Protect Students and Teachers, Not the Old Way of Education
Illustration · not a service recordIf this AI boom ended next year, what would Taiwan be left with? The Singapore prime minister’s answer made me think of Taiwan’s classrooms.
On the evening of October 8, Singapore’s Prime Minister Lawrence Wong sat down with Steve Forbes at the Forbes Global CEO Conference. Forbes asked him: how much longer can Singapore’s AI boom hold up?
The first sentence of his answer was: “No boom is forever.”
He said a correction always comes. It is just that no one knows when or how it will happen, or who will get hurt.
What makes the remark interesting is when he said it. Singapore has grown about 5% a year for three straight years, with productivity growth above 3%. For a country with per capita income this high, those results are almost too good to be true. To remind everyone, unprompted and in the middle of good times, that good times end is a rare kind of honesty.
Then I looked at Taiwan’s numbers. According to the central bank’s September report, Taiwan’s economy grew 14.15% in the first half of this year, and the full-year forecast was raised from 9.45% in June to 11.48%.
A country growing at 5% is already preparing for a correction. What about us, growing at 11%?
(Friends who bought shipping stocks in 2021 probably understand best what “No boom is forever” means. I won’t ask, and you don’t have to tell.)
1|Lawrence Wong actually said three things
After reading the transcript, I think what he said about AI can be broken into three layers.
The first layer: the boom is real, but it is also temporary. He said frankly that Singapore does not make the most advanced AI chips. It relies on major makers of memory and specialty chips, plus an entire semiconductor supply chain. He did not describe this boom as proof of national strength. He described it as “external demand that happened to arrive.”
The second layer: a window is something to trade for capability, not something to enjoy. He said Singapore’s approach is to make good use of this window while external demand is still there, to build new capabilities and create better jobs for its people. The key word is “trade”: turning money that is here for a moment into something that will still be there in ten years.
The third layer, and what I consider the most important line of the whole session:
Not to protect the job, but to protect every worker.
Do not protect the job; protect every worker.
He acknowledged that AI’s effect on employment is still in its early days, but some jobs will certainly disappear and many more will have to change. Singapore’s choice is not to block change but to give people the ability to get through it. They treat unions as partners, not obstacles. Unions help administer the government grants for company transformation, on the condition that companies help retrain their employees. Layoffs can only be a last resort, and they require advance notice and compensation, after which the union stays with the workers until they find new jobs.
There is one more passage that is easy to skip over. Speaking about AI safety, he said regulators in every industry have to answer the same question: which decisions must stay in human hands? Where in the process should a person stand? I will come back to this question shortly.
2|Translating that line into education
When I read “not to protect the job, but to protect the worker,” another line came to mind on its own:
Not to protect the old way of education, but to protect the students and the teachers.
Do not protect the old way of education; protect the students and the teachers.
To be clear, the old way of education is not a bad thing. Standard answers, uniform pacing, exam rankings: this system worked very well for decades, and many of the engineers holding up this AI boom were taught exactly this way.
But it is like that “job”: having once been good does not mean it should be protected forever.
What does protecting the old way look like? The moment AI appears, the first reaction is to discuss how to ban students from using it. Effort goes into keeping to the schedule rather than making sure children have really learned. Teachers spend large amounts of time filling in forms and running through procedures, only because “this is how it has always been done.”
Then what does protecting students and teachers mean? It has to be told in two halves.
3|Students are the future
Taiwan had 208,000 newborns in 2016. In 2025 there were only 108,000, a new low for the tenth year in a row. In less than ten years, the number fell by half.
This means each child’s future now carries more weight for society as a whole.
Officials at the Directorate-General of Budget, Accounting and Statistics (DGBAS) have also noted that over the past two years AI-related industries have performed brilliantly while traditional industries have clearly lagged. The economy is splitting into two speeds. If we are not careful, education will split into two speeds too: children who know how to use AI as a tutor run faster and faster, and children whom no one has taught how to use it are left where they are.
Protecting students does not mean keeping them away from AI. It means making sure every child has the ability to use it well. And what needs protecting is not only their scores, but the drive to keep learning when they face a world with no standard answers.
