English version of this article is after the Chinese version. Ther English version is translated by ChatGPT 4o and slightly edited by Leo Huang. 本文的英文版在中文版本之后。英文版由ChatGPT 4.0翻译，并由Leo Huang稍作编辑。
Photo created using ChatGPT4o (DALL. E precisely).
我最近比较频繁地使用ChatGPT，也看了不少有关人工智能和ChatGPT的文章和书。似乎有了一些相关的认知和使用经验，因此逢人都想聊聊这个话题，以及AI对我们的工作和学习可能会产生的影响。我觉得影响肯定是已经开始了，只是现在大家都在想这个影响到底有多大多深多远而已。
李飞飞在她自传的《我所见的世界：人工智能黎明的好奇、探索与发现》说，“我相信我们的文明正处于一场技术革命的风口浪尖，这场革命有能力重塑我们所知的生活。” 这本自传的英文名字副标题中有“at the dawn of AI”，所以在我翻译的中文名字中用了“黎明”这个词。
无独有偶，可汗学院的创始人Salman Khan在他的新书《新勇敢的话语：人工智能将如何革命教育（以及为什么这是件好事）》中也表达了目前人工智能只是在教育中初期运用。他在书讨论AI对教学的辅助作用时也用了“黎明（dawn ）”这个词，他描绘的是“A I助教的黎明”的模样。书中另外一个章节的标题是“AI家教的崛起”，他用的是 “rise”，从描述的内容来看也是在表达一种初期新事物的意思。当然个性化的AI助教才是Khan这本书的主题。
我解释这两个英文单词的含义并不是讨论翻译的方法，实际上我是想强调现在这个时代的特殊性。 要知道从目前AI尤其是Chat GPT发展的趋势来看，它很可能会成为一种像蒸汽机、电力、内燃机和电脑一样的通用技术 （general purpose technology）,而这些技术的作用是在所有经济领域提高生产力。
因此，我想给在读这篇文章的教师或其它领域的朋友注入一点兴奋感和紧迫感。兴奋是因为我们可能正在见证一项伟大的技术的早期快速发展，紧迫是因为这项技术和之前的似乎有很大的不同，它不仅直逼人类的智能，而且它的发展速度实在是超过了我们人类智能的想象。
李飞飞在回顾自己30年的学习和研究经历时看到，二十一世纪第二个十年之前人类花费了超过半个世纪的时间才使人工智能产生的前提条件趋于成熟，而之后不到短短五年时间AI就飞速地改变了世界。她书中有这么一段话，“所有企业都发生了变革，数十亿美元投入AI研发，从行业分析师到政治评论员再到哲学家，每个人都在赶紧努力理解这种似乎一夜之间从学术小众变成全球变革力量的技术。”
从2022年11月第一个Chat GPT版本发布以来，这个世界肯定已经发生了一些变化了。有些工作已经或者说正在消失，比如初级的翻译和编程。有些新的领域已经或者正在出现，比如如何制作和训练个性化的GPT和教人如何有效使用这个这个工具，而越来越多的公司或者开始整合AI技术或者干脆基于人工智能而发展。
可汗学院最近上线了一个基于ChatGPT的AI 辅导老师，学院的创始人Salman Khan谈到为何要使用AI时说，这个这个潘多拉的盒子已经打开了，已经不可能再让世界变回原来的样子了。
我在教育教学和学校管理第一线，从ChatGPT一面世就感觉到了它对我的工作的影响。和很多同事一样，当时最大的忧虑甚至恐惧就是“完了，学生现在有了一个超级作弊工具了”。尤其是写作类的作业，老师所看到的文章很可能就是学生和ChatGPT共同完成的。传统的作弊定义也被迫修改了，学生不一定是在抄袭现有的文章，他只是找了一个强大的帮手一起完成而已。Salman Khan说，这和在家里找一个学习很好的哥哥帮了一下忙有什么区别呢？
今天的查重工具，比如TurnitIn所提供的报告也只是告诉教师学生的作业由AI参与完成的概率，最终的判断还需要教师根据对学生情况的了解而作出。(https://help.turnitin.com/ai-writing-detection.htm)
这个变化显然对教师提出了新的更高的要求，因为这种介于客观、主观和直觉之间的判断是极难作出的。老师既要了解AI是什么，也要更加了解他们的学生。我感觉有不少老师可能就是完全按照TurnitIn给出的概率来判断了，实际上这个概率本身大概率也是用AI算出来的！你看，这到底是谁在依赖AI呢？有个老师跟我开玩笑说，那以后我们就索性用AI来批改学生的作业吧，反正学生也在用AI来写作业了。
当然，即使是最高版本的ChatGPT 4o也还处在发展过程中，虽然其进阶速度可能挺会很快，但离人们期待中所谓的AGI (Artificial General Intelligence) 还有些距离。 “ChatGPT 可能会犯错误”，这句话是在ChatGPT使用对话框下面清清楚楚地写着的。当然人也会犯错误啊，所以这个提醒似乎有点多余。不过这个提醒至少让我们清醒地认识到不能完全依赖这个工具，不能从此不懂脑筋地使用它给出的答案或者设计。
