts Majors Start Using “Philosophy” to Command AI, Does the Programmer’s Moat Still Exist?
——The “Average” Tragedy of Medicine: We Are All Unique Patients
Is the doctor saying “80% effective”? Revealed! The “Banality of Evil” Behind Modern Medicine
Hello, friends. I am the Financial Veteran from [FinSages].
It is early February 2026. Outside the window, AGI is reshaping the atomic structure of every industry at a dizzying speed. Yet, I find that in the face of such drastic change, many people’s anxiety remains stuck in the inertia of the previous era.
Two days ago, I attended a parent-teacher meeting for my niece. In that private school claiming to be “connected to the future,” I witnessed an absurd scene: a group of highly educated parents were crowding around a tutor from a training institution, scrambling to sign their children up for a “Children’s Python Programming Class.”
That teacher was shouting, with spittle flying: “In 2026, not knowing how to write code is the new illiteracy! Python is the ticket to the future!” The parents’ eyes shimmered with a mixture of fear and desire, as if swiping their cards would guarantee their children’s future.
Watching this, I felt a deep sense of sorrow. It reminded me of the parents in the late 19th century who tightened their belts to send their children to learn “Morse Code.” In that era when the telegraph was just rising, telegraph operators were the highest-paid professionals. People naively believed that only by mastering that complex set of “dot-dash” coding rules could they control the power of information.
However, just a few years later, the telephone became popularized. Humans spoke directly with their mouths, and voices traveled along copper wires to the other side of the ocean. Morse code instantly lost its value as a “universal skill” and degenerated into an extremely niche professional skill.
Today, speaking as a veteran who has navigated the games of finance and risk for 30 years, I must tell you a conclusion that will overturn your cognition with the utmost responsibility:
In 2026, Python is the Morse code of our time. Continuing to regard “learning programming syntax” as a core competency is your greatest strategic misjudgment of the future.
The explosion of AGI has not made programming disappear, but has caused a fundamental “phase change” in programming. It has detonated that “Tower of Babel” standing between humans and machines.
I will strip away all the technological bubbles and take you back to the source of logic and language. What we are going to explore is not how to become a faster “code monkey,” but how to become a “creator of the new era.”
Part 1: The Fall of the Tower of Babel and Musk’s Prophecy
I. Echoes of History: The Twilight of the Telegraph Operator
To see the present clearly, we must first look back at history.
In the 1890s, if you could send and receive Morse code at a speed of 40 words per minute, you were the “gold-collar” worker of that era. You held the privilege of translating complex human thoughts into machine-transmittable pulses. This “translation capability” constituted your professional moat.
However, when Bell invented the telephone, and when human voice could directly drive the transmission of information, that intermediate class of telegraph operators responsible for “translation” collapsed rapidly.
Today, the scene we witness on GitHub bears a striking resemblance to the telegraph offices of yesteryear. According to the latest data, over 80% of the new code generated globally in 2025 was created by AI Copilots or fully automated programmers like Devin. In 2026, a product manager with zero programming background says to the screen: “I want to build a family expense tracking app, minimalist style, capable of automatically recognizing supermarket receipts.” Minutes later, tools like Cursor or Windsurf automatically generate thousands of lines of perfect Python or Swift code in the background, completing compilation, packaging, and deployment.
In this process, Python, C++, and Java still exist, just as Morse code still exists in the underlying signal transmission. However, “humans no longer need to write them personally.” This validates Elon Musk’s famous prediction from an interview a few years ago: “In the future, no one will need to program. Our goal is for computers to understand human language. In fact, humanity’s native language is the highest-level programming language.”
At the time, many mocked him as a madman. Now, we have to admit, he was the first to see the “Tower of Babel” fall.
II. Economic Judgment: The Code Market Under the Four Forces Model
If we examine the programming market of 2026 using the “Four Forces Model,” you will see an extremely brutal landscape of supply and demand.
First is the explosion of “Expansion Force.” In the past, code was a scarce resource because training a qualified programmer required four years of university plus two years of project nurturing. Now, the supply of code is infinite. The marginal cost of AI-generated code approaches zero. As long as there is computing power, you can have as much code as you want. When the supply of a resource becomes infinite, according to the law of supply and demand, its price must crash to the floor.
