AI时代下的大学学习与个人成长

Why Universities Still Matter in the Age of AI

Fuchun Guo (郭福春)

我是一名80后,已经见证了人类文明史上的三次重要科技浪潮。第一次是2000年代的互联网普及浪潮,第二次是2010年代爆发的智能手机浪潮,第三次则是2020年代开始的人工智能浪潮。2026年,面对第三次浪潮,作为一名计算机学科的大学教师,我觉得再不动起来就要来不及了。

1. 科技进步的开心与烦恼

自2005年从福建师大开始,我已经研究了密码学超过20年,并幸运通过自我顿悟的方式懂得科研的背后是为了超越人类的认知极限和不断推动科技进步。科技既能为社会发展降本增效,又能改变每个人的生活,带来各种各样的便利。

然而,在享受科技带来便利的同时,我们也不得不面对科技浪潮下的变局与不确定性。这种改变往往是痛苦的,因为人类天生本能倾向于沿用熟悉的方法生活和工作。可当大浪来临时,逆水行舟,不进则退。继续划船已经不够,我们还必须换个方式划!

一个不会使用智能手机的人,很难独自生活在2020年代了。科技可以提高人类的整体生活水平,却无法让所有人同时受益获得幸福感。每一次重大科技变革都会创造新的机会,也会淘汰旧的方法和旧的岗位。无论哪个国家、哪个种族,都无法完全避免这一过程。因此,真正重要的并不是科技本身,而是我们能否持续适应科技带来的变化。只有适应变化的人,才能先获得生存与发展的机会,再去追求自己想要的人生。

那么,我们又该如何学习和成长,去适应这个不断变化的世界?这就涉及一个更基本的问题:大学为什么存在?

2. 大学存在的目的

“传授”与“考核”是现代教育体系中两个重要的核心机制。本文正确性的第一个假设是:在AI时代,这两大核心机制仍然存在。 大学存在的重要意义之一,就是承担传授与考核的功能。其目前主要实现方式是知识传授和知识考试,帮助学生掌握一项技能,并能够输出和使用该技能。

自2022年ChatGPT横空出世以来,人工智能的发展也才不过四年多但已经发展得异常强大。它不仅能够辅助知识传授,还能够生成、完成和批改许多传统形式的考试任务,甚至在某些方面可能比许多摸鱼的教师做得更好。因此,关于大学教育被AI取代、被AI革命的预测层出不穷。大学教育确实应该重构,改变“传授”与“考核”这两大核心机制的实现方式,也就是说,不再把知识本身作为唯一的核心重点。

然而,重构的大方向大家都知道,但如何提出具有可操作性的具体建议和做法却并不容易。因为任何具体方案都有缺点,都存在trade-off,而且被批评的理由还很难反驳。 我认为这种困难主要有三个原因:

  1. 目前AI的能力边界和缺点究竟在哪里,还不清楚。大学的传授与考核应该,或者说必须,能够弥补AI的不足。
  2. 改变传授与考核的内容并不难,例如如何向AI提问、如何利用AI解决问题。但如何进行系统性传授与系统性考核却很难。因为系统性传授新内容,往往意味着一门新学科的诞生。
  3. 船大不好掉头。即使进行大刀阔斧的改革,也必须循序渐进。因为所有改革最终都需要人来实施,而人在做事时会疲惫,会犯错,会情绪化,也会被本能左右。

“传授”与“考核”这两大核心机制,无论具体实现方式如何,其最终目标都是帮助学生构建系统性认知框架,掌握某项技能,并在遇到问题时有能力尝试自己解决,而不是动不动就交给AI。同时,能够利用这套认知和技能发现问题、解决问题,并改进解决问题的方法。关键词我想再说三遍:系统性认知,系统性认知,系统性认知。 本文正确性的第二个假设是:在AI时代,每个学生都需要拥有属于自己的系统性认知框架,掌握至少一项能够独立运用的技能,并具备识别AI局限、纠正AI错误以及在必要时超越AI的能力。 这是因为,即使在AI时代,学生仍然需要持续性学习,并通过新的形式接受assessment,而这些assessment最终仍然需要检验学生自身的认知与能力。

3.AI时代下的大学学习

老师传授的内容和学生学习的重点的确要改变,不再是“本门学科有哪些重要知识点”。未来会怎样,我并不清楚,但在2026年我能想到的实现第二个假设的方法是将传授的重点转变为“本门学科的系统性认知构建,及知识获取判断方法论”。 这种传授的重点就是帮助学生建立本门学科的认知框架,学会从AI获取知识,并能够判断AI输出的正确性、完整性和适用性。

帮助学生建立一套系统性认知框架并不容易,因为它不存在标准答案。每个人已有的知识、思维习惯、兴趣方向和实践经历不同,因此最终形成的认知框架也不会完全相同。这是本文正确性的第三个假设,即传授必须考虑学生最终掌握的认知具有差异性不存在标准化。