4|Teachers are the catalyst of the future
How much high school chemistry do you still remember? (I know many people gave it back to their teachers long ago.)
The definition of a catalyst: it makes a reaction happen faster without being consumed itself.
Teachers are the catalyst of education. But look at what the start of this school year was like on the ground: across Taiwan, elementary and junior high schools opened 6,464 positions for regular teachers and actually hired only 5,676. As of August 25, 614 substitute teacher positions still could not be filled, for reasons including pay, administrative burden and working conditions.
Our catalyst is being consumed. Something that gets used up in the reaction is no longer a catalyst.
Back to Lawrence Wong’s question: which decisions must stay in human hands?
In education, my answer is MCP. (Yes, it shares a name with the MCP that AI engineers talk about every day. That is on purpose.)
Mentor: knows who the child is, and cares about what kind of person the child will become.
Coach: sees where the child is stuck, and practices alongside until the child gets it.
Pacer: judges when to give a push and when to wait a little.
Think about it, and many of the best teachers have always taught this way. AI will be able to share more and more of the work of passing on knowledge. But these three things, only a person can do.
5|Two things we are doing
Over the past few years Junyi Academy has been answering the same question: how to make AI a help to students and teachers, rather than something that replaces them or wears them down. We are currently doing two things.
First, building a training and learning platform for AI literacy. We break AI literacy down into four words, Ask, Use, Manage, Build (問用管造), so that teachers and students alike can learn step by step to work together with AI. This content is now being used even in Singapore. Against the “two speeds” I mentioned earlier, this is a bit of leveling that we can do.
Second, continuing to refine our education AI, and asking a third party to study whether it actually works. Our AI tutors for math, science, Chinese and English have scaffolding, and we are also strengthening the guardrails. Scaffolding lifts a child up one step at a time instead of handing over the answer; guardrails are there to block what should not happen.
What many parents and teachers worry about most is this: will children who use AI become lazy and stop being able to think? The third-party study found that on exams where AI could not be used, students’ scores did not drop, and their ability to ask questions actually improved.
In the age of AI, being able to ask good questions may matter far more than being able to memorize answers.
6|Two things we need the government to do with us
Some things remain scattered points no matter how much the private sector does; only when the government joins in can the points become a whole surface.
First, from “a tablet for every student” to “Tokens for every teacher.”
Over the past few years the government has given students tablets to use, and that step was important. The next step is to give every teacher enough Tokens (think of them as an allowance of AI usage), so that chores such as grading, organizing and administrative work can be handed to AI and the time given back to students. Singapore lets unions help administer the grants for company transformation, on the premise that companies take good care of every worker. Education can follow the same logic: put resources directly in teachers’ hands, with the aim of bringing teachers back to teaching.
Second, finding, within teacher education, the teacher’s MCP role in the age of AI.
With knowledge work, AI will gradually be able to help. The focus of teacher preparation should also slowly evolve from “how to explain knowledge clearly” toward Mentor, Coach and Pacer. This is not asking teachers to become something else. It is turning what the best teachers have always done into abilities every new teacher can learn.
7|The window is still open
The window Lawrence Wong spoke of is one Taiwan has too, and ours is open wider than Singapore’s.
The question is what we will trade it for.
If we use these good times to hold on to the old way, then on the day the boom recedes (and it will), all we will have left is a system that is very familiar and no longer fits.
If we use these good times to invest in students and teachers, then when the boom is gone, the people will still be here. And people are exactly where the next boom grows from.
One is the future. One is the catalyst of the future.
One is the future; the other is the catalyst of the future.
No boom is forever. But what we are left holding when the boom ends is something we can choose.
- Original: Ray Lu’s personal Facebook post (published 2026-10-09)
Course titles follow the source. Check each course page for its teaching language.
Generative AI Leader Certification Learning Path
Five preparation courses are free and address business adoption and transformation. Course completion is not certification; the separate exam costs US$99 plus applicable tax. Its 90-minute duration is exam time, not course length.
AI capabilities and limitations
Understand next-token prediction, knowledge, working memory, and steerability to diagnose hallucinations and context limits. No technical prerequisites.