Educating for the Generative AI Era: A Few Thoughts
Generative AI will turn everyone into a manager of AI assistants. This post proposes five abilities that education should focus on: expression, logical thinking, reading comprehension, judgment, and just-in-time learning.
Generative AI is having a big impact on every one of us, and the trend is unavoidable. In the end it will become part of our life and work, like the internet and the smartphone did. People call it the next industrial revolution. Whether or not that label fits, it raises a real question for education: what abilities should we build so that people can adapt to the Generative AI era, and how should education change?
After using AI every day, I found that the result I get depends a lot on abilities that have nothing to do with coding: how clearly I can say what I want, whether I can break a task apart, and whether I can tell a good answer from a bad one. This post puts those observations into a short list.
Everyone becomes a manager
Picture the near future. Each of us has one or more AI assistants that help us at work and in life. We hand tasks to them and they hand results back. Whatever our job title is, each of us becomes a manager.
Managing well is not about doing the work yourself. It means:
- saying clearly what you want;
- breaking a big task into parts and giving each part to the right assistant;
- reading what the assistants hand back;
- deciding whether the result is good enough and correct;
- learning along the way whatever else you need to know.
Each of these points to an ability worth building.
Five abilities for the Generative AI era
1. Expression
Expression decides how well you can communicate with AI. The assistant can only work from what you tell it, so the quality of the result is limited by the quality of the instruction. Describing clearly and completely the result you want, and the work you want the assistant to do, is the first skill of a good manager. However strong the assistant is, a vague request can only get a vague result.
2. Logical thinking and problem solving
Suppose you have a task and want several AI assistants to work on it together. First you have to break the task down: what parts does it have, which parts depend on each other, and which assistant is best for each part? Only then can you hand out the work and coordinate the results. Breaking down a task takes thinking, not tool skills. If you have not worked out the structure of the problem yourself, no assistant can do this step for you.
3. Reading comprehension
AI assistants produce a lot of information, far more than a person would write by hand. The bottleneck moves from producing content to absorbing it. You need to go through a lot of text quickly, pick out the key points, and then make a decision.
Reading belongs in the same category as expression. Expression is what you send to the assistant, and reading is what you take back from it. Together they make up the communication loop you work in every day.
4. Judgment and evaluation
When an assistant hands back a result, we must be able to judge it: does it meet the requirements, and is what it says correct? AI output can read smoothly and sound confident while being wrong or off target. The person is still responsible for the outcome, so the ability to evaluate results cannot be outsourced.
5. Learning ability: just-in-time learning
Here we need to change how we think about knowledge. Traditionally, educational institutions have put most of their effort into teaching students domain knowledge. I think that role will matter less in the future, because what we should promote is just-in-time learning: when a task arrives, first understand the task, work out which related domain knowledge is needed and which key points you must understand to do it well, and then learn only those. It also helps you communicate better with the AI assistant that works on the task.
This makes the ability to learn fast very important. You must be able to grasp the basics of an unfamiliar field in a short time, and the main way to do that is to ask good questions. Knowing how to ask an AI assistant, and how to find the key points through the back-and-forth of questions and answers, is an ability in itself. If you ask well, you get the domain knowledge a task needs, right when you need it.
What this means for education
If these five abilities are the key, education should shift its focus toward them:
- Expression: practice writing precise, complete instructions, and compare the different results that different instructions give.
- Logical thinking: break real projects into parts and decide who or what should do each part.
- Reading: under time pressure, summarize a large amount of text and pull out the points you need to decide.
- Judgment: review AI-generated answers critically, find what is wrong or missing, and explain why.
- Learning: pick up an unfamiliar field in a short time by asking an AI assistant questions, and show what you learned.
These are only examples, but they show the direction: spend less time teaching knowledge for its own sake, and more time building the abilities that let a person direct, evaluate, and learn.
The last two abilities pull against each other. To judge the quality of AI output, you usually need some understanding of the topic. If all you have is something you learned a few minutes ago, you may not spot a subtle error. So I do not think domain knowledge disappears from education. A solid base is still what makes judgment possible, and just-in-time learning also works best when it builds on that base rather than starting from zero. What changes is the balance: less stress on memorizing what can be looked up, more on understanding things well enough to question and verify them.
Closing
None of these abilities is new. Clear writing, structured thinking, careful reading, good judgment, and curiosity have always been valued. What the Generative AI era changes is their weight: when assistants can take over the actual work, these abilities are what is left to us, and how much we get out of AI will depend on how well we have built them.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article