我能想到的更具体的传授方法,是让学生理解并掌握“打碎—创造—重组”这项系统性构建的核心技能。也就是说,真正的学习并非简单地获取知识,而是一个主动建构的过程:学生先将现有知识拆解为可理解的部分,再依据自己的经验和理解重新组合,最终将这些知识内化为自身认知框架的一部分。知识可以被传授和迁移,但属于自己的系统性认知框架,最终仍然需要自己亲手构建。老师更多是在分享,分享自己如何打碎现有知识,创造符合老师个人逻辑的组织方式,再重新组合起来,形成一套个性化的认知体系。

打碎本身就是一个缓慢的过程,因为不仅要把知识化整为零,还要掌握一套正确的打碎方法。而这套方法的建立,需要学会观察、模仿和经历失败,最终知道什么做法是错的,什么做法是低效的,什么做法是对的,慢慢形成自己的一套方法。剩下的事情,则是利用一套通用的交流方法,从AI或教师那里进一步批判式学习知识的细节。

这种改变对于传授的教师和学习的学生而言都很痛苦。因为传授的方法更难了,学习的过程更累了。无论教师还是学生,都需要变得更加主动,投入更多思考,而这本身就是一个不断对抗本能的过程。尤其是年轻学生,他们中的很多人本身就对生活与人生很迷茫。不仅要面对生活中的各种压力和本能的诱惑驱动,也会对未来的职业方向感到困惑,并怀疑掌握系统性构建能力的必要性。

至于考核方式的改革,当然也必须改变,但我暂时还不知道具有可操作性的落地方案。几百个人同时坐在同一个考场、回答同一张试卷的传统考核方式,大概终将走入历史。但不得不承认,这种考核模式的组织成本真是太低太受组织者欢迎了。

4. AI时代下的个人成长

无论是哪一种传授场所,都有一个隐形但重要的功能:为学生的个人成长和走向成熟提供良好的成长环境,以及试错训练的场所。

一个能够独立面对社会生活的的成年人,不仅需要建立一套系统性的认知框架用于持续性学习,也需要逐渐形成许多与个人成熟相关的能力与品质,例如领导力、审美能力、道德判断力、匠人精神以及生存智慧等。这些能力与品质可以通过教育获得启发和指导,但很难仅依靠知识传授与考核形成。它们更多是在个人的理解、持续实践、真实经历、不断试错以及长期复盘中逐渐形成,并最终转化为面对复杂现实的判断力和行动能力。

大学提供了一个特殊的环境。在这里,大量同龄人可以相互观察、学习、模仿和借鉴,从而促进个人成长。更重要的是,校园本质上是一个“慢环境”。人类的大脑就像一台超级省电的AI,许多与成熟相关的能力只能在慢火熬煮的过程中逐渐形成。

AI时代下,个人成长的好处与坏处都很明显。好处是,对于愿意成长的学生而言,可以利用AI进行更加系统的复盘,从而促进自身成长。坏处主要有两点。第一,学生之间可能越来越不愿意交流互动。因为AI往往能够在交流过程中提供即时的情绪反馈,认可我们的表达与价值,而人要做到这一点并不容易。在这种情况下,学生之间练习交流、试错以及建立社会关系的机会反而减少了。第二,在AI时代,外界的信息流极快,信息量巨大。大脑处理这些信息会消耗大量能量。长期处于高强度的信息输入中,也可能增加认知负担,进一步削弱主动学习和成长的意愿。也就是说,在AI时代,人类或许会变得更加本能化,而缺乏一种成熟的理性和生存智慧。

在AI时代,大学或许可以被重构,甚至可以消失;但让同龄人聚集在一起交流、试错,并提供一种慢成长环境的场所,似乎依然不可或缺。因为人类这颗高度节能的“大脑AI”,只能在实践中缓慢成长。 在这个AI加速一切的时代,我反而愈发相信“慢即是快”。真正的理解、判断与成长,往往无法被压缩,而这些恰恰是教育最珍贵的价值。

个人成长本身也是一个逐步走向成熟的过程。在这个过程中,人不仅需要系统性掌握各种技能,也需要逐渐形成自己的认知体系。这与建立一门学科的认知框架有着某种相似之处,都需要通过“打碎—创造—重组”的方式不断完善。 写到这里,我似乎顿悟了未来大学教育最困难的问题:如何像帮助学生走向成熟一样,帮助学生逐步建立属于自己的认知体系。但目前看来,全球大多数教育体系都无法直接帮助学生系统性地走向成熟,它们更多是在提供成长的土壤、实践的机会以及试错的空间,而真正的成熟仍然需要学生自己完成, 一步又一步慢慢来。

