On the Prospects of AI in Education 人工智能与教育的前景

On the Prospects of AI in Education

Introduction

In the past, there were many specific instances where technology significantly transformed traditional paradigms in education. The invention of the printing press in the fifteenth century revolutionized access to information by making books widely available, resulting in the dissemination of knowledge and exponential increases in literacy rates; The invention of the radio and television in the twentieth century delivered educational content remotely to a broader audience – anyone with a receiver could watch educational programs; The advent of the age of the internet and personal computers in the late twentieth century transformed education by enabling computer-assisted learning, further changing how students research and acquire knowledge; Smartphones and mobile learning took a step further by providing on-the-go access to educational resources, creating countless “classrooms” outside traditional contexts. Today, though, the world is at the cusp of a revolution in education hitherto unseen in modern history. The dawn of the age of artificial intelligence (AI) challenges conventional educational models and bears incredible promises to make quality education universal.

While it is undeniable that AI boasts the capacity to revolutionize education for the better, educators worldwide must simultaneously caution against the accompanying risks of equal significance. AI’s potential risks in education exist far beyond the notion that students can use AI software to cheat or conduct other kinds of academic malpractice. As this essay will explain and surmise, the AI age delivers non-negligible ramifications on the foundations of education, including human connections, competition, the teaching of values, and the relation between the learner and knowledge. However, on a more fundamental level, educational AI developers must ponder the question: After all, what is an educational AI?

What is an Educational AI Anyways?

Today, many AI developers believe that educational AI is already a well-defined term. It is a computer system designed for use in educational situations, able to perform tasks normally requiring human intelligence. Unfortunately, if these developers and educators truly desire to apply AI’s power to all stages of education – from early kindergarten to late high school – then they must define educational AI in a far more approachable and understandable way that even a Grade 1 student can easily grasp. To this end, this essayist champions the adoption of the metaphor created by AI developer Mustafa Suleyman that AI, especially in education, is an independent “digital species,” a companion to a human educator or learner.

This metaphor suggests that the perception of education AI as just a tool is too conservative. Generally, a tool is something that humans can take full control of. The results of the usage of a tool are usually relatively predictable. A “species,” a companion, implies that educational AI will demonstrate creativity, decision-making skills, proactive responses to changes in environment, and even empathy. In fact, some educational AI models, such as “Khanmigo”, developed by Khan Academy, already boast almost all these abilities.

Educational AI’s identity as a digital species separates itself from previous revolutionary products in education, which were only designed as mere tools to augment the educational experience. Given the speed at which AI software advances, AI might shake the foundations of previous educational methods and philosophies. It displays the potential to reconstruct the educational industry. Thus, educators should not underestimate AI’s impacts and must brace for a most powerful blow.

AI’s Potential Disruptions to Socialization Under an Educational Context

A dangerous opinion on AI’s impact on the element of socialization in education is that it might only alter the traditional student-teacher relationship. This essayist believes that with AI serving as potentially more patient, knowledgeable teachers, the traditional bonds between human students and teachers will surely reform or vanish altogether. AI can do much more than simply reshape student-teacher relations. It can undermine relations between fellow students as well, thereby further influencing their social-emotional well-being.

First, consider the education of young children, which this essayist will arbitrarily define as children from ages five to twelve. In the future, AI software will provide tailored services to children of this age. One of the major outcomes of adopting AI in the educational process is that there will be more individualized learning and reduced face-to-face group work opportunities. Face-to-face group work is a prime opportunity for youngsters to learn interpersonal interaction principles in action. Scholars Ying Xu and Mark Warschauer claim that if these interpersonal interactions become mostly replaced by AI substitutes, children will be less likely to view human connections as integral to their emotional well-being. In their paper “What Are you Talking To?: Understanding Children’s Perceptions of Conversational Agents,” the two scholars show that after interacting with AI agents, many kids before the age of six believe that these agents possess feelings, thoughts, and abilities that are highly comparable to a human companion. As a result, they demonstrate less interest in interacting with each other.