Following closely is the cleansing by the “Contraction Force.” Those “junior programmers” (colloquially known as CRUD Boys) who can only translate explicit requirements into code are witnessing their salaries cliff-dive. Business owners are discovering that hiring a veteran employee who understands the business, paired with a super AI, is far more cost-effective than maintaining an expensive development team.
Therefore, if you are still forcing your child to learn `if name == ‘main’:`, you are effectively forcing them to learn an “inflationary currency” that is rapidly depreciating.
Chapter 2: The Prison of Syntax—The Pain of “Dimensional Reduction”
To understand why “not learning Python” is progress, we must first deeply understand: What exactly was programming in the old era?
From a philosophical perspective, traditional programming is essentially a painful “dimensional reduction.”
I. The Castration of Thought
Human thinking is divergent, three-dimensional, fuzzy, and full of metaphors.
For instance, as a Chief Risk Officer, when I evaluate a company, my intuition tells me: “This boss has shifty eyes; although the financial statements are perfect, he changes the subject when asked about supply chain details. This company has the ‘scent’ of an impending implosion.”
This is a high-dimensional judgment synthesizing experience and logic.
But in the old era of programming, if I wanted to hand this judgment over to a computer, I had to “castrate” it into rigid syntax:
“If current_ratio < 1.5…”
“If accounts_receivable_days > 90…”
In the process of translating “scent” into “code,” a vast amount of subtle, critical information (Nuance) was lost.
To accommodate the machine’s low intelligence, we had to forcibly compress our high-dimensional thoughts into the cage of if/else logic.
This is why many senior business experts always feel that systems built by programmers are “missing something.” Because that system only inherited the expert’s skeleton, not their soul.
II. Memory of 30 Years Ago: The Soul Tortured by Syntax
Recalling 30 years ago, when I first entered the industry, I taught myself early programming languages to handle credit data.
Back then, I often sat withered before the computer all night for a missing semicolon or a misalignment in indentation.
My brain held clear risk control logic, but I could not tell the computer. I was like a gagged orator, struggling to express myself through sign language (code).
That frustration was etched into my bones. I wondered: “Humans invented machines, so why must humans become slaves to machines?”
III. AGI’s Jailbreak: From Syntax to Semantics
The emergence of AGI is a magnificent “jailbreak.”
It is the first time machines have acquired the ability to understand Semantics. In 2026, I no longer need to translate “shifty eyes” into “liquidity ratio metrics.” I can throw the interview recording to a multimodal large model and tell it directly: “Please analyze this manager’s tonal stress values and micro-expressions, and combining this with industry averages, assess his integrity risk.”
The machine understands. At this moment, we no longer need to dimensionally reduce ourselves to accommodate the machine. For the first time, we can use “full-bandwidth” human language to command silicon-based intelligence.
Those who are still complacent about “mastering complex syntax” are actually showing off their shackles. They have not realized that the prison door is already open.
Chapter 3: The Misread “Technical Barrier”
In society’s general perception, there remains a huge misconception: “People who can write code possess technical barriers; people who cannot write code are easily replaced.”
In 2026, this view is not only wrong but fatal.
I. Code is Not a Barrier, It is Infrastructure
Let us return to the first principles of business. The essence of business is “discovering needs” and “satisfying needs.” Code is merely the infrastructure material in the process of “satisfying needs.” It is like bricks and cement on a construction site.
In the past, because laying bricks (writing code) was hard, bricklayers (programmers) held high status. Now, with 3D printers (AI), laying bricks has become extremely simple.
So, where has the real barrier shifted? It has shifted to the hands of the people “designing the blueprints.”
II. Case Decomposition: Feature Stacking vs. Soul Weaving
Let us look at a real detective decomposition case. Suppose we want to develop a “Companion App for Seniors with Alzheimer’s.”
Contestant A: Senior Python Engineer (Focused on Syntax). He masters various libraries and writes code fast. The APP he makes: Powerful features, precise medication reminders (using complex cron job algorithms), perfect location tracking (calling high-precision GPS interfaces), and an interface full of high-tech vibes.
