文 | 脑极体
表妹就读的高中在去年秋天引进了两台人形机器人，一台搁在教学楼大厅做迎宾，一台在科技节上跳过一段舞。学校公众号发过推文，标题里带着“科技感拉满”，照片里机器人披着红色绶带，旁边围了一圈好奇的面孔。三个月后，表妹对这两台铁家伙的评价只剩一句：“笨笨的，动作迟缓还挡路。”
走廊不算宽，课间上百人涌出来，机器人杵在拐角，移动迟缓、避障迟钝。学生们起初还围着看，后来便改成了绕道走。“高峰期是我们避让它，很耽误抢饭的时间。”表妹说，“有一回一个男生跑得太急差点撞上去，机器人也只是僵硬地挪动了一下”。
这件事不免让人感到荒诞，一台被寄予“科技启蒙”厚望的机器在日常校园生活里最稳定的功能居然是制造通行障碍。他并没有让同学了解到具身智能的技术逻辑，也没有成为科技课堂的实验帮手，反而成了需要小心珍藏的“麻烦制造者”。
如果把视线从这一条走廊拉开，会发现类似的故事在过去一年里密集上演。机器人以“新同学”“迎宾员”“表演嘉宾”的身份扎堆涌入中小学校园。
下一所学校，下一场典礼，同一批机器，同一套动作……机器人到底为什么要扎堆进入校园呢？智能化进校，到底是不是一件有意义的事？
2026年春季开学季，人形机器人和机器狗几乎成了各地中小学开学典礼的标配。
多所中小学校把“春晚同款”请进了校门。麓城外国语小学的操场上，人形机器人作为“领跑员”带着学生跑步；松岗第二小学门口，机器狗作揖握手，被孩子们围住合影；龙岗区一所学校一口气请来七位“机器人嘉宾”，写福字的、耍功夫的、跳街舞的各司其职。绵阳一所乡村小学的开学典礼上，机器人披着红色绶带在校门口迎宾，又和老师合唱了一首《奔跑》，孩子们簇拥着它走进教室，“好奇地观察、兴奋地交流”。上海、佛山、济南、南京，类似的场景在同一周内反复出现。
场面热闹，来路却大多潦草。这些机器人中的大多数是租来的外援、并非学校资产。一台能完成舞蹈、互动动作的人形机器人，日租金在数千元量级；学校通常需要租两台以上，租期两天、彩排一天、正式演出一天，在开学典礼上完成一套挥手、作揖、跳舞的动作，然后被装车拉走。两台机器两天的开销，足以覆盖一所普通小学一个学期的社团活动经费。如果是偏远地区的学校，这笔钱的分量更重。
不过，学生站在台下看机器人跳舞与在科技馆看机器人跳舞，在体验上并无本质差异。甚至科技馆的展项更丰富，有讲解，有互动，还不需要承担设备运输和现场调试的风险。把机器人租到学校里表演，本质上和请一个表演团队来撑场子在教育逻辑上区别不大。
有校长会辩护：“沉浸式体验带来的思维启蒙不能用钱衡量。一个从未见过机器人的乡村孩子，第一次伸手摸到机械臂，那种触感可能埋下一颗种子。”这话有道理，但种子要发芽，需要后续的土壤。
如果典礼结束，机器人被拉走，教室里的一切照旧，那颗种子大概率停在摸过的层面。表妹学校那两台区里配发的机器人因为没有配套课程和师资最终退化成走廊摆设，就是现成的例证。配发的尚且如此，租来的更不必说。
这是一种基于追赶热潮的结构性仓促：技术尚未成熟，课程尚未配套，师资尚未准备，但仪式已经等不及了。
表妹觉得机器人“笨笨的”，这个看法与行业现状高度吻合。
2026年4月，新疆一所高校的运动会开幕式上，一台参与表演的人形机器人突然失控，倒地后开始朝周围学生胡乱踢打，最后被工作人员抱离现场。同月，西安一场校园活动中，表演机器人突发故障“抱住”了同舞的女生。
这些视频在社交平台上流传时，评论区最常见的一句话是：“还好没出事。”
