文 | 定焦One，作者 | 金玙璠、王汉星、雷晶、王璐、陈丹，编辑 | 王汉星
2026年还剩两个多月，大厂打工人也马上要熬过被AI支配的一年。
年初，黄仁勋的一句“如果一名年薪50万美元的工程师，一年的Token消耗不到25万美元，会令我深感不安”，带起Tokenmaxxing浪潮，所有人都在用AI，都在烧Token，似乎Token用得少了，工作就没做到位。
到了6月，大厂发现Token有点烧不起了，又开始转而关注AI使用的回报率，这不仅没有降低大厂员工的AI FOMO（错失恐惧）情绪，反而让员工更难通过使用AI来“取悦”老板，他们要证明AI正在帮助自己的工作提效，交付的内容一定要有AI含量，还要证明自己消耗的每一个Token都在产生价值。
对于一部分大厂员工来说，AI能否真的提效是其次，更要紧的是要向团队和管理者展示“我已经在用了”。因为对他们而言，不会用AI不要紧，重要的是不要让老板觉得你不会。
为此，他们有人把“让AI把老板半小时的会议录音转成文字”写成“搭建语音信息结构化流程”，还被评为AI应用的优秀案例；有人专门找一些不重要的工作场景，做成Skill，再包装成一个听起来很厉害的名字给老板看；还有大厂PM（产品经理）用AI去干RD（研发工程师）的活儿，干完之后还要找RD同事重新再改一遍。
这种现象如今在海内外职场上越来越普遍。不久前，商业情报公司Visier参与的一项调查显示，在1000名受访的美国全职员工中，48%承认向同事或上级夸大了自己的AI使用频率或专业知识。
可以预见的是，随着AI在各种工作场景中的不断深入和普及，这种“表演”还会继续下去。
01.为了让周报有AI含量，十分钟的活我干了半小时
李昭 | 92年 杭州 电商公司产品运营
我在杭州一家电商公司做产品运营。今年三月，OKR要求我们做到“AI深度应用”。双月会上领导挨个问，你上周用AI解决了什么。答不上来的，会后主管会私聊。
三月到五月是真卷。组里有用量排行，行政每周导出数据发到群里。那会儿用AI不为干活，为的是榜上别掉队。我问排第一的姑娘怎么做到的，她说睡前让AI帮她写全年规划，跑完也不看。我学她，一到工位就让AI跑“竞品监控”，然后自己刷刷短视频，看看娱乐八卦，算是蒙混过关。
六月风向变了，不看用量，看产出。周报里要体现“AI提效案例”，要有数据。我开始琢磨怎么凑数。最滑稽的一次，我把“让AI把老板半小时的会议录音转成文字”写成“搭建语音信息结构化流程”，这条后来还被评为优秀案例。还有一次是这个月初，我拟一份渠道调价通知，原本十分钟的事，为了让周报有AI含量，我让AI起草，它给我编了好多数据，最后我翻出去年双十一的通知，在那版基础上改了改交上去。但周报里我还是写了“AI辅助撰写”。
凑数归凑数，活还得干，我慢慢摸出一套方法。工位两台显示器，左边跑AI，右边工作。方案类的活，把brief丢进去等框架，不反复调指令，它问什么我答什么。它跑第一块的时候，我写第二块的要求；等第二块跑起来，我回头检查第一块的结果，顺手把第三部分细化完，几个小时连轴转。AI也确实能提效。上周四十多条客服投诉，我让它按问题分类，十分钟出了张表。还有整理会议纪要，开完会大家刚走到工位，我已经把纪要发群里了。
但AI跑东西没准点，三十秒和一小时都有可能。我试过趁这工夫做点别的，改到一半它跑完了，切回去看结果，再切回来，经常忘了刚才做到哪一步。后来干脆刷手机，边刷边盯进度。那段时间我明显感觉专注力在下降，屏幕切来切去，一天下来很累，又说不出累在哪。我只好给工作分类，做A的时候坚决不用AI做B，润色这类小活攒到一起，每天抽一两个小时集中处理。
最近，部门开始让每人汇报“AI使用心得”。我看了领导在群里发的汇总文档，说综合了大家的心得。不用看就知道，都是AI总结的AI。
02.我写的Demo根本用不了，但领导特别满意
李默｜27岁 北京 互联网大厂产品经理
我是一家北京互联网大厂的产品经理，今年27岁，在这家公司干了两年。
年初公司开始要求全员搞AI、写Skill。最早波及的是研发岗的同事，后来就蔓延到了各个岗位。其实我之前也断断续续用过AI，写写文案、整理整理需求，属于“用用看”的阶段。
大约今年3、4月份，我们组领导找我谈，说以后PM如果不会写代码，很快就会被淘汰，要求我们都学着用AI写代码，一些简单的需求尽量自己解决，不要事事都找研发。
那之后我确实动手写过一些小demo，也做过一些小的UI设计。说实话，那些东西都不太能用，给到研发的同事，他们还得重新改一遍，等于我写得再热闹，最后还是人家兜底。但领导很满意。那段时间我明显感觉到，汇报交差反而比以前容易了，反正东西能不能落地是其次，“你在用AI”这个姿态先立住了。
6月以后，我们组变动很大，前端后端都有不小的调整，我的汇报线也换了。新Leader对AI产出和提效的考核细化了很多，像之前那样自己随便写写、再交给研发同事改改，已经交不了差了。
于是我换了一种“AI提效”的方式。我把需求反馈、市场调研、产品框架这些步骤都写成Skill，其实就是一些很简单的页面读取和整理，每个环节都让AI来处理，交付的东西基本都是AI直出。这里面其实真正有价值的东西不多，“表演”的成分更重。
一来，这些工作里很多根本不需要AI，我自己做也花不了多少时间；二来，AI给出的不少需求，同事拿到手其实不好推进，经常会发现逻辑错误，我们还得重新开会碰一遍。表面上看流程跑起来了，实际效率未必比以前高。
不过这么干也有好处，Leader觉得我们很忙，AI工作流也搭起来了，过程数据好看，汇报也好看。
我自己对这件事的判断是，AI确实能提效，尤其在写代码和写文档这两个环节，是真的省时间，但它不是总能提效，硬套上去反而添乱。而我个人觉得最大的问题，是Leader一旦觉得你效率提升了，就会想让你去干别人的活儿，提效的红利落不到自己头上，反而成了加活的依据，这就得不偿失了。