Another study found that AI agents have the potential to break the most rudimentary trust between young children and their peers or teachers. After conducting a series of observations on a selected group of children, the authors concluded that young children involved in the study thought smart, AI-programmed algorithms are more reliable than humans when answering fact-based questions, such as “Who is the first United States President to drive a car?” These studies reveal that educational AI’s risk of discouraging crucial human interactions and trust in the early stages of education cannot be ignored.

Second, consider the education of older students aged twelve to eighteen. Educators assume that these students are more experienced and conscious about their choices. Still, if AI software proliferates through the educational systems in place for these students, their socialization abilities and incentive to interact with others will likely remain stagnant. Most educational AI comes in the form of a virtual assistant, providing extremely efficient, emotionless interactions that require minimal social effort. Consequently, students who share daily conversations with an AI assistant might struggle to grasp the nuances involved in human conversation with their peers. This difficulty might further induce individuals to seclude themselves, a highly detrimental approach that triggers psychological issues.

AI’s Impact on Unhealthy Social Competition

Even if educators shrug off the argument above and claim that human interactions will not completely disappear as long as institutions implement necessary precautions, they will find it challenging to tackle the issue that AI might exacerbate, not mitigate, unhealthy competition on a larger, social scale. Some people posit that since educational AI will accelerate the intake of knowledge by the student, then surely, they will be less anxious about their academic performance and shift their focus to other activities. This belief is quite weak upon further examination.

Most fundamentally, unless AI wholly transforms how educators assess the performance of their students, otherwise the competition between students for a “stronger” proof of their learning abilities will only intensify with the proliferation of AI software. AI provides extensive learning resources, which might induce students to prioritize grades or other verifications of ability over everything else, especially in this age when acquiring knowledge and achieving higher “scores” seem to be the ultimate pursuits of academic life. In fact, today, ChatGPT is mainly used to elevate the quality of students’ writing, a large portion of which is submitted for school grading and comparison with fellow students. The more knowledge that AI provides, the more likely students will – through systems of gamification, AI-driven testing, and ranking – compete to demonstrate their proficiency in mastering such knowledge.

The AI age might even witness the “evolution” of unhealthy competition to new dimensions. In the novel Klara and the Sun, written by Nobel laureate Kazuo Ishiguro, while AI technology seemingly erases the competition to master knowledge, it “raises” the dimension of competition to a point where students’ socialization abilities are graded and compared. The book’s content is fiction. Nonetheless, it is highly pertinent to the trend of AI development today. One way to fundamentally solve the issue of increasingly white-hot unhealthy competition is to overhaul and restructure the assessment system, redesigning the criteria to measure performance. It is unsure whether educators can and are willing to use AI for this purpose.

Another common belief is that since AI makes knowledge truly accessible to everyone through the most efficient network, there will be greater equity in distributing educational resources, narrowing previously existing disparities and reducing unhealthy competition. AI will probably bring more equality but not equity. The AI educational world is not one in which developers wave their algorithmic wands and educational inequities or inequity-induced competition magically vanishes. After all, there will still be marked resource disparities between the elite and non-elite families who need these resources more desperately. For example, AI is unlikely to run the whole educational system. Humans will be involved in providing support in the usage of AI software. It might still, unfortunately, be elite families that capture these human resources and hence enhance the power of AI to transform their children’s education. Such disparities might further create conflicts and fervent competition between families of different social statuses.

