In the AI Era, Ability Is Cheap, Trust Is Precious

We of this generation are quite accustomed to believing one thing: as long as one is capable enough, opportunities will eventually come.

This logic is not flawed. Exams, degrees, papers, competitions, certificates, and portfolios are all, in essence, tools that convert personal ability into a public signal. Especially for young people without family backgrounds or social connections, a relatively standardized evaluation system once provided a very important channel for upward mobility.

But AI might be changing part of this.

When researching information, writing code, conducting analyses, generating proposals, and even completing quite complex knowledge work becomes increasingly easy, “Can I do it?” and “Will others dare to entrust the task to me?” are gradually becoming two different questions.

The former is a question of ability; the latter is a question of trust.

And what a young person just entering society often lacks most is not learning ability, but that initial bit of credit that can be recognized by others.

I. When Opinions Become Cheaper, Credibility Becomes More Expensive #

In recent years, I’ve become increasingly accustomed to not rushing to express an opinion when encountering an interesting phenomenon, but rather first seeking out the research and patterns behind it.

Upon seeing a news story, hearing an anecdote, or encountering something that evokes anger, excitement, or emotion, it’s natural for us to form a judgment: “I think this is how things are.” In the past, when information was difficult to access, such experiential judgment was sometimes the best one could do. But today is different.

The specific thing you encountered is very likely not an isolated incident, but just one sample within a certain category of problems. Sociology, economics, psychology, management science, and even computer science might have debated similar issues for decades. The truly interesting question thus shifts from “How do I see this matter?” to “Why do these kinds of things happen repeatedly?”

What mechanism is at play here? Under what conditions does it occur? Under what circumstances does it not? Are there counter-examples? Are there overlooked costs?

Following these questions often reveals that although the real world is extremely complex, the recurring structures behind many phenomena are not as numerous as one might imagine. Goodhart’s Law, incentive misalignment, information asymmetry, principal-agent problems, path dependence, unintended consequences, collective action dilemmas… many seemingly new social news stories, after a change of superficial appearance, may still be repeating some ancient problems.

Of course, this doesn’t mean “there’s nothing new under the sun.” The truly interesting part lies precisely in the fact that some new phenomena can be well explained by old theories, some only partially, and others indeed expose what existing theories cannot cover.

After AI greatly reduced the cost of researching information, comparing viewpoints, and tracking literature, I actually feel that the demands on writers should be higher. Not because every article must be a research paper, but because simply relying on intuition to declare a grand conclusion has become increasingly cheap.

Opinions are cheap; what’s truly valuable is: why you believe that opinion.

This is, in essence, a form of credit.

II. Goodwill, Ability, and Intelligence Cannot Automatically Translate into Trust #

Another thing that is easily confused is goodwill.

We often feel that as long as a person’s intentions are good enough, their actions should be worthy of support. But many organizations in reality tell us precisely this: goodwill and reliability are not the same thing.

David Brooks wrote about individuals who dedicated themselves to public service in The Road to Character. One deep impression that book left on me was: what truly sustains an endeavor is often not some grand passion of “I want to change the world,” but rather the ability to consistently handle a large number of琐碎 (trivial) tasks. What to do if funds are insufficient, who audits the accounts, how a certain person receives treatment, who prepares today’s meal, who replies to letters, who ultimately handles a small problem no one wants to deal with—these things may not seem grand, but they constitute the majority of the real world.

A person can be very kind, but not necessarily reliable; very enthusiastic, but not necessarily able to finish tasks. Sometimes, over-reliance on emotion and goodwill can even lead to resource misallocation. Disasters, illnesses, or animals that easily evoke empathy naturally receive more attention; other problems, which are equally worthy of help but have less strong communicative impact, may remain out of sight for a long time.

Therefore, modern society cannot rely solely on individual goodwill to allocate resources. We need budgets, audits, rules, professional institutions, and systems that can transcend individual emotional fluctuations. The problem is not that goodwill has no value, but that goodwill must pass through responsibility, ability, and systems before it can become actionable and entrusted.

The same applies to AI.

Imagine a sci-fi scenario: a company consults AI before every decision, and the AI’s advice is proven correct time and again. Then one day, the AI suggests that for the company’s long-term development, the CEO should be replaced, and this suggestion is seen by everyone in the company.

What happens next?

This question doesn’t really touch upon whether AI will “seize power,” but rather: why do we grant power to a particular entity?

If it’s merely because “it’s smarter,” that’s insufficient. What’s more likely to happen in the future is not that one company suddenly acquires an almost omniscient and omnipotent super AI, but that everyone has AI. Intelligence is increasingly becoming infrastructure, much like electricity, search engines, or cloud computing.