我还是觉得，各自做好自己擅长的事，在这个基础上把AI用好，才是更好的状态。不然公司这么搞下去，只是在散播AI焦虑、浪费资源，最后逼得大家都在演戏。
03.AI使用率不到70%扣绩效，我们卷出一堆没人用的Skill
小楠 | 95后 深圳 活动策划
我在深圳一家企业做活动策划。我们老板非常热衷AI，希望把它嵌入每个工作环节，要求大家尽可能用AI辅助完成工作，还鼓励我们建立自己的AI工作流。
后来甚至把AI使用率纳入了绩效考核：使用率达不到70%，绩效就要扣20%。
起初，大家还觉得没什么，毕竟AI确实能为我们提效。做一份活动策划方案，以前少说也要3天。现在有了AI，一天之内就能完成初稿。还有竞品分析、PPT制作等琐碎工作，AI也都能帮上忙，确实节省了不少时间。
但慢慢的，事情开始变得不对劲了。公司内部搞起了AI比赛，比谁能用AI覆盖更多工作场景、搭建更多工作流、制作更多Skill。只要能证明自己用AI打通了某个工作环节，就可以算作成果。
一开始，大家做出的东西确实有用，比如活动策划相关的Skill，能减少一些重复性的琐碎工作。
但有价值的场景毕竟有限，为了完成任务，大家开始卷数量。有人专门找一些不重要的工作场景，做成Skill，再包装成一个听起来很厉害的名字；还有人用AI做个小程序，汇报时展示自己又解决了什么问题、节省了多少时间。
老板看到这么多新工具，觉得大家使用AI的积极性很高。但实际上，有些东西对工作帮助不大，做出来之后根本没人用。
我感觉大家渐渐形成了一种微妙的默契。相比AI到底有没有帮上忙，能不能拿出看得见的成果更重要。毕竟，做出了多少Skill、覆盖了多少工作场景，很容易“伪装”，但究竟节省了多少时间、解决了多少实际问题，没那么容易衡量。
除了这些专门用来展示的AI成果，日常工作中也开始出现一种现象：AI在制造新的工作量。
有个同事用AI快速生成了一份活动方案，配上精美的PPT，拿去向老板汇报。老板看完很满意，觉得效率特别高，马上让下面的人按照方案推进。结果到了执行阶段，大家才发现方案里有不少漏洞，很多实际问题没有考虑进去。最后还是得靠人一点点补充、调整。
更让我无奈的是，AI的存在，还让修改方案的次数变得更多了。以前一份策划方案做完，领导看一遍，提些意见，改完就可以推进。现在领导自己看完，还会交给AI审核一遍。AI往往能提出一大堆修改意见，有些确实有价值，但有些不符合实际情况，我们还得逐条检查、解释和修改。
我觉得挺矛盾的。老板希望AI帮公司节省时间，我们却要花额外的时间参加比赛，证明自己确实在用AI。好不容易借助AI提高了效率，又要承担更多审核、修改和交付任务。这样看来，使用AI反而成了一笔不划算的账。
04.我们对着AI演戏：字正腔圆，微笑要听得见
王敏 | 95后 山西 软件开发公司销售
入职这家公司5个月，我算见识了什么叫“AI水土不服”。
公司主营业务是软件开发，已经成立了十几年，员工大概有100人。老板六十多岁，对AI的认知基本来自营销号。去年他提出AI转型，今年年初直接升级成年度战略。大半年下来，部分业务用AI后产出翻倍，但更多时候，是离谱。
先是员工的绩效考核新增了AI参与度，绩效评定要看AI使用占比。
于是大家开始各显神通地“凑AI”。比如，前天做了一个skill，昨天做了一次数据清洗，今天给客户从0到1做了全生命周期规划，AI占比从30%一路卷到80%。AI干没干活不重要，只要“参与过”就行。
以前写周报是写给老板看的，现在变成了一套循环流程，我把想法发给AI，AI写完周报，我检查一遍后发给老板，老板用AI提出几个问题发给我，我再发给AI，如此反复。我每周用AI干得最多的事，是让AI帮我编，这周用AI干了什么。
这还不够，老板还采用了一套AI监督系统。
他觉得员工日常对着电脑敲字做方案就是摸鱼，总说打字打不出业绩，要多打电话多沟通。于是AI客户管理系统上岗了，电话一拨自动录音，AI旁听、总结、给建议，结束生成日报。
现在员工在公司打电话跟“上镜”似的。说错一句后台记一笔，语气不好AI给建议，客户没聊成日报自动标红。老板没事就刷APP回放沟通记录，听到哪句不顺耳直接截图甩大群。每次看到群消息弹出，我们都后背发凉。
后来我们学会了怎么演给AI看，比如，说话字正腔圆，微笑要听得见，挂电话前先道谢。演得越好，AI评分越高。
AI给公司带来了多大提效，我不太清楚，反正有一次，AI助理把全公司的工资表暴露了。
某天老板开完会，说要搭建内部AI机器人，以员工为核心做AI化改造，解决资料共享、提升效率。人事同事照着网上教程，把考勤、绩效、工资表全传了上去，一天就搭完了。然后不知道谁在公司群问了句工资情况，AI直接把明细甩了出来。
这下炸了锅。有人问出同岗不同薪，有人查出绩效被压分，还有人把全公司薪资扒了个遍，查出三个工资照发、但从来没来上过班的名字。出乎所有人意料的是，老板没在第二天的总结会当众批评人事，反而夸她执行效率高。但散会后，老板联系AI公司技术加了权限限制，系统重新上线。
明明AI是个好东西，却越用越不对劲，我们原本指望它省力提效，结果反倒要演给它看。说到底，暴露的不是AI的问题，工资表泄露，是公司缺乏数据权限的基本常识。员工表演式上班，是老板把AI当成了新的监控工具。AI放大的是管理本身，流程顺，它就放大提效，猜忌多，它就放大内耗。
想明白这层，我反而释然了。即便没有考核，我依然离不开AI，做PPT、写规划书、整理文档，这些杂碎工作它确实帮我省了不少劲。但凡需要专业判断的内容，我绝不交给它。
05.领导要的不是提效，是“全员AI化”的样子
林舟 | 00后 广州 软件研发工程师
前几天，一个不认识的同事突然找到我，问我发布在内部插件市场的插件怎么用。我还挺开心的。毕竟这是我花时间研究出来的东西，有人主动使用，说明它确实解决了实际问题。
