What AI Can Learn Will Never Be Your Additionality

For the past thirty years, the career advice Taiwanese parents gave their children has been much the same: study a good major, get certified, join a good company, start at the bottom and slowly climb.

The premise of this script is that the entry-level positions will always be there, waiting for young people to learn and endure.

This year, the script started to change.

In a study updated by Stanford University economists this August, using anonymized payroll data from the US payroll provider ADP, researchers found that in the occupations most exposed to AI, employment of 22-to-25-year-olds is now 19% lower than in less-exposed occupations; a year earlier the gap was 13%. The researchers note that the gap comes mainly from companies hiring fewer people, not from layoffs.

The feeling in Taiwan is just as real. A survey this year by the Taipei City Department of Labor and 1111 Job Bank found that workers rated their worry about being replaced by AI at an average of 6.29 out of 10, and expect their work to be noticeably affected within 2.56 years on average.

The question really worth asking is a different one: if the bottom rung of the ladder is getting shorter, what can make someone the person without whom things would not happen?

One study, two kinds of knowledge

There is a distinction in this study that made me stop when I first read it.

The researchers divided occupations by the knowledge they require. One kind is codified knowledge, which can be learned through textbooks, classrooms and written procedures. The other is tacit knowledge, which usually accumulates only through experience, mentorship and repeatedly facing real situations.

They found that the more an occupation relies on codified knowledge, the slower employment growth has been for young workers; the more it requires tacit knowledge, the stronger employment growth has been for mid-career and senior workers.

In other words, what textbooks can teach, AI can learn too.

In the language of impact investing, we use additionality to test a sum of money: without this investment, something would not happen. My own extension is that people can test themselves by the same standard.

Skills many people have are where involution begins; skills AI also has don't even qualify for that race. A person's additionality increasingly lies outside the textbook.

The most valuable things I have were not taught in textbooks

Looking back on my own path, this distinction is especially clear.

I studied mathematics at university. Mathematics gave me rigorous training in logic, but if I had relied only on that diploma, I would probably have been just one of many analysts in finance.

What really set me on a different path were the three years I spent doing microfinance in rural areas after 2008. Small loans often serve families with no financial statements and no collateral. Judging whether a loan can be made and whether the money will come back depends on understanding people, communities and livelihoods. No textbook can fully teach you that.

Later I had the chance to join the Temasek group in Singapore to work on SME investment and financing, and then to take part in transforming Anue (cnYES) in Taiwan from a financial media company into a financial group. Each time I was approached, what mattered was not what I could calculate, but that I had been somewhere very few people had been, worked there, and fallen down there.

What others cannot copy has never been what you studied, but where, and at what cost, you have lived through something.

Three sources of additionality

Putting these experiences together, I believe a life's additionality has roughly three sources.

First, intersections. The intersection of mathematics and finance, of finance and rural communities, of media and trading: places that few people from either circle cross into are where "without you, the two sides won't connect" positions are most likely to appear. My role at the association these past few years is similar: sitting in many meeting rooms with different positions, and translating between the languages of finance, families, corporates and the public sector.

Second, time. Being willing to stay long enough in a place others don't want to be. Tacit knowledge cannot be rushed; it is built up by time. This is also what AI finds hardest to replace: it can read every textbook in a second, but it cannot live in a village for three years on your behalf.

Third, commitment. Additionality usually comes with a cost: a pay cut, leaving a familiar environment, entering a field with no certainty. The cost itself is not the goal, but it is a threshold, and precisely because there is a threshold, that position is not crowded.

Intersections make you scarce, time makes you deep, and commitment makes you credible.

To be clear: this is not dismissing textbooks or popular majors

I must rule out a misreading. I am not saying what schools teach is useless, nor that young people should avoid AI-related majors.

The same study also notes that higher education still offers some degree of protection. Textbook knowledge is the foundation; without it, tacit knowledge has nowhere to grow. The problem is not what you learned, but stopping at what you learned.

Another limitation must be stated honestly: additionality comes with no guarantee. I can write about these experiences because those choices later worked out; the same risk might have brought only hardship that no one ever saw. And additionality must rest on "someone needs it." Doing something unique where no one needs it is self-admiration, not additionality.

Additionality is not a strategy that guarantees returns, but a way of asking questions that makes one's presence more meaningful.

For the next generation: not answers, but questions

In recent years I have often had the chance to talk with second- and third-generation members of families, and with young people just entering the workforce. The question parents ask most often is: "What is the safest thing to study?"

In the age of AI, that question may no longer have a standard answer. A better question might be: "Where are you willing to stay a little longer than others?"

For families with resources, the best inheritance may not be paving the road for their children, but giving them time and room to "walk into places few people go": an unfamiliar city, an unfamiliar industry, a place that needs them but may not pay them well. Three years of tacit knowledge often goes further than a diploma.

Looking back on my own road, many of its turns were less something I calculated than something I was led through. But every time, it began with being willing to walk into a place few people went, and only then did the doors open one after another.

AI will be able to do more and more; that much is certain.

What is uncertain is something else: in an age when everything can be generated, which part of your life is something that would not have happened without you?

The author is Chairman of DoublePortion Capital, and also a Partner at SIC (Sustainable Impact Capital), Executive Director of the Taiwan Impact Investing Association, Executive Director of the Cross-Border Innovation & Entrepreneurship Association, and a deacon at Taipei Bread of Life Church.