More importantly, many truly difficult organizational decisions inherently lack a single correct answer. Should quarterly profits be increased, or should long-term R&D investment continue? Should staff be cut to improve efficiency, or should safety redundancy be maintained? How should shareholder returns, employee stability, innovation speed, and systemic risk be balanced?

These certainly require calculation, but they also involve value prioritization, responsibility bearing, and interest distribution. AI can provide analysis, but it cannot decide for everyone what is most valuable.

So whether it’s goodwill, ability, or intelligence, none can naturally acquire power.

What’s truly scarce is the qualification to bear responsibility.

And the qualification for responsibility is, in essence, long-term accumulated trust.

III. The First Bit of Trust Is Often Not Earned #

This brings us back to the problem of young people.

If the premium on ability itself gradually declines in the future, and reputation, connections, a track record of responsibility, and the qualification to vouch for others become increasingly important, then what should a recent graduate do? They have no assets, no industry reputation, no decade-long project history, and perhaps no opportunity yet to prove themselves.

This is not a problem that only emerged in the AI era. Human society has always had a troublesome credit cold start problem.

Historically, the most common solution was actually quite simple:

The first bit of trust is often not earned, but borrowed.

Whose child are you, who did you study under, which teacher recommended you, who did you work with previously, is there an acquaintance willing to introduce you—these things that seem like “connections” essentially serve some form of credit guarantee function.

Today, when recruiting, looking at schools, degrees, and internship companies actually involves similar logic. The name of the school, in a sense, tells the employer: at least one screening system that we generally recognize has verified this person.

Of course, such guarantees are not perfect. Students from prestigious schools are not necessarily outstanding, and ordinary schools also have many highly capable individuals. Degrees, workplaces, and recommendations can all lead to biases and even perpetuate unfairness.

But from an organizational perspective, they at least attempt to solve a real problem:

Facing a complete stranger, why should I trust them?

This is also why the term “关系” (connections) should not be understood solely as “getting ahead through improper means.” Connections also have a very important social function: reducing information costs and credit risks.

IV. What’s Truly Valuable About Connections Is Someone Willing to Put Their Name Behind Yours #

Someone who has worked with you for a long time will usually have a much more accurate judgment of you than a forty-five-minute interview.

They know when you’ll procrastinate, whether you’ll conceal problems when they arise, whether you’ll make excuses or actively remedy mistakes after a failure, and whether you’ll still do a good job when unsupervised. This kind of information is often difficult to obtain through resumes and interviews.

More importantly, connections are not just an information mechanism, but also a responsibility mechanism.

When someone introduces you to others, they are actually assuming part of the risk. If they tell others: “I’ve worked with this person, you can entrust tasks to them,” they are using their own years of accumulated credit to vouch for your as-yet-unverified parts.

So the truly valuable aspect of a professional connection has never just been “they know me,” but rather:

“They are willing to put their name behind my name.”

These two things are completely different.

Many young people in the early stages of their careers tend to interpret “building connections” as meeting more people, adding more WeChat contacts, and attending more dinner gatherings. But truly valuable professional relationships usually don’t start with acquaintance; they start with collaboration.

Having done one real thing together is often more important than meeting ten times.

Because only through real collaboration can a judgment of a person’s reliability emerge.

V. Trust Grows Through a Series of Small Commitments #

Ideally, a young person’s professional credit would go through a process like this: initially, teachers, schools, friends, former colleagues, or an organization lend you a small portion of their credit. You then get an opportunity.

This opportunity usually won’t be particularly grand. It might just be organizing a dataset, writing a module, leading a small event, completing an analysis, resolving a customer issue, or maintaining an inconspicuous open-source project.

What’s truly important is not how dazzling the task is, but that it forms a very clear commitment:

If this task is given to you, can you complete it?

Succeed once, and next time, others might entrust you with a bit more. Succeed again, and the boundary of responsibility continues to expand. Gradually, a person moves from executing tasks to being responsible for projects, from being responsible for projects to bearing judgment, and then from bearing judgment to bearing outcomes.

This is how credit grows, bit by bit.

Moreover, being “reliable” doesn’t mean never making mistakes. Things in the real world naturally fail. What truly impacts credit is often how a person handles failure when it occurs: did they explain it early, did they cover up the problem, did they actively remedy it, did they clearly summarize the reasons?

The same failure can leave completely different credit records for different people.

So what young people truly need to seek in the early stages of their careers is usually not an opportunity to “achieve instant fame,” but rather a series of small commitments that are clearly defined, have manageable consequences, and verifiable outcomes.

You don’t need to prove you can carry a mountain from the start.

First proving that the stone others entrusted to you won’t be lost might be more important.