入职以来，我被分配了不少AI化任务。研究工具、开发插件、搭建工作流，想办法把日常业务中重复的环节自动化。这些事我其实很愿意做，也确实从中获得了成就感。
但我的+1（直属上级）似乎不太满足于此。
他的目标是让整个部门都用上AI，最好实现端到端的标准化流程。从个人使用到全员推广，从某个环节提效到建立完整的SOP，AI逐渐变成了一项必须完成的部门任务。
我有些困惑。自己摸索出来的工具确实好用，但每个人的工作内容和习惯不同，为什么非要用同一套流程？更奇怪的是，组内那些工作五年以上的老员工似乎并不着急。有老员工私下告诉我，他其实无所谓。
后来，我想明白了。老员工的核心竞争力在于业务经验和专业知识，AI对他们而言只是辅助工具，怎么方便怎么用，反正不与绩效挂钩。真正与AI挂钩的，是号召部门AI化转型的+2、积极响应的+1，以及倒霉被分配到具体任务的新员工，譬如我。
这也解释了为什么+1如此执着于建立SOP。
最近部门新来了几个外包同事。我隐约感觉，这个版本开发结束，他们可能就会离开。等下一个项目开始，又会有新人进来。人员流动如此频繁，最理想的管理方式自然是把工作流程标准化，让新人无需太多培训就能上手。人可以走，技能留下。
想通这一点，我才意识到，我和领导对AI的理解可能完全不同。
我认为AI的发展方向是个性化、可定制化。过去是人适应工具，现在应该是工具适应人，每个人都能根据自己的需求建立工作流。但+1似乎认为，AI可以把一切业务变成SOP。说难听点，我觉得他的想法一半是许愿，一半是表演。真正提高多少效率未必最重要，重要的是让上级看到整个部门正在AI化转型。
当然，从管理角度看，这套逻辑完全合理。通过AI降低培训成本、减少对个体经验的依赖，都是组织的正常诉求。
只是我原本以为，自己在帮助同事减少重复劳动，后来才意识到，我可能也在帮助公司降低对员工的依赖。更微妙的是，这件事与我的个人利益并不冲突，甚至对我有利。我能学习新技术、积累项目经验，也能获得实际的成就感。
我依然喜欢做AI提效，只是突然看清管理层可能的目的后，心里多了一层芥蒂。这下我终于理解，有人说道德感太高不适合本司。最难受的不是被迫做一件坏事，而是明明在做自己喜欢的事，却发现它同时服务于一套自己不认同的逻辑。
*应受访者要求，文中李昭、李默、小楠、王敏、林舟为化名。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Wen, focus on One, by Kim, Wang Han, Lui, Wang Jing, Chen, editor Wang Han Star
More than two months remain in 2026, and workers in large factories will soon survive the year at the disposal of AI.
At the beginning of the year, Huang In-hoon's sentence, “If an engineer with an annual salary of half a million dollars, Token spends less than $250,000 a year, I would be deeply disturbed” led to a wave of Tokenmaxing, where everyone was using AI, burning Token, as if Token was running out of space.
By June, the major factory discovered that Token could not afford to burn and began to focus on the return on the use of AI, which not only did not reduce the mood of the large factory’s employees at AI FOMO, but made it more difficult for employees to “please” their bosses by using AI, who had to prove that AI was helping to make their work work work, that the content of the delivery had to have an AI content, and that every Token they consumed was producing value.
For some of the major plant employees, the second most important thing is to show the team and the manager that “I am working”. Because for them, it doesn't matter to use AI. It's important not to let the boss think you won't.