事故并非偶然。人形机器人目前所处的技术阶段，用“学步期”来描述是准确的。北京首钢园有一个被媒体称为“机器人幼儿园”的训练场地，十五台不同形态的机器人按照能力分班，从软硬件适配到本体感知再到社会化协作，逐级训练。具体看，机器人走路会踉跄、会停顿、会卡住，偶尔摔倒、互撞，工作人员得时不时把它们“抱”出场地。入园标准里有一条是皮实抗造，还要加装柔软外壳，“为减少跌倒或人机交互时的伤害”。
换句话说，在行业最前沿的训练场里，人形机器人连稳稳走路这件事都还在靠一次次摔跤来迭代。把这些尚在“幼儿园”阶段的机器，放进几千名中小学生中间，指望它们承担迎宾、引导、表演的功能，本身就是一种错位。
更深层的问题在于，这种错位被“科技感”的修辞掩盖了。一台在开学典礼上完美跳舞的机器人，展示的是编排好的、经过反复调试的成功动作。它不暴露失败、不呈现调试过程、不让学生看到“机器为什么会笨”。学生接收到的是一个成品，而不是一个可被理解、可被质疑、可被改造的技术对象。
而这种草台班子样的表演做法与科技教育的目标恰好相反。
科技教育想让学生获得的，说到底是对技术的理解力，知道一台机器怎么动起来、为什么会失灵、哪里可以改。可表演的逻辑是不许出岔子，出了岔子就赶紧把机器人抱走，异常成了要遮掩的事故。学生的位置也成了站在安全距离之外的观众，鼓掌或者受惊，而不是动手拆解、追问、改造。
而这反而会让学生离科技越来越远。
如果说具身智能还处在蹒跚学步的阶段，那AI最好完全不进入校园吗？
如果要把“智能是否值得引入学校”作为一个可判断的问题，标准不在于它是否足够新奇、足够上镜，而在于它是否进入了课程结构，是否给了学生“折腾”它的空间。一台能被学生写代码、调参数、反复试错的机器人，哪怕走得摇摇晃晃，它的笨拙本身就是教学素材。一台只能在台上完美跳舞、然后被装箱拉走的机器人，再流畅也只是消费品。
就具身智能而言，北京经开区2026年5月提供了一个值得参照的样本。他们把机器人足球赛上“退役”的二十台人形机器人交付给区内五所中学，希望融入人工智能、机器人等校本课程及社团活动，开发专项教学模块，鼓励学生围绕其开展项目式学习与研究。接收学校的一位副校长曾提到：“期待在一线教师的带领下，这些人形机器人可以助力学生二次开发。”
二次开发与观看表演之间隔着一条清晰的分界线。前者意味着学生要写代码、调参数、面对失败、反复试错。一台走不稳的机器人，在表演场景里是缺陷，在开发场景里是素材。它越笨拙，学生调试的过程就越有教学价值。后者则只需要学生坐在台下，在适当的时候鼓掌。
这个逻辑同样适用于机器人之外的智能技术。大模型进入校园的速度比人形机器人更快，风险也更隐蔽。智能批改系统可以把教师从重复劳动中解放出来，智能问答可以让学生通过趣味化、一对一的方式了解学科知识，趣味智能游戏也能激发学生的学习兴趣、寓教于乐。
值得注意的是，这些案例里AI扮演的角色都不是替学生完成，而是把学生从低效环节里解放出来，让他们去做更值得做的事。批改、数据采集、初步反馈这些机械性环节被机器接走，学生和教师省下的注意力投向设计、追问和迭代。
不过，眼下涌进中小学的这批机器人大多属于展示性技术，热闹归热闹，并没有真正落到教育价值上。学校真正该引入的是能进入课程结构、允许失败和重来的技术。
两万块一场的科技感换来的是一篇推文和几张照片；两万块一套的机器人套件换来的是一个学期里几十次失败、几十次重试。前一笔钱花完就结束了，后一笔钱花完，学习才刚刚开始。
本文由 AI宝库自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Man, the brain.