More significantly, the idea that educational AI will provide equal resources and opportunities to everyone rests on the premise that the AI software market must be a monopoly or oligopoly, with only one or a few sellers providing the same service to every consumer. This premise fails to convince. The educational AI landscape will be incredibly diverse in the future. Suleyman believes that there will be no single algorithm that dominates the market. Instead, countless companies and individual developers will write their own AI codes based on publicized knowledge of AI computing. As a result, different AI software will bear different qualities – some will be equipped with the most cutting-edge technologies or features, while others will be less capable. This essayist predicts that educators and students will compete fiercely to obtain the most capable software in the market. There will be individuals with more widespread resources who are lucky enough to hold the most advanced algorithms in their hands. And there will be large groups of individuals who fail to do so. When computers and the internet unleashed new potential to learn more effectively and efficiently, some families could afford and obtain the best, immediately elevating their children’s learning, and some families struggled then and even today to access these technologies. Thus, similarly, far from creating a more equitable educational environment, the proliferation of AI might be the progenitor of a trend where educational resources become more unfairly distributed due to differences in individuals’ social status, exacerbating social competition.

AI and the Education of Values

As the essay finished to examine the risks that educational AI brings to the learning environment or the social atmosphere at large, it shall turn to a discussion of AI’s relationship with the fundamental purpose of education: To convey a set of values and moral principles that an individual will live by. An education system that concentrates excessively on the absorption of knowledge alone, which is increasingly the case in many countries today, fails to fulfill this imperative purpose. Traditionally, human teachers or mentors have been the drivers behind the education of values. In the future, as AI becomes ever more omnipotent, will this technology render human teachers obsolete in this task of educating values? This essayist claims AI will be a poor substitute. And if institutions opt to root AI solidly in their educational systems, they must face the risk of casting the fundamental purpose of education into oblivion.

To illustrate, one can consider this story first: “Over four millennia ago, the god Theuth offered the King of Egypt the gift of writing, which would be a ‘recipe for both memory and wisdom.’ But the king was not impressed. According to the king, ‘[Writing] will allow [a teacher] to tell [their students] many things without teaching them anything, [the teacher] will make them seem to know much, while for the most part they will know nothing. And as men filled not with wisdom but the conceit of wisdom, they will be a burden to their fellow men.” Previously, before the advent of writing, education was verbal – students must learn by listening, not by reading. This mode allowed for the conveying of conventional wisdom and values. When the mode of education dramatically changed following the invention of writing, the King of Egypt argued that the delivery of “wisdom” would be impossible because “[the teacher] will make [students] seem to know much, while for the most part they will know nothing.”

This story is a fitting analogy to the AI revolution in education. AI is like “the invention of writing,” surely to revolutionize the form of education. However, AI tutors will fail to teach social values due to the inherent incompatibility between AI and these values. The latter are the products of historical and conventional human wisdom, rooted in the depth of culture, which even the most advanced AI software will find extremely challenging to decipher and interpret. Therefore, AI algorithms are ineffective and dangerous substitutes to human teachers in this sense. They will produce students who are knowledgeable but may lack the basic moral principles to inform their actions.

AI and the Alienation of Knowledge

Finally, educational AI risks alienating students from knowledge itself. Karl Marx proposed his theories of alienation in the mid-1800s, amidst the fervor of the Industrial Revolution. In his manuscripts, Marx argues that alienation can be defined as a process in which work estranges an individual from humanity. In the context of the Industrial Revolution, he declared that workers did not own what they produced. The goods were owned and sold by capitalists instead.

In the context of education, this alienation theory seems especially pertinent. AI does provide a turbo boost to the rate at which knowledge is passed from a source to the learner. Nevertheless, does the learner “own” this knowledge? Or are they “alienated” from this knowledge under the power of AI? AI is so efficient that it takes away the processing time one must spend to digest knowledge and interpret it based on personal perspectives. Without processing, knowledge is mechanic memory, not food for thought. Hence, if students over-rely on AI, they must cope with the challenge of “owning” knowledge and converting it into creative and critical ideas. Otherwise, knowledge and the knower shall evolve into separate entities.