我最近就差点上了ChatGPT的当。我和一个数学老师讨论一个高中12年级的排列组合题目。我和他的做法有分歧，我们就试着让ChatGPT做了一下，结果是和我们中的某一个做法一致，而且还分析得头头是道的。然而，我和那个数学老师再次讨论后发现，其实ChatGPT是用了一个错误的方法。我然后告诉ChatGPT做法有误，它就老老实实地重做了一遍不过还是同样的错误。我进一步告诉它错误的原因试图教育它一下让它纠正错误，结果它吐出了长长的好几页的乱码，就像一个人被突然指出错误后语无伦次胡言乱语一样。这个例子让我体会到Chat GPT毕竟还不是在“思考”！
不管怎样，ChatGPT就应该被我们当成提高学习和工作效率的工具。这一年半以来，我相信越来越多的人已经开始使用它，并逐渐把它当成一个必不可少的辅助工具或人工智能助手。这个工具会在全球范围内引起这么大的反应甚至争议的主要是因为它的未来发展方向，既最终成为通用人工智能（AGI）的可能性。AGI目前可能还没有一个大家公认的定义，它大致已经可以理解为“可以在一系列任务上与人类相匹配或超过”的人工智能。(https://www.technologyreview.com/2023/11/16/1083498/google-deepmind-what-is-artificial-general-intelligence-agi/)
我们人类大概是挺难接受一个工具超过我们自身的智能的。我们早已经非常习惯于在智能方面远远地站在所知的任何生物和事物的顶端了，虽然在其它很多方面我们自己创造的工具其实早已超过了个体人的能力了，比如起重机和收割机等。即使是我们每天使用的计算器和手机，它们的某些能力，比如说存储信息的能力也超过了人。
然而我们若是在通用智能上被超越那就显得很可怕了，驾驭或者说努驭惯了的人类突然有了被努驭的危险。
目前来看，ChatGPT4的有些方面的智能确实可以跟人类匹配了，比如说它参加法律和生物奥林匹克竞赛的成绩已经大大超过了大部分常人了。（https://www.coursera.org/articles/chat-gpt-3-vs-4）
如果这么一个接近智能的工具能被我们有效的使用，那么可以想象我们每个人的学习和工作的能力和效率都会极大提高的，因为我们每个人的大脑一下子增加了极大的存储、搜索、计算和创造的能力。我想正因为如此全球范围内现在很多人已经越来越多地把它变成自己工作、学习和生活中不可或缺的部分了。
前两天当我正在使用ChatGPT工作的时候经历了一次它突然罢工的事件。当时我的感受可以说是相当焦虑的，就好像我的一个得力下属突然离职不干了似的。我不仅反复地尝试重新启动这个工具，还在网上疯了一样到处找解决方法。最后发现在OpenAI的论坛里竟然有很多人在抱怨ChatGPT突然不工作的问题，有人说突然没有它不知道怎么工作了，有人说哈哈让我们重新学习一下用脑子工作吧，也有人说马上有个植入了ChatGPT的商业展示活动，也有人说要OpenAI退费。
在近三个小时的ChatGPT罢工过程中，论坛里可以看到抓狂、愤怒、沮丧、担心、无奈和期待等等各种情绪样样都有。这些来自于全球各地的用户应激反应说明了这个工具已经成为很多人一刻不能离不开的必需品了。
我在学校工作，因此思考得最多的还是人工智能对教育教学的影响。Salman Khan在他的新书New Brave Words里描绘了一个个性化的AI辅导老师Khanmigo，它可以随时和学生关于任何话题进行答疑解惑，并记录下所有过程性的资料，为教师和家长的介入提供资料，也提供了安全性保障。Khanmigo还可能可以为教师针对指定的课题创造教学计划和素材。它可能是目前第一个已经投入使用的基于ChatGPT的学生学习和教师教学的辅助工具。
这些AI新工具的出现肯定是在倒逼学校里的教师思考如何改变现在的工作方式，包括如何备课、如何设计作业和考试、以及如何为学生提供个性化的指导等等。但是就我目前的经历来看，老师们跟我讨论得最多的还是如何防止和甄别学生用ChatGPT作弊的问题。我觉得绝大部分教师可能还处在观望或被动接受人工智能已经来到我们身边的状态。不夸张地说很多老师也可能还不知道最新版本的ChatGPT到底具备了哪些超过了他们想象的能力。
ChatGPT能做数学这件事是让我很惊讶的，虽然它有时还会出错。不过我们每个人不是也会出错吗？我相信因为大语言模型 (LLM, Large Language Models)的学习能力和速度，很快ChatGPT做数学题的准确率会大大提高，甚至超过大部分一般水平的学生和教师的。也是前几天，我还在中国的某个网站上截屏了一个有相当难度的有关集合的数学题，把图片上传到ChatGPT对话框要求它给出答案。十几秒之后，一个正确且完整的分析和解题过程就呈现在了我的面前。
几个月前我曾经参加过一个加拿大的教师培训课程，其中有不少课堂教学设计的作业。若是以前的话，我肯定会化更多的时间在收集资料和考虑初步设计框架这个步骤上。然而现在这个工作完全由ChatGPT替代了，既节省了我的学习时间，又让我获得了更加全面的信息和角度。在AI所提供的初稿的基础上，我再根据自己对课题的认知和所需教学设计的场景进行最后修改和完善，大大地提高了我的学习效率。顺便说一句，我的每个作业和最后课程的分数平均都是A以上。
在我的平时工作中ChatGPT也开始显示出它的辅助作用。比如，它曾经帮我设计过调查问卷、翻译过文章、制定过学校政策等。平时让它帮助搜索和整理一些我所需要的信息则已是家常便饭似的事情了。我也在尝试用它制作一些教学计划，以及学生测试和作业。虽然我所得到的初稿目前还远没有达到我满意的程度，但是AI在这些方面的潜力，尤其是这种通过不断地提示(prompt)使得结果越来越个性化，越来越接近用户期望的方式让我对它的未来能力抱有极大的希望。这个过程似乎有点像逐步地引导一个新招聘的人类助手，逐渐对工作场景和细节增加认知，工作的效率和成果也就越来越符合我的要求了。