Result: The elderly cannot use it at all. Those complex buttons and cold notification sounds make them feel panic and frustration.
Contestant B: Psychology Major (Focused on Semantics). He cannot write a single line of code. But he has read dozens of books on geriatric psychology and volunteered in nursing homes for three years. He does not understand Python, but he understands “loneliness.” He commands the AI Agent using natural language:
“The interface should not have buttons; it should look like an old photo album.”
“The medication reminder sound must not be an alarm clock; synthesize the voice of the senior’s granddaughter, make the tone coaxing, not commanding.”
“When it detects the senior staring blankly for a long time, automatically play their favorite Peking Opera snippets from their youth.”
The AI understood his intent and automatically generated all the code in the background to implement these functions.
Result: This APP pierced the hearts of countless families and became a hit.
III. Who is the Architect of the New Era?
In this case, who possesses the real barrier? Is it Contestant A? His code can be generated by AI in a second. It is Contestant B. That “empathy” for others, that “profound insight” into human nature, and that “systems thinking” to translate insight into product logic are things AI cannot pile up through computing power for a long time to come.
So, please remember a survival iron rule of 2026: The real technical barrier never lies within the technology itself, but outside of it. It lies in the “granularity” of your understanding of the world, the “acuity” of your perception of pain points, and the “imagination” to mobilize resources to solve problems.
Writing to this point, I believe you have begun to have a brand-new understanding of the word “programming.” The Tower of Babel has fallen, and the cage of syntax has been broken. But this is only the first step in dispelling the illusion.
Many will ask: “Since we are not learning Python, what should we learn? I also want to command AI, but I find AI doesn’t understand me, or the results are idiotic. Why is this?”
This leads to the heavyweight topic of our next part. Although we do not need to learn the machine’s language (Python), we must learn a higher-order language—”Logical, Structured Natural Language.” I will reveal to you: Why is “Prompt” not a spell, but a scaffold for thinking? Why is a veteran who understands “The Art of War” [Sunzi Bingfa], or a liberal arts student proficient in Kantian philosophy, the most formidable “Super Programmer” in the AGI era?
Chapter 4: Prompt is Not a Spell, It is a “Scaffold” for Thinking
Let us start with the biggest misconception currently in the market.
Even in 2026, I still see the internet flooded with merchants selling “Prompt Encyclopedias.” They package Prompts as “magic spells” from the Harry Potter world, as if as long as you memorize those mysterious word combinations—like “Unreal Engine 5 render, 8k, hyper-realistic”—you can summon a dragon.
This is an extremely superficial and dangerous cognition. If you pin your hopes of commanding AI on chanting spells, you are still doing “cramming,” which is no different in essence from rote memorization of Python syntax.
A real Prompt is not a spell at all; it is a “scaffold for thinking.”
Imagine, as a Risk Director with 30 years of experience, guiding a high-IQ intern (which is exactly the role of AGI) to write a credit analysis report, what would you do? Would you chant spells to him?
Of course not. You would build a clear “thinking framework” for him.
You would say: “Xiao Li, for this report, first establish the authenticity of this company’s ‘blood-making capability’ through cash flow analysis, this is the first layer of logic; if the blood-making capability is fine, the second step is to screen its related-party transactions to see if there is interest tunneling, this is the risk baseline; finally, combine the macro cycle of the industry to give a comprehensive rating recommendation. Remember, I don’t want a pile of data, I want the causal chain behind the data.”
Look, this is a top-level Prompt. There are no metaphysical vocabulary words here, only “structured thinking.”
In the AGI era, the so-called “programming ability” has undergone a complete isomorphic mapping—it is equivalent to your “writing ability” and “logical deduction ability.”
There is a famous concept in computer science called “Chain of Thought.” Actually, this is not a new invention at all. The outline you list when writing policy papers, the business model loop you deduce when writing a business plan, the argumentative system you construct when studying philosophy—these are all essentially “Chain of Thought.”
A liberal arts student trained in rigorous logic, or an analyst who has written hundreds of thousands of words in deep research reports, naturally possesses the ability to dismantle complex problems into ordered steps. When they migrate this ability to conversations with AI, they are performing the highest level of programming.