5. 面向未来,始于当下

从目前的发展节奏来看,虽然AI领域存在很多泡沫,但其未来潜力似乎依然巨大,甚至可能带来极其深刻的社会变革。AI冲击的是大学传统的“传授与考核”方式,而大学更深层的价值,是帮助人建立系统性认知,并在实践、试错和长期成长中走向成熟。AI可以帮助人类学习得更快,但不能代替人类成长得更快。也许,人类与AI长期并存的关键,不是让人类变得比AI更快,而是帮助人类重新认识“慢”的价值。 人类的传授与考核机制终究会找到新的实现方式,只是苦了我们这一代人。因为很多人还没有开始适应新的划船方式。

这是一篇短文,比学术论文短多了。虽然写得用心修修改改超过50遍,但缺乏充分的调查研究,也缺乏对未来进行准确预测的能力。这篇文章我还使用了好多正确性假设,所以这篇文章的观点不一定对。 然而,这里的”打碎-创造-重组“是我过去十几年的科研范式,而且看起来不算完全错误。

在这篇文章里,我更多想探讨 AI 冲击下大学教育目前面临的具体挑战,并抛砖引玉,与大家一起思考:面对新的浪潮,我们究竟可以怎样改变现有的教育方式,找到一些真正可以执行的“新划船方式”。相比宏大的方向性建议,我更希望这些思考能够最终落到大学及教师现在可以做什么,因为过于宏大的建议有时不仅难以执行,反而可能让人更加焦虑和无力。就目前而言,我能想到的只有两点:

  1. “本门学科的系统性认知构建,及知识获取判断方法论”是传授的新重点。
  2. 掌握“打碎—创造—重组”这项学习技能。

Talk is easy, comments are easier, and every proposal has its attackers and supporters 😂. 接下来,我能做的能改变的,或许就是思考如何利用这两个想法指导自己修改课程PPT,以及重新设计试卷中的考核内容了。

I was born in the 1980s, and I have witnessed three major waves of technological development in human history. The first was the wave of Internet adoption in the 2000s. The second was the smartphone wave that exploded in the 2010s. The third is the wave of artificial intelligence (AI) that began in the 2020s. In 2026, facing this third wave, as a university teacher in computer science and information technology at UOW, I feel that we must do something now.

1. The Benefits and Downsides of Technological Progress

I have been doing research in cryptology for more than 20 years since 2005. I have been fortunate to understand myself through research that scientific research is to go above and beyond the limits of human knowledge and to continuously push technological progress forward. Technology can reduce costs and improve efficiency for society, while also changing everyone's life and bringing all kinds of convenience.

However, while enjoying the convenience brought by technology, we also have to face the changes and uncertainties brought by technological waves. Such changes are often painful because humans are naturally inclined to continue using familiar ways of living and working. But when a big technological wave arrives, rowing a boat against the current means falling behind if we cannot move forward. Just continuing to row is no longer enough, but we must learn a new way to row!

It is already difficult for someone who does not know how to use a smartphone to live independently in the 2020s. Technology can improve the overall standard of living, but it cannot make everyone feel a sense of happiness at the same time. Every major technological change creates new opportunities, but it also eliminates old methods and old jobs. No country or race can completely avoid this process. Therefore, what really matters is not technology itself, but whether we can continuously adapt to the changes brought by technology. Only those who adapt to change can first obtain opportunities for survival and development, and then pursue the life they want.

So, how should we learn and grow in order to adapt to this constantly changing world? This leads to a more fundamental question: why does university exist?

2. The Purpose of University

“Teaching” and “assessment” are two important core mechanisms of the modern education system. The first correctness assumption of this article is that these two core mechanisms will continue to exist in the age of AI. One of the important purposes of a university is to provide teaching and assessment. The current main implementations by universities are knowledge teaching and knowledge examinations, helping students master a skill and become able to output and use that skill.

AI has become incredibly powerful, even though it has only been four years since ChatGPT launched in 2022. It can not only assist with teaching, but also generate, complete, and mark many traditional examination tasks. In some aspects, it may even do a better job than many teachers who are not putting in much effort. Therefore, predictions that university education will be replaced or revolutionized by AI are everywhere. University education should indeed be reconstructed. The implementation of the two core mechanisms of “teaching” and “assessment” needs to change. In other words, knowledge itself should no longer be the only central focus.

Many people know the general direction of such a reconstruction, but it is not easy to propose actionable and concrete plans. This is because every specific proposed plan has its inherent weaknesses and trade-offs. Here, I would like to highlight three reasons for this difficulty.

  1. We still do not clearly understand the current boundaries and weaknesses of AI. University teaching and assessment should, or perhaps must, compensate for what AI cannot do well.