The high school for cousins introduced two man-shaped robots in the fall of last year, one in the lobby of the school building as a guest, and one in a dance at the technology festival. The school, WeChat Official Account, sent a tweet with the title “Technology Full” in which the robot was wearing a red belt and surrounded by curious faces. Three months later, the cousin said to the two irons only one sentence: "Imbecile, slow, slow."
The corridor is not wide, hundreds of people are coming out of the classroom, robots are standing around the corner, moving slowly, avoiding too much. The students initially looked around, and then they turned to detours. "The peak period is when we avoid it, and it's too late for food." The cousin said, “One time a boy ran too fast and almost hit it, and the robot just moved in a rigid manner”.
This is a ridiculous thing to do, and the most stable function of a machine that has been entrusted with “scientific enlightenment” in daily school life is to create obstacles to movement. He has not taught his classmates the logic of smart technology, nor has he become an experimenter in the science class, but rather a “problemmaker” who needs to be carefully hidden.
If the sight is drawn from this corridor, similar stories will be found to have taken place intensively over the past year. The robots piled up as “new classmates” and “visiting guests” into primary and secondary schools.
The next school, the next ceremony, the same machine, the same move... why would the robots get together to get to the campus? Is smart going to school a meaningful thing?
In the spring of 2026, human robots and robotic dogs became almost the standard for the opening of primary and secondary schools.
Many primary and secondary schools have invited “spring and night” to the school. In the playground of the foreign language primary school in Changcheng, human-shaped robots run as “runners” with their students; at the entrance to the second elementary school of Matsuoka, the machine dog shakes his hand and is surrounded by children; seven “robots guests” are invited to a school in the Yongchang district; and the responsibilities of the masters who write, play kung fu and dance in the streets. At the opening ceremony of a village primary school in Myeongyang, the robots were greeted at the entrance to the school with a red ribbon and sang with the teacher "Running" and the children came into the classroom with it, “Inquisitively observing, excitedly communicating”. In Shanghai, Foshan, Chinan, Nanjing, similar scenes were repeated during the same week.
There's a lot of excitement, but most of the way is rough. Most of these robots are rented foreign aid, not school assets. A human robot capable of performing dance and interactive movements is rented on a daily basis at thousands of dollar scales; schools are usually required to rent more than two units for two days, a rehearsal day, an official performance of one day, and to complete a wave of handing, dancing and dancing at the opening ceremony, and then to be pulled out of the car. The cost of two machines for two days is sufficient to cover the cost of community activities for one semester in a regular primary school. In the case of schools in remote areas, the amount is more significant.
However, there is no fundamental difference between the experience of students standing under the stage watching a robot dance and a robot dancing in a technology museum. Even the exhibition in the Science and Technology Library is more informative, informative and interactive and does not have to bear the risk of equipment transport and on-site commissioning. It's not very different to rent a robot to perform in school, in essence, to have a performance team support the field.
There are principals who argue: “The idea of immersion cannot be measured in money. A country child who has never seen a robot reaches out to his arm for the first time, and that touch may have buried a seed.” That makes sense, but seeds need follow-on soils to germinate.
If the ceremony is over, the robot is pulled away, everything in the classroom is the same, and the seed probably stops at the touch level. The distribution of robots in the two districts of the cousin school is a case in point because of the absence of a curriculum and the eventual deterioration of teachers into corridors. The distribution is still so, let alone rented.
This is a structural hasty process based on a rush to catch up: technology is not yet ripe, curricula are not yet available, teachers are not yet ready, but rituals cannot wait.
Cousins think robots are “mute”, a view that is highly consistent with the state of the industry.
In April 2026, at the opening of a sports conference at a university in Xinjiang, a human robot involved in the performance suddenly went out of control, fell down and started kicking around students indiscriminately, and was finally held by staff members from the scene. In the same month, at a school event in Sian, a girl who performed a robotic failure “held” the same dancer.