For example, when I want AI to help me develop an “Automated Public Opinion Monitoring System,” I will not agonize over which crawler library to use like a junior programmer. I will build such a logic tree using natural language:
“First, you need to act as an extremely sensitive ‘PR Crisis Accountant.’ Your task is not to collect all information, but to calculate ‘Emotional Debt.’ Step one, crawl discussions about our company from the entire web; Step two, don’t just look at keywords, establish a ‘Sarcasm Index’ model to identify comments that praise on the surface but are ironic in reality—this is a difficulty in human language, you need to invoke irony detection algorithms; Step three, once this index exceeds the threshold, immediately generate a brief containing ‘Source of Emotion, Propagation Path, Suggested Response Strategy’ and push it to my phone.”
When this paragraph is input to the Agent, the complex instruction set executed by the machine far exceeds the dimensions that a typical programmer can achieve by hand-writing thousands of lines of code. Because contained within is the definition of “irony” and the logical judgment of “crisis.”
So, the conclusion is inspiring: If you have honed your logical thinking and improved your structured expression ability over the past thirty years, then congratulations, you do not need to relearn programming. You only need to learn to “translate.” Translate that magnificent logical edifice in your mind into clear native language for the AI to hear.
Chapter 5: Why Are “Veterans” Better Suited to Be Architects?
If logic is the skeleton of programming, then “experience” is the soul. And on this point, industry veterans possess advantages that young people cannot match.
The ultimate problem of programming has never been the complexity of algorithms, but the “ambiguity” and “edge cases” of the real world.
Junior programmers or young Prompt Engineers often tend towards “linear thinking”: If the user clicks A, execute B. Their world is idealized and frictionless. But the real world is full of lies, fraud, misoperations, and unpredictable Black Swans.
At this time, a “veteran” who has rolled through the mud, their value is demonstrated vividly.
Let me tell you a real “detective decomposition” case. This is 2026, a scene from a fintech company developing a “Next-Generation Credit Approval Agent.”
There were two types of people on the team. One type was fresh graduates with Master’s degrees in CS from top universities, proficient in various algorithms; the other type was like me, a retired bank branch manager who had done credit approval for 30 years and seen countless fraud tricks.
When the young master directed the AI, the instruction was: “Please train a scoring model based on the applicant’s credit report, income proof, and bank statements. If the score exceeds 600 and income covers debt by more than 2 times, automatically approve.”
This logic sounds perfect, meets textbook standards, and the code runs fast.
But I took one look and shook my head. I knew that if this system went live with this logic, it would be drained dry by black market gangs within three days. Because in the real world, perfect statements can be forged, and perfect credit histories can be farmed.
So, I took over the console. I did not write code; I just used my 30 years of experience to say a paragraph to the AI, a paragraph later known as “The Veteran’s Gaze”:
“AI, listen up. From now on, you are not an accountant who only knows how to count; you are a naturally suspicious ‘old detective’ who has been deceived many times. When seeing a perfect application document, your first reaction should not be to pass, but to suspect.”
“You need to focus on the following ‘counter-intuitive’ patterns: First, if the applicant’s bank statement shows money entering and leaving on fixed days every month, and the balance is almost zero, no matter how large the amount, mark it as ‘bridge loan’ suspicion; Second, if the applicant’s mobile number is a new one activated within three months, but the address book has five hundred contacts and none are family, this is extremely likely a ‘white list’ forged to deal with debt collection; Third, compare the applicant’s company address with his mobile location late at night—if the distance exceeds 50km and is consistently so, mark as ‘shell company’ risk.”
“Only when these hidden suspicions based on human nature are ruled out can you call that scoring model.”
When the AI received this instruction, the risk control model it generated in the background underwent a qualitative mutation. It began to call geolocation APIs, conduct complex graph association analysis, and simulate the human “chain of suspicion.”
In the end, this Agent “trained” by the veteran intercepted over 90% of hidden fraud. While that young master’s model became our internal negative example.
In this case, who wrote the code? The AI. But who endowed the “Risk Control Soul”? The veteran.
Experience is the best de-noising algorithm; vicissitude is the highest level of training data.