ChatGPT以及类似的工具虽然并不是为了教育而创造的，但是它们在教育中将有很多应用可能性，也不可否认地已经给教育产生了深远的影响。然而，从我的观察来看，对于这个潜力非凡的新工具很多教师学习和使用其的意愿可能太低了，速度也可能太慢了。
当AI正在成为一个被广泛使用的工具，而你却被动观望或者拒绝接受的话，那结果就是变成如以色列历史学家尤瓦尔所说的与历史进程“不相关”了。做个不怎么太恰当的类比，就像支付宝或者微信支付，如果一个生活在中国大陆的人到现在还不会使用它们的话，那他很可能就被这个丰富而繁忙的商业社会抛弃了。
所以，我觉得每一个现在正在从事或者以后希望从事教育教学工作的教师都应该从被动接受或观望AI，转变为积极主动地学习和使用这个技术和相关工具。Khan在他的书的最后呼吁人们应该对AI有一种他称为 “受过教育的勇敢 （educated bravery）”的态度。他写道，“人工智能的海啸已经从岸边拉起，现在它正向我们奔涌而来。面对逃避或迎接的选择，我相信应当全力投入，同时采取适当的预防措施，以免被裹挟其中的漂浮物击中。”
作为一个国际教育的教师和学校管理者，我想我应该首先采取行动。我不仅频繁地使用 ChatGPT ,也时刻保持对AI技术，尤其是它在教育教学领域发展的关注。更重要的是，我想成为一个AI技术与教育教学结合的实践者和推广者。
我最近把我的个人网站名字改成了 www.edtinker.com，以后我会专门解释我组合并使用edtinker这个词背后的思考。我在想这个网站是不是可以成为一个促进教师对人工智能及其对教育教学影响和作用认知的社区。如果可能的话，我想让这个网站慢慢发展成教师分享如何在教育教学工作中使用AI的平台。我的这些想法肯定也是在“黎明”阶段，但我希望能够吸引很多有同样的兴奋感和紧迫感的教师来加入我，我们一起成为走入AI黎明的教师。
你可以登录我的网站，或通过网站首页上的电子邮件与我联系。我希望能够很快找到一些志同道合的教师一起讨论和建设这个网站平台。
English version (translated by ChatGPT 4o, slightly edited by Leo Huang)
Walking into the Dawn of AI as a Teacher
I have recently been using ChatGPT frequently and have read many articles and books about artificial intelligence and ChatGPT. I feel like I have gained some understanding and experience, so I want to talk about this topic with everyone I meet and discuss the potential impact of AI on our work and study. I believe the influence has already begun, and now everyone is just wondering how deep and far-reaching this impact will be.
In her autobiography "The World I See: Curiosity, Exploration, and Discovery at the Dawn of AI," Fei-Fei Li states, "I believe our civilization stands on the cusp of a technological revolution with the power to reshape life as we know it." The subtitle of the English version includes "at the dawn of AI," which is why I used the word "黎明" (dawn) in the Chinese translation of the title.
Coincidentally, Salman Khan, the founder of Khan Academy, in his new book "The New Brave Words: How AI Will Revolutionize Education (And Why It's a Good Thing)," also expresses that artificial intelligence is currently in its early stages of application in education. When discussing the auxiliary role of AI in teaching, he also used the word "dawn," portraying the "dawn of AI assistants." Another chapter title in the book is "The Rise of AI Tutors," using the word "rise," which, based on the content, also conveys the idea of an early-stage phenomenon. Personalized AI assistants are indeed the main theme of Khan's book.