Final Thoughts

Overall, this essay revolved around some of the major risks looming as educators and AI developers march toward a misty future. That is not to say educators should stop integrating AI into education. As Sal Khan, the founder of Khan Academy, rightly points out, AI bears huge positive potential to improve education. For example, AI might solve the persisting “two-sigma problem,” meaning that by serving as a 1-on-1 tutor, AI will turn an average student into an excellent one and a below-average student into an average one. More impactfully, for countries with poor educational resources, AI creates a turning point for their governments to provide quality education for the youth. Unlike the last Industrial Revolution, the wave of AI will sweep the shores of most, if not all, countries. Furthermore, AI software promises to deliver experiential learning to every student. Khanmigo, for instance, Khan Academy’s AI chatbot, can engage in conversations with a student in the voice of a historical figure, making the study of history much more vivid. One can only imagine, then, how many more physicists, mathematicians, historians, engineers, AI might help produce for the world.

Educators and AI engineers, though, must be meticulous when making any conclusion about the prospects of AI. Generally, it is an overstatement to claim that any improvement in educational AI will benefit all students around the globe. Like any other groundbreaking invention, AI will impact different groups of students and teachers from various socioeconomic backgrounds. To take the greatest caution, educators must not only discuss which features to not put on the table but also study, case-by-case, the varied impacts that advancement in AI technology will bring to these different groups of students and teachers. The AI age dawns, and if the world picks its steps carefully, it will undoubtedly be the most productive era in human history, brimming with infinite creativity and opportunities.

Another related piece is: Links Between East and West 55-AI and Academics 东西方的连接55-AI与学术
‍‍‍‍‍‍‍‍‍‍‍

人工智能与教育的前景

导言

过去，有许多技术显著改变了传统教育的模式：15 世纪印刷术的发明使书籍广为流传，从而彻底改变了人们获取信息的方式，促进了知识的传播与识字率的指数级增长；20 世纪广播和电视的发明将教育内容远程传送给更广泛的受众--任何拥有接收器的人都可以收看教育节目； 二十世纪末，互联网与个人电脑时代的到来改变了教育，实现了计算机辅助学习，进一步改变了学生研究和获取知识的方式；智能手机与移动学习更进一步，提供了随身获取教育资源的途径，在传统环境之外创造了无数 “教室”。如今，世界正处于一场现代史上前所未有的教育革命的风口浪尖，人工智能（AI）时代的到来对传统教育模式提出了挑战，并有望在全世界普及优质教育。

不可否认，人工智能有能力彻底改变教育，让教育变得更好，但全世界的教育工作者必须同时警惕随之而来的同样重要的风险。人工智能在教育领域的潜在风险远不止学生可以利用人工智能软件作弊或进行其他类型的学术不端行为。正如本文将解释和推测的那样，人工智能时代对教育的基础，包括人与人之间的关系、竞争、价值观的教育以及学习者与知识之间的关系，都带来了不可忽视的影响。不过，首先，在更基础的层面上，教育人工智能开发者必须思考一个问题： 什么是教育人工智能？

到底什么是教育人工智能？

如今，许多人工智能开发人员认为，教育人工智能已经是一个定义明确的术语，它是一种专为教育环境设计的计算机系统，能够执行通常需要人类智能才能完成的任务。遗憾的是，如果这些开发人员和教育工作者真的希望将人工智能的力量应用到教育的各个阶段--从幼儿园早期到高中晚期--那么他们就必须用一种更平易近人、更易于理解的方式来定义教育人工智能，即使是一年级的学生也能轻松掌握。为此，本文作者主张采用人工智能开发者穆斯塔法-苏莱曼（Mustafa Suleyman）提出的隐喻，即人工智能，尤其是教育领域的人工智能，是一个独立的“数字物种”，是人类教育者或学习者的伴侣。

这一比喻表明，将教育人工智能仅仅视为一种工具的看法过于保守。一般来说，工具是人类可以完全控制的东西，且使用工具的结果通常是相对可预测的。一个 “物种”，一个同伴，意味着教育人工智能将表现出创造力、决策能力、对环境变化的主动反应，甚至是同理心。事实上，一些教育人工智能模型，如可汗学院开发的“Khanmigo”，已经拥有了几乎所有这些能力。