Young technical personnel often indulge in the sharpness of tools, while veterans understand “awe.” In the AGI era, this insight into human nature, the smell for risk, and the understanding of complex interest relationships are barriers that no amount of computing power can brute-force crack.
So, a salute to all industry veterans: Do not feel inferior because you don’t know Python. The lessons hidden in your wrinkles are your most expensive assets in this era. As long as you learn to ask AI questions, you are the top-tier architect.
Chapter 6: The Counterattack of Liberal Arts Majors—The Granularity of Semantics
Having discussed logic and experience, we must turn our gaze to a group often slighted by the tech world—liberal arts majors. For a long time, literature, philosophy, and art were considered “useless learning,” receiving the cold shoulder in the job market.
But in 2026, the situation is undergoing a shocking reversal. Liberal arts majors are launching a hearty “revenge.”
Wittgenstein has a deafening quote in “Tractatus Logico-Philosophicus”: “The limits of my language mean the limits of my world.”
On the interface of human-computer interaction, this sentence can be rewritten as: “The granularity of your language determines the precision of AI creation.”
We know that current Generative AI (whether generating text, images, or video) is fundamentally based on probabilistic prediction. It requires the user to provide precise “semantic guidance.” If the user’s language is impoverished, what AI generates is mediocre; if the user’s language is exquisite, what AI generates is stunning.
It is like holding a peerless zither (AGI) in your hands, but if you only know how to play “Twinkle, Twinkle, Little Star,” the sound it produces is just a nursery rhyme; only when you have the grand chapters of a symphony in your heart and can touch the strings with precise fingering (language) can it play soul-shaking music.
Let us look at a “Video Generation” comparative experiment.
Suppose we want AI to generate a video of an “Autumn Forest.”
A user with a STEM mindset (accustomed to parameters) might input: “Forest, autumn, fallen leaves, HD, 4K, realistic style, sufficient lighting.” The video generated by AI would be standard: a pile of yellow leaves, bright, clear, but that’s it. It looks like a stock photo, with no emotional appeal.
How would a user with deep literary literacy (accustomed to artistic conception) input? He might mobilize the masterpieces he has read in his life and input a passage like this: “Please display an ancient forest in the bleak atmosphere of deep autumn. The light should not be direct white light, but that ‘amber color’ of ‘the setting sun is infinitely good, only near dusk,’ carrying a kind of decadent magnificence. The camera should capture the process of a leaf falling, embodying the heavy sense and fatalism of ‘the boundless falling wood rustles down.’ The background sound should have the whimpering of wind through dry branches, creating a melancholy of ‘homesickness on a journey.’ The color saturation should be low, like a faded oil painting.”
When AI receives high-granularity semantic symbols like “bleak,” “fatalism,” “whimpering,” and “faded oil painting,” it is forced to search the latent space for feature vectors that are deeper, more complex, and more artistic.
The final generated video might be a movie-level work of art. The falling of that leaf is no longer a physical motion, but a catharsis of emotion.
This is the “Granularity of Semantics.”
In the AGI era, people with a poor vocabulary can only command AI to generate “standard goods”; while liberal arts students who possess a rich vocabulary and understand synesthesia, metaphor, and rhetoric hold the “high-precision control knobs.”
Literary cultivation is no longer a useless decoration; it has become solid productivity.
You understand philosophy, you can command AI to write deep business insights;
You understand psychology, you can command AI to design product mechanisms that are addictive;
You understand aesthetics, you can command AI to draw soulful blueprints.
So, I want to say to all liberal arts students: Pick up your books. Read Shakespeare, read “Dream of the Red Chamber,” read Kant. Do not think these things are useless in the AGI era. On the contrary, it is these most brilliant crystals of human civilization that endow you with the “divinity” to harness silicon-based intelligence.
When others are bitterly reciting Python function libraries, you are honing your language, enriching your feelings, and elevating your aesthetics. When you need to build a world, you only need to speak softly, and AI will follow your word like law.
Friends, we have witnessed the fall of the “Tower of Babel” together, saw the twilight of Python as “Morse Code”; we also pushed open the door to the new world together, seeing how veterans with rigorous logic and liberal arts students with exquisite language sense complete a magnificent counterattack in the AGI era.