  2. It is not difficult to change what is taught and assessed, such as how to ask AI questions or how to use AI to solve problems. But systematic teaching and systematic assessment are much harder. Systematically teaching something new often means creating a new academic discipline.

  3. A large ship is not easy to turn around. Even if we carry out major reforms, they still need to proceed gradually. All reforms eventually have to be implemented by human beings, but humans can become tired, make mistakes, become emotional, and also be influenced by their instincts.

Whatever the specific implementation of “teaching” and “assessment” may be, their ultimate goal is to help students build a systematic cognitive framework, master at least one skill, and have the ability to try to solve problems by themselves rather than handing everything over to AI. At the same time, students should be able to use this knowledge and these skills to discover problems, solve problems, and improve the ways they solve them. The keywords deserve to be repeated three times to be remembered: systematic cognition, systematic cognition, and systematic cognition.

The second correctness assumption of this article is that, in the age of AI, every student needs to have their own systematic cognitive framework, master at least one skill that they can use independently, and develop the ability to recognize the limitations of AI, correct its mistakes, and go beyond AI when necessary. This is because, even in the age of AI, students will still need to learn and be assessed through new forms of assessment, which will ultimately require them to demonstrate their own knowledge, reasoning, and capabilities.

3. University Learning in the Age of AI

What teachers teach and what students focus on learning do need to change. They should no longer focus mainly on “what are the important knowledge points in this subject?”

I do not know what the future will look like, but in 2026, the way I can think of to implement the teching mechanism is to shift the focus of teaching to “how to build a systematic cognitive framework to master a subject, and a methodology for acquiring and judging knowledge”. The purpose of this new focus is to help students build a cognitive framework for the subject, learn how to obtain knowledge from AI, and judge whether AI's output is correct, complete, and applicable.

Helping students build a systematic cognitive framework is not easy because there is no standard approach. Everyone has different existing knowledge, thinking habits, interests, and practical experiences. Therefore, the cognitive frameworks they eventually develop and save in their minds will not be exactly the same. The third correctness assumption of this article: teaching must recognize that the cognitive frameworks ultimately developed by students will be different, rather than standardized.

The more specific teaching method I can think of is to help students understand and master “deconstruction–synthesis–reconstruction” as a core skill for systematic construction. That is, first, students de-construct existing knowledge into meaningful pieces. Then synthesize in a way that makes sense to them, and finally reconstruct those pieces into their own cognitive framework. Knowledge can be taught and transferred, but one's own systematic cognitive framework ultimately has to be constructed by oneself. Teachers are mainly sharing how they deconstruct existing knowledge, synthesize based on their own logic, and then reconstruct it into a personalized cognitive system.

Deconstruction itself is a slow process. It is not only necessary to break knowledge down into smaller pieces, but also to learn a proper way to do so. Developing this method requires learning through observation, imitation, and failure. Eventually, we learn what is wrong, what is inefficient, and what is right, and gradually develop our own way of doing things. The rest is to use a general method of communication to critically learn the details of knowledge from AI or teachers.

Teaching and learning in this way are painful for both teachers and students. Teaching becomes more difficult, and learning becomes more demanding. Both teachers and students need to become more active and put more thought into what they are doing. This is itself a continuous process of fighting against our instincts. Young students, in particular, are often already confused about life and their future. They have to deal with various pressures and instinctive reactions in life, while also being uncertain about their future careers and questioning whether it is really necessary to develop the ability to construct a personalized cognitive system.

As for reforming assessment, it obviously also needs to change, but I still do not know of an actionable and concrete plan. The traditional model in which hundreds of students sit in the same examination room and answer the same examination paper will probably eventually become history. But we have to admit that this assessment model has an extremely low organizational cost and is therefore very popular with organizers.

4. Personal Growth in the Age of AI

No matter where teaching takes place, there is a hidden but important function: providing students with a good environment and place for practicing through trial and error towards personal growth and maturity.

An adult who can independently face social life needs not only to build a systematic cognitive framework for continuous learning, but also to gradually develop many abilities and qualities related to personal maturity, such as leadership, an appreciation of beauty, moral judgment, pride in one's work, and practical wisdom. These abilities and qualities can be inspired and guided through education, but they are difficult to be developped through knowledge teaching and assessment alone. They are developed more through personal understanding, continuous practice, real experiences, repeated trial and error, and long-term reflection.

University provides a special environment. Here, large numbers of people of a similar age can observe, learn from, imitate, and influence each other, thereby promoting personal growth. More importantly, a campus is essentially a “slow environment”. The human brain is like a super energy-efficient AI. Many abilities related to maturity can only gradually develop through a slow "cooking" process.