VI. What if You Don’t Have a Prestigious School or Connections? #

There’s certainly a more realistic question here.

What if a young person graduates from an unknown school, has no social resources from their family, and no one around them who can lend them that initial bit of credit?

A very important answer provided by traditional society is examinations.

“Ability” has a big advantage: to a large extent, it’s something that doesn’t require permission. You can study on your own, practice on your own, prepare for exams on your own, and then through postgraduate exams, civil service exams, professional qualification certificates, or other standardized evaluations, convert your private effort into public credit.

This is also a very important layer of social value of the examination system. It is certainly not perfect, but it at least provides a pathway for a stranger to bypass private connections.

Will AI weaken this mechanism? It’s hard to conclude right now.

Perhaps some traditional exams will lose their discriminatory power due to AI, but on the other hand, a person’s ability to learn consistently, understand complex concepts, complete systematic training, and persevere to the end still says something about them, such as conscientiousness, self-management, comprehension, and sustained commitment.

So while future exams may change, the demand to “convert private effort into public credit through some public mechanism” will not disappear.

What’s truly worth considering is, if AI makes many abilities increasingly easy to acquire, and the credibility of traditional evaluations gradually declines, then what should society use next to accomplish this conversion?

Perhaps the answer is: gradually moving from “unapproved learning” to “unapproved contributions.”

VII. From “Unapproved Learning” to “Unapproved Contributions” #

Credit has a problem: it cannot be completely built up by oneself at home.

You can learn to code alone, but you cannot independently prove yourself to be a trustworthy programmer. You can read many management books, but only by actually working with others will people know what you’re like under pressure.

So the most important step in cold-starting credit is to put ability into the real world.

Internships are one way, community volunteer activities are another, student projects are a third, and open-source collaboration offers an especially interesting example.

In a public project, you submit code, modify documentation, reply to issues, participate in discussions, and fix bugs. None of these things are grand individually, but they leave a record. Others can see what you’ve done, whether your contributions were accepted, how you communicated when disagreements arose, and whether you were still there six months later.

At this point, ability slowly begins to transform into a form of credit that can be verified by a third party.

A merge record on GitHub, a long-term stable contribution in a community, a recommendation from a real project—these might all be more informative in the future than a line on a resume saying “proficient in X technology.” Because the former doesn’t just state “I know how,” but rather:

“I have successfully done things in a real-world environment.”

However, this problem cannot be entirely shifted onto young people.

If we simply tell young people that they should build a personal brand, participate in open-source projects, get to know industry veterans, and accumulate credit themselves, that’s still turning a structural problem into individual responsibility.

Every industry needs new blood. Since new blood is needed, some “yet-to-be-proven people” must be allowed to enter. Otherwise, the system will eventually fall into a paradox: inexperienced people don’t get jobs, and because they don’t get jobs, they never gain experience.

Therefore, a healthy society should provide some form of “public infrastructure for first credit.”

In the past, this infrastructure might have been the imperial examination system; later, it became schools, exams, professional qualifications, internship systems, and graduate programs; open-source communities are also one such form. In the future, new forms may emerge: real task platforms, public collaboration records, community contributions, short-term projects, peer evaluations, and even some verifiable human-AI collaboration records.

What’s truly worth designing in the AI era might not be more and more “ability tests,” but rather more environments like these:

Where a person without a background can do a real thing with very low risk; and after the task is completed, they can receive a record recognized by others.

This might be more important than adding another exam.

Conclusion: What’s Truly Scarce Might Be “Trustworthiness” #

For the past few decades, young people have often been encouraged to accumulate knowledge, enhance skills, and obtain better academic qualifications.

These things are still important.

But AI is reducing the scarcity of some of these abilities. More and more people can write code, more and more can write reports, and the cost of quickly learning a new skill is getting lower. Even “intelligence” itself might become less scarce than in the past, precisely because we can always call upon a very intelligent AI.

At this point, another set of qualities may re-appreciate in value: who is willing to work with you long-term, who is willing to recommend you, whether you actually completed tasks entrusted to you, whether you are the person who stays on site when problems arise, and whether you have the ability to fulfill a commitment, rather than just showcasing an ability.

From this perspective, what’s most worth accumulating in the early stages of a young person’s career might not be one huge success, but a series of small but clear records:

I promised, and I delivered. The next task can also be entrusted to me.

The first bit of trust might be very small, perhaps even borrowed from someone else.

But if such instances repeat often enough, borrowed credit will slowly transform into one’s own track record, and that track record will slowly become a reputation.

Until one day, people no longer primarily ask where you graduated from, nor who introduced you.

Your name itself will have become a guarantee.