But this journey is not over yet.
If “Natural Language” and “Logical Thinking” are the “Technique” (Shu) with which we harness AI, then today, I hope to take you through one final cognitive leap, to explore the “Way” (Dao) hidden deep within the code.
In 2026, when “How” becomes extremely cheap, we must place all our bets on “Why” and “What.” This is not just a shift in programming paradigms; this is a philosophical adventure for human intelligence finding new coordinates in the silicon ocean.
Chapter 7: The Dimensional Leap from “How” to “Why”
I. The Twilight of Engineers and the Dawn of Philosophers
In the past thirty years, the core education system of computer science has revolved around “How” (how to implement).
Whether it is “Data Structures and Algorithms” in universities or “Master Java in 30 Days” in bootcamps, they are teaching you: How to sort with less memory? How to query databases faster? How to build a high-concurrency server?
This was completely correct at the time. Because computing power was expensive, memory was scarce, and networks were congested. Programmers had to be like micro-carving masters carving on a grain of rice, calculating meticulously, using extreme skill to compensate for hardware deficiencies.
But by 2026, the environment changed. Computing power is not only in surplus, but intelligence itself has become an infrastructure like water and electricity. When you say to AGI: “I want an e-commerce system capable of handling ten million concurrent users,” it will deploy a perfect architecture based on Serverless for you in seconds, automatically completing load balancing and disaster recovery backups.
After such a “technological singularity,” the “How” that was once seen as a heavenly book instantly depreciated into a basic, even freely available resource.
At this time, if you are still obsessed with studying “how to write a bubble sort algorithm,” you have made a strategic error. It is like in modern warfare, you are still practicing “how to sharpen a bayonet,” while the opponent is already sitting at the console pressing the launch button for a drone swarm.
II. Returning to First Principles: Entropy Reduction of Value
So, what should we focus on? As a risk control veteran, I am used to thinking with “First Principles.” Stripping away all the fancy code, architecture, and algorithms, what is the essence of software engineering and business models?
The essence is just two words: Entropy Reduction.
The world is entropy-increasing (tending towards chaos), while human need is to seek order.
We developed Didi (ride-hailing) to reduce the chaotic time of “waiting for a car by the roadside”;
We developed WeChat to reduce the friction loss of “information transmission”;
We build risk control systems to reduce the uncertainty of “trust establishment.”
Programming is essentially a means to “counteract entropy increase,” not the end.
In the AGI era, whoever can counteract entropy more efficiently is the winner. If writing 1000 lines of Python code solves the problem, that is a means; if saying one sentence to AI solves the problem, that is also a means. And the latter is a thousand times more efficient.
Therefore, the super programmers of the future must complete the leap from “focusing on implementation details” to “focusing on the value loop.” You need to constantly interrogate yourself with three questions, which constitute the new “Ultimate Algorithm”:
1. Why: Is this need a real pain point or a pseudo-need? How much entropy (time, money, emotion) can it reduce for the user?
2. What If: If I combine the model of Domain A with the data of Domain B, will new value emerge? (For example, combining “Veteran’s Risk Logic” with “AI’s Computing Power”).
3. So What: Even if it is made, what irreplaceable “soul” does it have in this red ocean flooded with AI?
When you start thinking with these three questions, you are no longer a programmer; you have evolved into a product philosopher.
Chapter 8: Future “Programming” is the “Engineering of Philosophy”
I. Thought Experiment: “One Sentence Prophecy” in 2030
Let us fast forward the timeline a bit to 2030. The development tools (IDE) of that time might have disappeared, replaced by a simple dialogue box.
You say to it: “I want to make a mutual aid platform for empty-nest elderly in the community. It should feel like the courtyard life of the 60s, full of human touch. People can exchange skills, like Uncle Wang can fix light bulbs, Aunt Li can cook braised pork. Points are not called points, they are called ‘Favor Coupons’.”
Just this one paragraph. The system will automatically generate the App side, the Mini-program side, the backend management system, the point clearing logic, and even automatically generate promotional posters and community conventions in the background.
In that era of “One Sentence Prophecy” (where words immediately become reality), the difficulty of the act of “building” itself returns to zero. So, where does human competitiveness lie?