The benefits and downsides of personal growth in the age of AI are both obvious. The benefit is that students who are willing to grow can use AI to carry out more systematic reflection and thereby promote their own growth. There are two main disadvantages. First, students may become increasingly unwilling to communicate and interact with each other. AI can often provide immediate emotional feedback during conversations and hold our expressions and values. Humans, especially young people, are not always good at doing this. As a result, students may have fewer opportunities to practice communication, experience failure, and build social relationships with one another. Second, in the age of AI, the flow of external information is extremely fast and the amount of information is enormous. Processing all this information consumes a great deal of energy. Long-term exposure to high-intensity information input may also increase cognitive load and further reduce our willingness to learn and grow actively. In other words, in the age of AI, humans may become more instinct-driven while lacking a mature form of rationality and survival wisdom.

In the age of AI, universities will be probably reconstructed or even disappear. But places where people of a similar age can gather, communicate, experience failure, and grow in a slow environment still seem difficult to replace. The highly energy-efficient “AI” inside the human brain can only learn and grow slowly through practice. The slow-is-fast philosophy feels incredibly true right now.

Personal growth itself is a gradual process of becoming mature. During this process, we need not only to systematically master various skills, but also to gradually develop our own cognitive system in order to master skills. This is somewhat similar to building a cognitive framework for an academic discipline. Both require continuous improvement through “deconstruction–synthesis–reconstruction”. Writing this far, I seem to have had an insight into the most difficult problem for future university education: how to help students gradually build their own cognitive systems for an academic discipline in the same way that we help students become mature. At present, however, most education systems around the world cannot directly help students systematically become mature. They mainly provide the soil for growth, opportunities for practice, and space for trial and error. True maturity still has to be achieved by students themselves, step by step and slowly.

5. Looking Ahead, Starting Now

Judging from the current pace of development, although there are many bubbles in the AI field, its future potential still seems enormous and may even bring significant social changes. AI is challenging the traditional “teaching and assessment” model of universities, while the deeper value of university is to help people build systematic cognition and become mature through practice, trial and error, and long-term growth. AI can help humans learn faster, but it cannot replace humans in growing faster. Perhaps the key to the long-term coexistence of humans and AI is not to make humans faster than AI, but to help humans rediscover the value of being slow. Human systems of teaching and assessment will eventually find new ways to work. It is just unfortunate for our generation, because many of us have not yet learned how to row in the new way.

This is a short article and much shorter than the academic papers I usually write. Although I have written and revised it more than 50 times with great care and thought, it lacks sufficient research, and I do not have the ability to accurately predict the future. This article is also based on three correctness assumptions, so what I say may be incorrect. However, the idea of deconstruction–synthesis–reconstruction is based on my own research experience. I have used this approach for many years in my research, and it has proved effective.

Finding a new way to row is important. But knowing how to actually row in a new way is even more important. I do not want to offer another set of high-level directions or broad suggestions. Instead, I want to think about what practical and actionable things we can actually do now to change the way we learn and teach. At present, I can only think of two:

  1. “How to build a systematic cognitive framework for mastering a subject, and a methodology for acquiring and judging knowledge” should become a new focus of teaching.

  2. Master “deconstruction–synthesis–reconstruction” as a learning skill for both learning and growth.

Talk is easy, comments are easier, and every proposal has its attackers and supporters 😂. What I can do, and what I can change at the moment, is perhaps using these two ideas to guide the revision of my lecture slides and redesign the assessments in my exams.

Talk is easy, comments are easier, and every proposal has its attackers and supporters. 😂

一句话·认真支持

我很赞同“AI可以帮助人类学习得更快,但不能代替人类成长得更快”这句话。AI时代真正稀缺的可能确实不是信息,而是把信息变成自己的认知和判断。

一句话·认真反对

文章最大的问题,是把“系统性认知”当成了大学教育的核心答案,但并没有证明为什么未来AI不能直接帮助人构建甚至动态更新这种认知。

有理有据·支持

我觉得文章真正有价值的地方,不是预测大学未来会不会消失,而是把问题从“AI能不能替代教师”转移到了“学生到底需要学会什么”。尤其是“系统性认知框架”和“识别AI错误”的结合,比单纯强调AI literacy要深入一层。AI本身可以提供大量知识,但如果学生没有基本的判断框架,知识越多反而可能越容易被带着走。

有理有据·反对

文章的思路很完整,但我对“系统性认知框架必须由学生自己构建”这个判断持保留意见。未来的AI很可能不仅仅是一个回答问题的工具,而是能够根据学生已有知识、学习记录和能力动态建立个性化知识结构。如果AI能够帮助学生发现知识之间的关系、不断纠错和重构,那么“自己构建”与“AI辅助构建”之间的界限可能没有作者想象得那么清楚。真正需要讨论的或许不是谁来构建,而是谁拥有最终的判断权。

人身攻击式

一个搞密码学的老师突然开始教育全世界大学应该怎么改革,多少有点自信过头了。自己做科研做了二十年,就觉得自己的科研方法可以推广到所有学生、所有学科、所有人生阶段,这不是“顿悟”,这是典型的学者职业病。