My explanation of these two English words is not to discuss translation methods but to emphasize the uniqueness of our current era. Considering the development trends of AI, especially ChatGPT, it seems likely to become a general-purpose technology like the steam engine, electricity, internal combustion engine, and computers, which enhance productivity across all economic sectors.
Therefore, I want to instill a sense of excitement and urgency in the teachers or other professionals reading this article. The excitement comes from potentially witnessing the early rapid development of a great technology, and the urgency stems from the unique nature of this technology compared to its predecessors. It not only closely rivals human intelligence but also develops at a speed beyond our imagination.
Reflecting on her 30 years of study and research, Fei-Fei Li observes that before the second decade of the 21st century, it took humanity over half a century to mature the prerequisites for artificial intelligence. However, in less than five years afterward, AI has rapidly transformed the world. She writes, "Businesses were transformed, billions of dollars were invested, and everyone from industry analysts to political commentators to philosophers was left scrambling to make sense of a technology that had seemed to explode, overnight, from an academic niche to a a force for global change."
Since the release of the first version of ChatGPT in November 2022, the world has undoubtedly changed. Some jobs have already disappeared or are disappearing, such as basic translation and programming. New fields are emerging or have already emerged, such as how to create and train personalized GPTs and teaching people how to effectively use this tool. More and more companies are either integrating AI technology or developing based on it altogether.
Khan Academy recently launched an AI tutor based on ChatGPT. Salman Khan explained why they adopted AI, saying this Pandora's box has been opened, and it's impossible to revert the world to its original state.