教育人工智能作为数字物种的身份将自己与以往教育领域的革命性产品区分开来，后者仅仅被设计为增强教育体验的工具。鉴于人工智能软件的发展速度，人工智能可能会动摇以往教育方法和理念的基础，它显示出重构教育行业的潜力。因此，教育工作者不应低估人工智能的影响，必须做好准备，迎接最强有力的打击。

人工智能对教育背景下社会化的潜在破坏

对于人工智能对教育中社会化元素的影响，有一种简单的观点认为，它可能只会改变传统的师生关系。该文作者认为，随着人工智能可能成为更有耐心、知识更渊博的教师，人类学生与教师之间的传统纽带必将经历改革或完全消失。实际上，人工智能能做的远不止重塑师生关系这么简单，它还能破坏同学之间的关系，从而进一步影响他们的社会情感。

首先，考虑一下幼儿教育，本文作者将把幼儿定义为 5 至 12 岁的儿童。未来，人工智能软件将为这个年龄段的儿童提供量身定制的服务。在教育过程中采用人工智能的一个主要结果是，个性化学习将更多，面对面小组合作的机会将减少。面对面的小组合作是孩子们在行动中学习人际交往原则的最佳机会。学者 Ying Xu 和 Mark Warschauer 声称，如果这些人际交往大多被人工智能替代品所取代，那么孩子们就不太可能将人与人之间的联系视为其情感幸福不可或缺的一部分。在他们的论文《你在和谁说话？》中，两位学者指出，在与人工智能代理互动后，许多六岁以下的儿童认为，这些代理拥有的情感、思想和能力与人类同伴非常相似。因此，他们对同伴之间彼此互动的兴趣降低了。

另一项研究发现，人工智能代理有可能打破幼儿与同伴或老师之间最基本的信任。在对一组选定的儿童进行了一系列观察后，作者得出结论，参与研究的幼儿认为，在回答基于事实的问题时，例如 “谁是第一位驾驶汽车的美国总统？”，人工智能编程的智能算法比任何人类都可靠。因此，孩子们更会相信人工智能，而非人类同伴或老师。这些研究表明，教育人工智能在早期教育阶段阻碍关键的人类互动和信任的风险不容忽视。

其次，考虑高年级学生（12 至 18 岁）的教育。教育者认为，这些学生更有经验，对自己做出的选择也更有意识。尽管如此，如果人工智能软件在这些学生的教育系统中传开，他们的社交能力以及与他人互动的积极性很可能会停滞不前。大多数人工智能教育软件都是以虚拟助手的形式出现，提供极其高效、不带情感的互动，只需极少的社交努力。因此，与人工智能助手进行日常对话的学生可能会发现，他们很难掌握与同伴进行人际交往时的细微性。这种困难可能会进一步诱发个人自我封闭，这是一种非常有害的方式，可能会引发心理问题。

人工智能对社会恶性竞争的影响

即使教育工作者对上述论点不屑一顾，声称只要教育机构采取必要的预防措施，人与人之间的互动就不会完全消失，但他们也会发现，要解决人工智能可能会加剧而非缓解更大范围内的社会恶性竞争这一问题，还是极有挑战性的。有些人认为，既然教育人工智能会加速学生对知识的吸收，那么他们肯定会减少对学习成绩的焦虑，并将注意力转移到其他活动上，这种想法在进一步研究后会发现非常薄弱。

最根本的一点是，除非人工智能彻底改变教育工作者评估学生成绩的方式，否则学生之间为证明自己的学习能力“更强 ”而展开的竞争只会随着人工智能软件的普及而加剧。人工智能提供了大量的学习资源，这可能会诱导学生将成绩或其他能力验证看得比什么都重要，尤其是在这个获取知识和取得更高的 “分数 ”似乎成为学术生活终极追求的时代。事实上，如今 ChatGPT 主要用于提升学生的写作质量，其中很大一部分是提交给学校评分和与同学比较的。人工智能提供的知识越多，学生就越有可能通过游戏化系统、人工智能驱动的测试与排名系统，竞相展示自己掌握这些知识的能力。