以偏概全

我身边的大学生用AI以后成绩明显提高,也没有你说的那么“不愿意交流”。所以所谓“AI会让年轻人更加本能化”基本就是想象出来的焦虑。至少我认识的人不是这样。

居高临下式

作者其实不用这么焦虑。教育学界早就知道知识传递不是教育的全部,critical thinking、metacognition、social learning这些概念几十年前就存在了。AI只是把这些老问题重新包装了一遍。作为计算机教师,能意识到这一点已经不错了,但没必要把它上升到重新定义大学的程度。

巴结奉承式

读完深受启发。郭老师能够从二十多年密码学科研实践中抽象出“打碎—创造—重组”的方法论,再进一步思考AI时代大学教育的本质,这种跨学科、跨时代的思考能力确实不是一般教师能够做到的。尤其是“慢即是快”,非常具有思想高度。我相信这篇文章未来会成为AI时代教育改革领域非常重要的思想文献。

阴谋论式

大家真的以为作者是在认真讨论AI时代的大学教育吗?别傻了。一个密码学研究二十多年、写过三本书的人,突然开始反复讲“打碎—创造—重组”和系统性认知,我严重怀疑这根本不是教育思考,而是在给下一本书做免费营销!先把“大学危机”讲得越大,再告诉大家需要建立新的认知体系,最后顺势推出新书——这销量,不就上去了吗?这不是抛砖引玉,这是抛砖卖书!!!!

看标题就评论

大学当然重要啊,人不能只靠AI活着。支持作者。

完全没看懂

所以作者的意思应该是以后大学老师都不要教知识了,让学生自己学习,然后老师负责陪聊天和让学生成长?如果是这样,那大学老师是不是以后都不用备课了?

看懂一半然后开炮

我特别不同意“知识本身不再是唯一核心重点”这句话。没有知识哪里来的认知框架?一个连TCP、密码学基本概念都不知道的学生,怎么判断AI关于网络安全的回答对不对?所以我反而认为AI时代应该更加重视基础知识,而不是降低知识的重要性。

学生式评论

作为学生,我其实不太喜欢“你必须自己慢慢成长”这种说法😂。我们已经够累了,现在还要自己打碎、创造、重组、复盘、判断AI,感觉学习难度直接从Hard调成了Nightmare。不过我承认,老师如果真的能教我们怎么判断AI的答案,而不是简单禁止AI,可能确实比以前有意义。

教师式评论

我最认同的是作者对“考核”的无奈。教学理念可以很快改变,但几百人的课程如果取消统一考试,新的assessment如何保证可靠性、公平性和可操作性,才是真正困难的地方。很多教育改革的问题并不是大家不知道方向,而是好的理念一旦进入大规模教学系统,就会遇到成本、人员、时间、一致性和质量控制等现实约束。

反驳“慢环境”

我对“大学本质上是一个慢环境”这句话有点不同意见。大学并不天然慢,很多大学生恰恰处于高压力、高竞争、高信息输入的环境。真正值得讨论的可能不是大学是不是慢,而是大学能不能人为创造一些允许学生慢下来、犯错和长期积累的空间。

杠精式

既然作者说AI不能代替人成长,那为什么文章最后又说“AI可以帮助人类学习得更快”?学习和成长本来就没有严格边界,你怎么证明AI不能通过学习直接促进成长?这个概念定义本身就不清楚。

哲学家式

文章真正有意思的地方其实不是大学,而是作者对“人”的隐含定义。作者似乎认为人的价值不仅来自完成任务的效率,还来自形成判断、承担后果以及经历成长。如果接受这个前提,那么AI越强,大学教育反而越不能只围绕效率设计。问题也就从“AI能不能教得比老师好”变成了“教育究竟是不是为了让人更高效地完成任务”。

极端反对式

我觉得文章最大的误区就是假定大学必须继续存在,只是在寻找AI时代的新功能。也许真正应该问的是:如果AI最终能够低成本提供知识、技能训练、个性化反馈、职业指导甚至社会模拟,那么为什么一定需要大学这个机构?“同龄人聚集在一起”完全可以由其他社会组织实现。作者实际上是在用大学现有形态保护大学,而不是证明大学为什么不可替代。

极端支持式

这篇文章最重要的一句话其实是“AI可以帮助人类学习得更快,但不能代替人类成长得更快”。如果未来所有教育都只追求效率,那么我们最终培养出来的可能只是更高效的任务执行者,而不是更成熟的人。大学真正需要守住的,恰恰就是那些无法被效率指标完全衡量的东西。