On the frontlines of education, teaching, and school management, I have felt the impact of ChatGPT since its inception. Like many colleagues, my initial concern and even fear were, "Oh no, students now have a super cheating tool." Especially for writing assignments, the articles teachers see might be collaboratively completed by students and ChatGPT. The definition of cheating had to be revised. Students might not be copying existing articles but simply seeking help from a powerful assistant. Salman Khan compares it to getting help from a highly knowledgeable sibling at home.
Today's plagiarism detection tools, such as Turnitin, only provide reports indicating the probability of AI involvement in students' assignments. Ultimately, teachers must judge based on their understanding of the students. This change clearly sets higher demands on teachers, as making judgments that lie between objective, subjective, and intuitive is extremely difficult. Teachers need to understand what AI is and know their students better. I feel some teachers might completely rely on the probabilities given by Turnitin, which, incidentally, are likely calculated by AI! So, who exactly is relying on AI here? A teacher jokingly said to me, "Then we might as well use AI to grade students' assignments since students are using AI to write them."
Admittedly, even the highest version of ChatGPT 4 is still developing. Although it might progress quickly, it is still some distance away from the so-called AGI (Artificial General Intelligence) that people anticipate. "ChatGPT might make mistakes," is clearly written under the ChatGPT dialogue box. Humans make mistakes though, so this reminder seems a bit redundant. However, this reminder at least keeps us aware that we cannot fully rely on this tool and must not use its answers or designs unthinkingly.
I recently almost got fooled by ChatGPT. I was discussing a senior year high school permutation and combination problem with a math teacher. We had differing methods, so we tried letting ChatGPT solve it. The result matched one of our methods and was well-reasoned. However, upon further discussion, we realized ChatGPT used an incorrect method. I informed ChatGPT of its mistake, and it obediently redid it but still made the same error. When I further explained the reason for the mistake, trying to educate it to correct itself, it spit out several pages of gibberish, like a person rambling incoherently when suddenly pointed out a mistake. This example showed me that ChatGPT is still not "thinking."
Nonetheless, ChatGPT should be considered a tool to improve learning and work productivity. Over the past year and a half, I believe more and more people have started using it, gradually treating it as an indispensable assistant or AI assistant. The significant global response and even controversy surrounding this tool is primarily due to its potential future development towards becoming AGI. AGI currently lacks a universally accepted definition but can be roughly understood as "artificial intelligence that can match or exceed human abilities across a range of tasks."
Humans might find it difficult to accept a tool surpassing our intelligence. We are accustomed to being at the pinnacle of known intelligence, although in many other aspects, the tools we have created have long surpassed individual human capabilities, such as cranes and harvesters. Even the calculators and smartphones we use daily have certain abilities, like information storage, that exceed human capacity.
However, being surpassed in general intelligence seems frightening. Humans, who are used to controlling and commanding, suddenly face the danger of being controlled or even slaved.
Currently, some aspects of ChatGPT4's intelligence indeed match humans. For example, its performance in law and Biology Olympiads far exceeds that of most people.
If such an intelligent tool can be effectively used, it is imaginable that each person's learning and work capabilities and efficiency would be greatly enhanced, as our brains would suddenly gain immense storage, search, calculation, and creation capabilities. Therefore, many people globally are increasingly making it an indispensable part of their work, study, and life.
A few days ago, while working with ChatGPT, I experienced an outage. My anxiety was comparable to that of a competent subordinate suddenly quitting. I repeatedly tried restarting the tool and frantically searched online for solutions. Finally, I found many people complaining about ChatGPT's sudden malfunction in OpenAI's forum. Some said they didn't know how to work without it, some joked about relearning how to use their brains, some mentioned upcoming commercial demonstrations using ChatGPT, and some demanded refunds from OpenAI.
During the nearly three-hour ChatGPT outage, the forum was filled with various emotions—panic, anger, frustration, worry, helplessness, and anticipation. These reactions from users worldwide show that this tool has become an essential part of many people's lives.