人工智能时代甚至可能见证恶性竞争向新维度的 “进化”。在诺贝尔文学奖得主石黑一雄的小说《克拉拉与太阳》中，虽然人工智能技术看似消除了掌握知识的竞争，但却将竞争的维度 “提升 ”到了对学生的社交能力进行评分或比较的地步。这本书的内容是虚构的，然而，它对当今人工智能发展的趋势却具有很强的针对性。要从根本上解决日益白热化的恶性竞争问题，一个办法就是彻底改革和重构评价体系，重新设计衡量成绩的标准。目前，教育工作者是否能够并愿意利用人工智能来实现这一目的尚不确定。

另一个普遍的观点是，由于人工智能通过最高效的网络让每个人都能真正获取知识，因此教育资源分配将更加公平，缩小之前存在的差距，减少恶性竞争。人工智能可能会带来更多平等，但不会带来公平。人工智能教育世界并不是开发者挥动算法魔杖，教育不公平或由不公平引发的竞争就会神奇地消失。毕竟，精英家庭和非精英家庭之间仍然会存在明显的资源差距，而后者更迫切需要这些资源。例如，人工智能不太可能管理整个教育系统，在使用人工智能软件时，会有人类参与提供支持。不幸的是，精英家庭仍有可能掌握这些人力资源，从而增强人工智能改变其子女教育的力量。这种差距可能会进一步造成不同社会地位的家庭之间的冲突和激烈竞争。

更重要的是，教育人工智能将为每个人提供平等的资源和机会这一观点的前提是，人工智能软件市场必须是垄断或寡头市场，只有一个或少数几个卖家为每个消费者提供相同的服务。这一前提无法令人信服，未来的教育人工智能领域将无比多样化：苏莱曼认为，不会有单一的算法主导市场，相反，会有无数的公司和个人开发者根据公开的人工智能计算知识编写自己的人工智能代码。因此，不同的人工智能软件将具有不同的质量与能力--有些软件将配备最前沿的技术或功能，而有些软件则能力较弱。本文作者预测，教育工作者和学生将展开激烈竞争，以获得市场上功能最强大的软件。有些人将拥有更广泛的资源，有幸掌握最先进的算法，同时，也会有一大批人未能如愿。过去，当电脑与互联网释放出更有效、更高效学习的新潜能时，有些家庭能够负担得起并获得最好的东西，从而立即提升孩子的学习水平，而有些家庭在当时甚至今天都难以获得这些技术。因此，与此类似，人工智能的普及非但不能创造更加公平的教育环境，反而可能导致教育资源因个人社会地位的差异而分配不公，加剧社会竞争。

人工智能与价值观教育

本文在探讨教育人工智能给学习环境或整个社会氛围带来的风险之后，将转而讨论人工智能与教育的根本目的之间的关系： 教育的根本目的是传递一套价值观和道德准则，让个人以此为生。当今许多国家的教育系统都过分注重知识的吸收，这就无法实现教育的这一根本目的。传统上，人类教师或导师是价值观教育的推动者。未来，随着人工智能变得越来越无所不能，这项技术是否会让人类教师在价值观教育的任务中被淘汰？本文作者称，人工智能将是一个糟糕的替代品。如果教育机构选择将人工智能扎根于其教育系统，那么它们就必须面对教育的根本目的被遗忘的风险。