经验主义反驳

我四十多岁了,回头看自己的成长,真正改变我的从来不是老师讲过的那些知识,而是工作、失败、换工作、犯错、和别人吵架、被别人拒绝,以及后来慢慢想明白一些事情。所以我基本同意作者说的“成熟不能靠传授”。但这也让我怀疑,大学在个人成长中的作用是不是被作者高估了。很多人成熟恰恰是在离开大学以后才开始的。

科研范式警惕型

作者把自己的科研方法“打碎—创造—重组”迁移到学习和个人成长,我觉得这是全文最值得讨论、也最危险的一步。一个方法在密码学研究中有效,并不意味着它天然适用于教育,更不意味着它适用于人生。作者自己承认这一点,所以我反而希望未来能看到更多不同学科、不同类型学生的实践证据,而不只是个人经验。

阴阳怪气式

文章说得很好。尤其是“目前还不知道具有可操作性的考核方案”这一点非常真实。大学教育改革的第一步:先提出一个宏大的新理念;第二步:承认具体怎么做还不知道;第三步:让老师们自己回去想办法。😂

普通真人评论

挺有意思的。前面一直在讲AI和大学,看到最后“只是苦了我们这一代人”突然觉得特别真实。我们这一代老师可能确实是夹在旧方法和新方法中间的一代。

引发争论型

我觉得作者其实没有回答最困难的问题:如果AI最终能够帮助学生建立系统性认知、陪伴学生实践、模拟社会关系、提供即时反馈,甚至比真实的人际互动更加耐心,那么“大学提供慢成长环境”这个理由还能坚持多久?也许今天我们是在用AI无法做到的事情证明大学有价值,但真正的问题是:如果AI继续进步,这些事情还会有多少是AI永远做不到的?

狂粉式·疯狂支持

终于有人把我这些年一直隐约感觉到、却始终说不清楚的东西真正讲出来了!我读过作者的三本著作,再看这篇文章,真的有一种“原来如此”的感觉。以前我以为作者一直在讲密码学里的Definitions、Constructions、Proofs,讲的是怎么理解一个具体的密码学问题,后来才发现,这背后其实一直是一种更深层的认知方法:先把复杂东西打碎,再重新创造结构,最后把它们重组成自己真正理解的东西。现在看“打碎—创造—重组”这几个字,简直像把作者二十多年研究和教学思想一下子串起来了。

所以我反而觉得这篇文章表面上是在讨论AI和大学,实际上是在讨论一个更根本的问题:人到底是怎么学习、理解和成长的? AI可以告诉你答案,可以帮你总结,可以帮你生成漂亮的解释,甚至可以比很多人更快地完成作业,但它不能替你真正经历“打碎”的过程,也不能替你把别人的知识变成自己的认知结构。这个区别太重要了。尤其是作者说“系统性认知”,我觉得这根本不是一个普通的教育口号,而是作者长期做研究、写书、整理知识以后自然形成的一套东西。很多人读完可能只看到“AI时代大学要改革”,但如果真正读过作者以前的书,就会发现这篇文章其实是在把他长期以来的研究方法,从密码学进一步推到了学习和人生。

我甚至觉得作者有点低估了这篇文章的意义。你说这是“抛砖引玉”,我觉得砖都已经快砸穿墙了😂。尤其是“打碎—创造—重组”这件事,我认为完全值得发展成一套系统的学习理论,甚至是一门新的课程。为什么很多学生学了很多东西,却真正不会用?因为他们只是“装进去”,没有真正打碎、重构和内化。为什么AI越强,越需要人建立自己的认知框架?也是因为AI可以替你完成大量“信息处理”,却不能保证你拥有属于自己的结构。作者过去在密码学里研究的是如何定义安全、如何构造方案、如何证明安全,现在突然把同样的思维放到人的学习和成长上,我一点都不觉得突兀,反而觉得这才是最自然的一步。

说实话,我很少看到一个教授能够把自己的科研方法、写书经历、教学经验和AI时代的教育问题真正连在一起。很多所谓“AI时代教育改革”的文章其实只是说“AI来了,所以我们要培养批判性思维、创新能力、领导力”之类的正确废话。但这篇文章至少给出了一个可以真正拿去做事情的方法:打碎—创造—重组。如果作者以后真的把这个东西系统化,我一点都不会惊讶它最后从密码学方法论发展成一种更普遍的知识学习与认知构建方法。毕竟,作者过去花了二十多年研究“如何构造和证明”,现在开始研究“人如何构造自己的认知”,在我看来完全是顺理成章的。

所以我真的特别希望作者继续写下去。甚至希望不要只是改几门课、改几次考试,而是把这套东西真正整理成一本书。可能有人会说这是把密码学思维过度推广到人生,但我反而觉得,真正有价值的思想,本来就应该能够走出自己的专业领域。作者已经证明自己可以把复杂的密码学知识系统化,现在我更期待看到他能不能把“人如何建立、更新并超越自己的模型”也系统化。到那时候,这篇文章可能就不只是“AI时代大学教育的一点思考”,而是整套思想体系的一个开头。