It lies in “Defining the Problem” and “Value Judgment.”
II. Value Judgment: The “Good” That AI Cannot Calculate
AI is a master of utilitarianism. If you ask it to design the “most efficient nursing home,” it might design a place like a chicken farm, where the elderly are intubated, data metrics are perfect, but there is zero dignity. Because it calculates “efficiency.”
But you—a human who has read “The Analects,” read “Les Misérables,” and experienced birth, aging, sickness, and death—you will say to the AI: “No. Efficiency is not the first priority. Dignity is the first priority. The design I want, even if it wastes a little space, must leave a public area where the elderly can sunbathe and play chess. Even if it increases management costs, they must be allowed to keep pets.”
This is the “Engineering of Philosophy.” Future programming is no longer writing `if (age > 60)`, but injecting your values, your ethics, and your aesthetics into the AI’s decision model. You are using your thoughts to “legislate” for this powerful silicon-based behemoth.
III. New Persona: The Shepherd of Intelligence
Therefore, I prefer to call the new generation of developers after 2026 “Shepherds of Intelligence.”
Under your command are thousands of AI Agents; they are your flock. Some are good at writing code, some at drawing, some at data analysis. They are energetic, even a bit crazy. And you, holding the whip of “Logic” and the sheepdog of “Philosophy,” guide them in the right direction. If your direction is wrong, the faster the flock runs, the greater the disaster. If your direction is right, you alone are an army of thousands.
In this sense, a liberal arts student who understands history, humanity, and has compassion, or a veteran who has rolled through the mud of this complex society, is obviously more suitable to be this “Shepherd” than a fledging science student who only knows how to solve algorithm problems.
Conclusion: The Finger Pointing at the Moon, Is Not the Moon
Friends, this long article is coming to a close. Reviewing these messages, we talked about the fall of Python, the rise of logic, the revenge of liberal arts majors, and finally the value of philosophy.
A thousand words can actually be condensed into a Zen saying from the “Platform Sutra of the Sixth Patriarch”: “The finger pointing at the moon is not the moon.” The finger points to the moon to let people see the moon. If a person stares dead at the finger and forgets to look at the moon, then he is a fool.
In the long river of computer history:
Assembly language was a finger;
C language was a finger;
Python was a finger;
Even today’s Prompt is a finger.
They are just tools pointing to that ultimate goal. What is that ultimate “Moon”? The “Moon” is the problem in your mind that you want to solve, the desire to create surging in your heart, the thought that you want to make this world a little bit better.
In the past, because the finger was too clumsy (code was too hard to write), we had to spend 90% of our energy decorating the finger, so much so that we often forgot to look up at the moon.
Now, AGI has made this finger incredibly dexterous, even making it transparent. The obstacle is eliminated.
So, for the last time, as a veteran, I issue the most sincere appeal to you in front of the screen—whether you are an anxious parent, a confused student, or a professional worried about a mid-life crisis:
First, please let go of the obsession with “tools.” Stop forcing yourself or your children to memorize those dry Python syntaxes, unless you are doing it for the pure fun of logical puzzle solving. Save that time.
Second, please pick up the honing of “thought.” Read logic, read history, read literature. Exercise your ability to explain a complex matter clearly (structured expression), exercise your ability to perceive the pain of others (empathy granularity). These things once considered “soft skills” are now the true “hardcore code.”
Third, please throw yourself into “real” life. Go to the vegetable market to bargain, go to the hospital to observe all walks of life, go to the factory to see the assembly line. AI has learned all the books on the internet, but it has no body, it cannot go to the scene. Your experience is your last line of defense against AI hallucinations, and also your strongest confidence in commanding AI.
Your mother tongue is the highest programming language. Your logic is the most rigorous algorithm. Your conscience is the lowest-level operating system.
This is the best of times, because technology is no longer the privilege of the few. This is the worst of times, because mediocre thoughts will have nowhere to hide.
I am the Financial Veteran from [FinSages]. I do not want to teach you how to become a more perfect screw; I want to accompany you to become the “Creator” who defines the soul of the machine.
On this ark to the future, I hope we travel together. Welcome to visit our website: finsages.org
See you next time.