Working in a school, I think most about the impact of AI on education and teaching. In his new book, "New Brave Words," Salman Khan depicts a personalized AI tutor named Khanmigo that can answer students' questions on any topic at any time, recording all procedural data to provide information for teachers and parents' involvement and ensuring safety. Khanmigo can also create lesson plans and materials for teachers on specific topics. It might be the first ChatGPT-based student learning and teacher teaching assistant tool in use.
These new AI tools inevitably force teachers to rethink how they work, including lesson planning, assignment and exam design, and personalized student guidance. However, based on my current experience, teachers mainly discuss how to prevent and identify student cheating with the help of ChatGPT. I feel most teachers are still in a state of observation or passive acceptance of AI's arrival. It is not an exaggeration to say that many teachers may not even know the abilities of the latest version of ChatGPT.
The fact that ChatGPT can do math surprised me, although it sometimes makes mistakes. But don't we humans all make mistakes? Given the learning ability and speed of Large Language Models (LLM), ChatGPT's accuracy in solving math problems will soon greatly improve, surpassing most average students and teachers. A few days ago, I took a screenshot of a relatively difficult math problem about sets from a Chinese website, uploaded the image to the ChatGPT dialogue box, and asked for an answer. Within seconds, a correct and complete analysis and solution appeared before me.
A few months ago, I attended a teacher training course in Canada, which included many lesson design assignments. In the past, I would have spent more time collecting information and considering the initial design framework. Now, this step was entirely replaced by ChatGPT, saving me study time and providing more comprehensive information and perspectives. Based on the initial drafts provided by AI, I then modified and refined them according to my understanding of the topic and the context of teaching, greatly improving my learning efficiency. By the way, my assignment and final course scores were all above A.
In my daily work, ChatGPT also shows its help as an auxiliary role. For example, it has helped me design surveys, translate articles, and develop school policies. Regularly, it helps search and organize the information I need, which has become routine. I am also trying to use it to create lesson plans, student tests, and assignments. Although the initial drafts I get are not yet satisfactory, AI's potential in these areas, especially the ability to increasingly personalize results through continuous prompting, gives me great hope for its future capabilities. This process is akin to gradually coaching a newly hired human assistant to increase their understanding of work scenarios and details, with the efficiency and results increasingly meeting my requirements.
Although ChatGPT and similar tools were not created for education, their potential applications in education are vast and undeniably have had a profound impact. However, based on my observations, many teachers are reluctant or slow to learn and use this powerful new tool.
When AI is becoming a widely used tool, and if you remain passive or refuse to accept it, the consequence could be, as Israeli historian Yuval Noah Harari put, "irrelevant" to the historical progress. To draw an imperfect analogy, it's like Alipay or WeChat Pay. If someone living in mainland China still doesn't know how to use them, they are likely to be left out of the rich and busy commercial society.
Therefore, I believe every teacher currently working or aspiring to work in education should shift from passively accepting or observing AI to actively learning and using this technology and related tools. In his book's conclusion, Khan calls for an attitude he terms "educated bravery" towards AI. He writes, “The AI tsunami has drawn back from the shore, and it is now barreling towards us. Faced with the choice between running from it or riding it, I believe in jumping in with both feet, while taking proper precautions so that we don't get hit with the flotsam.”
As an international education teacher and school administrator, I think I should take the lead. I not only frequently use ChatGPT but also constantly keep an eye on AI technology, especially its developments in education. More importantly, I want to be a practitioner and promoter of combining AI technology with education and teaching.
I recently changed my personal website name to www.edtinker.com. In the future, I will write about my thought behind combining and using the term "edtinker." I want to use this website platform to promote teachers' understanding of AI and its impact on education and teaching. Additionally, if possible, I hope this website gradually evolves into a platform where teachers can exchange and share their experiences of using AI in their educational work. These ideas are also surely in the "dawn" stage, and I hope to attract many teachers who share the same excitement and urgency to join me so that we can together become teachers stepping into the dawn of AI.
You can Log In my website or Keep In Touch with me by email at the front page of the site. I hope soon I can find some like-minded teachers to discuss and build this website platform together during our spare time.
Comments