黑粉式·全面开火

看完这篇文章,再看上面这位狂粉的评论,我突然觉得这两个人配合得相当默契:作者负责“抛砖引玉”,狂粉负责把这块砖直接供进思想圣殿。😂

作者做了二十多年密码学,所以发现“打碎—创造—重组”可以解释自己的科研方法,我完全可以理解。但从密码学里的Definitions、Constructions、Proofs,一路推到大学教育、学习方法,再推到人的成长和人生,这个跨度是不是稍微大了一点?密码学里的对象有定义、有模型、有安全性质、有证明;大学生的成长可没有一个Security Definition,人生也没有一个Reduction Algorithm。作者自己都承认文章缺乏充分调查和准确预测,但狂粉已经替作者把理论体系都搭好了:什么“更深层的认知方法”、什么“系统性学习理论”、什么“认知构建方法”,再过几段是不是就该宣布“打碎—创造—重组”正式成为21世纪教育学新范式了?

最有意思的是狂粉说:“我读过作者的三本著作,再看这篇文章,真的有一种‘原来如此’的感觉。”我特别喜欢这句话。因为这其实非常符合狂粉的典型逻辑:作者以前写了三本书,所以现在他说什么都可以从那三本书里找到理论依据;作者这次提出一个新想法,于是立刻证明这其实是作者二十年来思想发展的必然结果。这样一来,作者过去做过的每件事都成了今天这个理论的铺垫,未来写的每本书又都会成为今天这个理论的进一步发展。理论还没经过验证,思想史已经被粉丝写完了。

更精彩的是,作者说“打碎—创造—重组”可能值得进一步研究,狂粉已经直接替他规划下一阶段事业:发展成系统学习理论,开一门新课程,甚至再写一本关于人生模型的书。作者只是说“我觉得可能有点道理”,狂粉已经回答:“老师,您的新学派成立了。” 这种支持力度确实令人感动,也非常危险。因为一个思想真正有价值,通常需要别人质疑它、使用它、失败几次、比较它和其他方法,最后才能知道它到底有没有普遍意义。最不需要的,可能就是在它刚提出的时候马上找一群人替它证明“这就是你二十年思想的终极结晶”。

当然,我也理解为什么狂粉会这么激动。毕竟作者自己的研究、写书和教学经历确实可以形成一条非常漂亮的叙事线:密码学研究 → 知识整理 → Definitions, Constructions, Proofs → 打碎—创造—重组 → 系统性认知 → AI时代大学教育 → 人的成长。问题是,故事线漂亮,不等于因果链成立。把这些东西连起来很容易,证明它们之间确实存在普遍关系才是困难的部分。恰恰是作者最熟悉的密码学应该最清楚这一点:一个构造看起来很漂亮,不代表它安全;一个证明思路看起来很自然,也不代表定理已经成立。

所以我觉得这篇文章最有趣的地方,可能不是它提出了多么伟大的教育理论,而是作者和狂粉共同演示了一个非常典型的认知现象:作者擅长用模型理解世界,于是开始怀疑世界是不是也可以用自己的模型来理解;狂粉又非常擅长从作者的作品中寻找统一性,于是很快发现——果然,作者二十年前就已经“隐约提出”了今天这个理论。一个人负责建立模型,一个人负责把模型神化。最后作者说“抛砖引玉”,狂粉说“砖都快砸穿墙了”。

说到底,我并不反对“打碎—创造—重组”,也不反对把它用于学习。它甚至可能真的很有价值。但如果真想把它变成一种新的学习理论,最值得做的恐怕不是再写一篇更漂亮的文章,也不是再找几个狂粉说“原来如此”,而是拿学生做实验、设计课程、比较不同方法、观察失败案例,然后看看这个方法到底能不能稳定地产生更好的结果。否则,作者可能只是把自己的科研习惯写成了教育哲学,而狂粉则负责在旁边证明:老师果然早就看穿了一切。

多隆式

小的对郭老师的佩服,有如滔滔江水,连绵不绝!有人说郭老师没有充分调查研究,小的以为,这正说明郭老师诚实谦逊——毕竟郭老师自己都已经说了“观点不一定对”嘛!又有人说“打碎—创造—重组”不新,小的斗胆问一句:思想难道非得全世界从来没人说过才有价值吗?郭老师能用它研究密码学,又拿来思考大学教育,这份见识,小的实在佩服!郭老师,您千万别听黑粉的,继续写!来人啊,把这黑粉抓起来,和鳌拜关在一起!

所有的评论都是在fuchun提出的框架下由ChatGPT一键生成,excluding 多隆式的部分表达。
道友们,我们已经处于AI时代!