Manifesto · Intelligence Equity

English · 中文

A self-taught marksman may one day beat a champion from Tsinghua.

This is not a claim that ordinary people have already surpassed scientists. It is a claim that the rules of intellectual competition have already started to change.

To anyone without a lab, a degree, or a grant — but with a question they can't put down.

1. After the rifle arrived, the champion was no longer the only strong one

In the age of cold steel, a martial champion trained from childhood: winters and summers, eighteen weapons, decades of drills. A farmer, however hard he worked, had little chance in a straight fight.

Then the rifle arrived.

An ordinary shooter with modest training could, in certain situations and certain terrain, beat a man who had trained for thirty years.

The champion did not get weaker. The technology changed the rules of competition.

AI is doing something similar to the world of intelligence.

2. Intelligence used to be a resource privilege, not just a capability

Think about what it took for an ordinary person to do research:

Get into university. Then a master's, then a PhD. Find an advisor, enter a lab, apply for funding. Before you genuinely start research, a dozen years may be gone.

And even then: a very good idea, without a lab, a team, or money, is very hard to verify.

The old research system was, at its core, a highly organized mode of intellectual production. Universities, institutes, and large companies held the most important tools.

3. What AI removed is not intelligence. It is the cost of starting

Today one person can use AI to read a thousand papers, map the state of the art, derive formulas, write simulations, analyze data, and sketch research plans.

Work that used to need a small group can now be done by one person plus AI.

AI makes mistakes. Real experiments remain irreplaceable. But one change is irreversible:

The cost of starting intellectual production is falling, fast.

In the past, you had to enter an organization to get the tools to innovate. Now, you may be able to innovate first, and let organizations find you.

4. The moat around credentials is moving

Tsinghua, Peking, MIT, and Stanford will not lose their value. A great scientist's judgment, accumulated taste, and long experience still matter enormously.

What changes is where the advantage sits:

From "how much you know" to "what question you can pose"
From "how many people you have" to "how much effective intelligence you can organize"
From "who you are" to "what you have made"

Coding used to be an edge; AI puts it within reach of more people. English papers used to be a wall; AI climbs it for you. Interdisciplinary work used to require a team; now one person can scout first, then find the right experts.

A factory engineer who has spent twenty years next to real production may spot a problem that has blocked the industry for decades. In the past he had neither the modeling skill nor the resources. Today he can turn experience into hypotheses, hypotheses into models, and then find someone to test them.

He may not be smarter than a PhD. But he may be closer to the real problem.

AI narrows the knowledge gap — and may amplify the value of asking the right question.

5. Real equity is not "everyone gets to publish"

There is a line that must be drawn.

Science equity does not mean everyone becomes a scientist. It certainly does not mean every unverified idea deserves to be called a discovery.

Science is science because of evidence, logic, experiment, and reproducibility.

AI can lower the cost of proposing a hypothesis. It cannot cancel the responsibility of testing it.

What real equity looks like:

An ordinary person designs a new battery chemistry and can find a lab to test it
A technician improves a process and can prove it with simulation and production data
An independent developer writes a new algorithm and anyone in the world can run and reproduce it
A farmer without a PhD finds a planting method that raises yields, and can prove it with a proper experiment

Science should ask about your evidence, not about your identity.

How we evaluate will change too:

Yesterday: what degree, which institution, how many papers? Tomorrow: what did you solve, can it be reproduced, how much faster, how much cheaper?

This is a shift from credential evaluation to result verification.

6. The root of involution is too little innovation

A hundred companies making the same product, with the same technology, the same cost, the same marketing — the competition ends in price war. You cut 10%, I cut 20%; you work 12 hours, I work 14.

But if one of them halves the cost through technology, or invents an entirely new product, the rules change.

We keep fighting over the same pond. Innovation is someone learning to build a new pond.

What is most promising about AI is not saving a few salaries. It is giving individuals and small firms who never had R&D capacity a first chance to explore new technology, new products, new business models.

China's largest research potential may not sit only on university campuses. It is also hidden in millions of factories and industrial sites.

7. The one-person lab: a third kind of innovation organization

We already talk about "one-person companies": one person plus AI, doing product, engineering, marketing, support.

One step further: why not a one-person lab?

One person finds the problem and poses the hypothesis. AI handles literature, derivation, simulation, experimental design, data analysis. When real experiments are needed, call on external labs, factories, testing services, and specialists. Results come back; AI helps analyze; the plan iterates.

The future research entrepreneur may not need to own a large lab. He needs the ability to call on labs — just as today's founders don't build data centers, they rent compute.

Underneath this, a new infrastructure may grow:

AI research agents + an open experimental network + a system of verifiable results.

It may not replace universities. But it could become a third form of innovation organization, alongside universities and corporate research institutes.

8. An explosion of intellectual productivity

Every major technological revolution redistributes some kind of productive power:

Steam: ordinary people commanding mechanical force far beyond muscle
Electricity: huge energy entering factories and homes
The internet: ordinary people connecting to global information
AI: ordinary people gaining part of the intellectual capacity that used to belong only to large organizations

If the industrial revolution mechanized muscle, the AI revolution may be the scaling and democratization of intellectual labor.

What is exciting is not that smart people get smarter. It is that a large number of people who never had a chance to participate in innovation now hold the tools to do so.

A significant innovation may come from a rented room in an urban village in Shenzhen, from a technician in an ordinary factory, from a retired engineer, from someone who never finished higher education but has thought about one problem for thirty years.

They may not have impressive résumés. But they have problems, experience, curiosity — and increasingly powerful AI.

Having tools does not guarantee results. Those who go the distance still need rigor, patience, experiment, and time.

AI will not turn everyone into a scientist. It may stop burying the people who already are one.

Permission to begin

What ordinary people lack is not tools. It is permission to start.

So here are three permissions:

Ask the "basic" question. Nobody is born knowing what to measure. Ask it, then take the next step.
Let AI be wrong. It gives you a draft, not a conclusion. Verification is your job — that is exactly where your value is.
Let the experiment fail. Failure is a result. Write it down so the next person saves hours.

You don't have to jump into deep water. Start with a 30-minute experiment. Dare to press start once.

What this manifesto becomes in this repo

The manifesto says	The repo does
The barrier is "nobody tells you what's next"	17 experiments, each with a way to start and a way to know it worked
The responsibility to verify cannot be cancelled	Acceptance criteria written first; results public, failures included
Judge evidence, not identity	No certification, no scoring, no vetting who you are
AI as a tool, not a substitute	AI helps you pick experiments and filter candidates; your hands do the work

This repo is the executable version of this manifesto. It is small. It is a start.

Source and boundaries

Adapted from 贺去病's essay "AI 智力平权时代：民间神枪手，可能打败清华武状元" (published 2026-10-20, written 2026-10-19).

The reference to Academician E Weinan's remarks at the Guanghua Thought Leadership Forum on 2026-10-17 comes from public reporting. The extensions and judgments in this text are the author's own — not the academician's words.

Author: 贺去病, independent developer and practitioner of AI-native business consulting. Writes about AI newcomers, science equity, one-person companies, and the next form of human–AI collaboration.

Closing

What will future competition actually be about?

Credentials? Titles? Institutional scale? Or the ability to find a problem, solve it, and prove it?

Our bet: the last one matters more and more.

Top universities will still matter. Great scientists will still be precious. But they are no longer the only entrance to innovation.

In the past, to change the world you first needed permission to enter a system. In the future, you may build something with AI first, and let the world find you.

The champion will not disappear.

But the age of the marksman has arrived.