为了说明这一点，我们可以先看这样一个故事："四千多年前，Theuth 神向埃及国王献上了文字这个礼物，这将是‘记忆和智慧的秘方’。但国王并不以为然。国王说：'（文字）会让（老师）告诉（学生）很多东西，却不教他们任何东西，（老师）会让他们看起来知道很多东西，而大部分他们什么都不知道。他们不是充满智慧，而是自负智慧，他们将成为同伴的负担'"。在文字出现之前，教育是口头的--学生必须通过听来学习，而不是通过阅读，这种模式可以传递传统智慧和价值观。当文字发明后教育模式发生巨大变化时，埃及国王认为，“智慧 ”的传递是不可能的，因为“[教师]会让[学生]看起来知道很多，而大部分情况下他们什么都不知道”。

这个故事用来比喻教育领域的人工智能革命再恰当不过了。人工智能就像 “文字的发明”，肯定会彻底改变教育形式。然而，由于人工智能与社会价值观之间固有的不相容性，人工智能辅导员在试图传授社会价值观时将会失败。后者是人类历史和传统智慧的产物，根植于文化的深处，即使是最先进的人工智能软件在解读或诠释这些价值观时也会发现极具挑战性。因此，从这个意义上讲，人工智能算法是人类教师无效且危险的替代品。它们培养出来的学生知识渊博，但可能缺乏基本的道德原则与价值观来指导他们的行为。

人工智能与知识的异化

最后，教育人工智能有可能使学生产生对知识本身的异化。卡尔-马克思于 19 世纪中期在工业革命的热潮中提出了他的异化理论。在他的手稿中，马克思认为异化可以定义为工作使个人与人性疏远的过程。在工业革命的背景下，他宣称工人并不拥有他们生产的产品，这些产品反而由资本家拥有及出售。

在教育领域，这一异化理论显得尤为贴切。人工智能确实为知识从源头传递给学习者的速度提供了动力。然而，学习者 “拥有 ”这些知识吗？还是他们在人工智能的力量下与这些知识 “异化”了？人工智能的效率如此之高，以至于剥夺了人们消化知识和根据个人观点诠释知识所必须花费的处理时间。没有加工，知识就是机械记忆，而非思考的源泉。因此，如果学生过度依赖人工智能，就必须应对 “拥有 ”知识并将其转化为创造性和批判性思想的挑战。否则，知识、知识者将演变成两个独立的实体。

最后的思考

总之，本文围绕教育工作者和人工智能开发者在迈向不可预测的未来时所面临的一些主要风险展开论述。这并不是说教育工作者应该停止将人工智能融入教育，正如可汗学院（Khan Academy）创始人萨尔-可汗（Sal Khan）正确指出的那样，人工智能具有巨大的积极潜力，可以更好地改变教育。例如，人工智能可以解决长期存在的“双标准差问题”，也就是说，作为一对一的辅导员，人工智能可以把平均学生变成优秀学生，把低于平均水平的学生变成平均学生。更具影响力的是，对于教育资源匮乏的国家来说，人工智能为其政府为青少年提供优质教育创造了一个转折点。与上一次工业革命不同，人工智能浪潮将席卷大多数国家，甚至所有国家。此外，人工智能软件有望为每个学生提供体验式学习。例如，可汗学院的人工智能聊天机器人 Khanmigo 可以用历史人物的声音与学生对话，使历史学习变得更加生动。可以想象，人工智能将为世界培养出更多的物理学家、数学家、历史学家和工程师。

不过，教育工作者与人工智能工程师在对人工智能的前景下任何结论时都必须非常谨慎。一般来说，声称人工智能教育的任何改进都将惠及全球所有学生，这种说法未免言过其实。与其他任何突破性发明一样，人工智能将给不同社会经济背景下的不同学生与教师群体带来不同的影响。为了慎之又慎，教育工作者不仅要讨论哪些功能不能被开发，还要逐一研究人工智能技术的进步会给这些不同的学生和教师群体带来哪些不同的影响。人工智能时代已经来临，如果世界谨慎地选择自己的脚步，这无疑将是人类历史上最富有成效的时代，充满无限的创造力与机遇。

关于人工智能的其他主题的文章，请阅读: Links Between East and West 55-AI and Academics 东西方的连接55-AI与学术