The most common sentence in the comment area, when these videos are being distributed on social platforms, is: "Well, it's okay."
Accidents are not accidents. The technical phase in which human robots are currently in is precisely described as a “step-by-step” period. In Beijing's first steel park, there is a training site called the “Droid Kindergarten” by the media, with 15 different forms of robots organized according to their capabilities, from soft and hardware adaptation to in-house awareness to socialization collaboration, and training by level. Specifically, robots walk, pause, jam, occasionally fall, crash, and staff “take” them out of the field from time to time. One of the criteria for admission to the park is dermal resistance and a soft and soft shell “in order to reduce the harm of falling or human interaction”.
In other words, in the training fields at the forefront of the industry, human-shaped robots are still walking on one fall. To put these machines, which are still in the “kindergarten” stage, among thousands of primary and secondary students, it is wrong to expect them to assume the functions of hospitality, guidance and performance.
The deeper problem is that this misalignment is masked by the rhetoric of “technology sense”. A robot that dances perfectly at the beginning of the school ceremony, showing well-organized and repeatedly tested successful moves. It does not expose failure, does not present debugging processes, does not allow students to see “why machines are stupid”. Students receive a finished product rather than a technical object that can be understood, challenged and transformed.
And this grass-roots performance is the opposite of the goal of science and technology education.
Science and technology education seeks to give students an understanding of technology, how a machine works, why it fails, where it can be changed. But the logic of acting is not to go astray, but to get rid of the robot as soon as it goes wrong, it becomes an accident to hide. The students are also seen outside safe distance, applauding or shocked, rather than breaking down, asking questions and transforming.
And it's going to keep students away from technology.
If intelligence is still in the learning phase, would it be better for AI not to go to school at all?
If “intelligence is worth introducing” to school as a question to be judged, the criterion is not whether it is new enough to be mirrored, but whether it enters the curriculum structure and gives students the space to “scratch” it. A robot that can be coded, calibrated, mistested by students, even if it moves, is self-defeating and is teaching asset. A robot that can only dance perfectly on stage and then be boxed out, and then flowing is a consumer.
In terms of intelligence, in May 2026, Beijing's opening region provided a reference sample. They have handed over 20 man-made robots “retired” from robotic football to five secondary schools in the region, hoping to integrate into their curricula and community activities, such as artificial intelligence, robotics, and the development of special teaching modules to encourage project-based learning and research around them. One of the vice-presidents of the receiving school mentioned that “it is expected that, under the leadership of first-line teachers, these morphological robots can assist students in redevelopment”.
There is a clear line between the secondary development and the performance. The former means that students write codes, adjust parameters, face failures, and repeat mistakes. An unstable robot is a flaw in the performance scene, and in the development scene is asset. The more clumsy it is, the more pedagogic the student debugging process. The latter would only require students to sit under the table and applaud in due course.
The same logic applies to intelligent technologies other than robots. LLM enters the campus faster than human robots, and the risks are more hidden. The smart adaptation system can free teachers from repeated work. Smart questions and answers can enable students to learn about their subjects in a fun, one-on-one way. Interesting smart games can also stimulate students ' interest in learning and fun.
It is worth noting that in these cases AI did not play a role for the students, but rather liberated them from the inefficiency and made them do more worth doing. The mechanical components of rectification, data collection and initial feedback were removed from the machine, and the students and teachers saved attention to design, inquiry and iterative.
However, most of the robots that now rush into primary and secondary schools are display technology, and they are popular and not really of educational value. What really needs to be introduced in schools is technology that can enter the curriculum structure, allow failure and re-emergence.
The 20,000-dollar technological sense is a tweet and several photographs; the 20,000-dollar robot kit is replaced by dozens of failures and dozens of retrys over the semester. The first one is finished, the latter is spent, and learning is just beginning.
Auto-fetched and translated by AI Gem Box. All rights belong to the original author.