To that end, some of them have written "Let's get the boss's half-hour recording translated into text" into "Structuring voice messages" and have been rated as a good example of AI application; some have looked for unimportant work scenes, made Skill, and packaged them into a name that sounds great to the boss; and some of the major plant's PM (product manager) uses AI to do RD (research and development engineer) work, and then to re-engineer with RD colleagues.
This phenomenon is now becoming more prevalent in the world of work, both at home and abroad. A recent survey involving the business intelligence company Visier showed that 48 per cent of the 1,000 full-time United States employees interviewed admitted to exaggerating their AI frequency or expertise to colleagues or superiors.
It is foreseen that this “demonstration” will continue as AI deepens and spreads across the work landscape.
I've been working for half an hour for 10 minutes to get the weekly AI.
Lee Choo-chul, '92.
I run a product at an electrician company in Hangzhou. Last March, OKR asked us to do "AI-Deep Applications". The leader at the bi-monthly meeting asked what you solved last week with AI. I can't answer that. After the meeting, the supervisor talks.
March to May is the true roll. The group has a useful line-up and the administration exports data to the group on a weekly basis. That's why I didn't work with AI. I asked the first girl how she did it, and she said to let AI write her whole year's plans before she went to bed and didn't watch. I followed her, and I told AI to run the competition monitor as soon as I got to the job, and then I brushed short video, and I looked at entertainment gossip, which was kind of cleared.
The June wind has changed in its direction, not in terms of quantity, but in terms of output. The weekly report needs to reflect "AI cases" and data. I'm starting to figure out what to do. In one of the most funny cases, I wrote "Let AI translate the boss's half-hour meeting into a text" into "Burning up voice messages to structure the process" which was later rated a case of excellence. On the other hand, at the beginning of this month, I drew up a channel call for ten minutes, and to get the weekly report to have an AI content, I asked AI to draft, and it made up a lot of data for me, and I turned it over to the 11th of the year and changed it on that basis. But I wrote "AI-assisted writing" in the weekly newspaper.
I've got work to do, and I've come up with a way. Two monitors, left running AI, right working. Program-type work, throw Brief in the frame, don't repeat instructions, ask what I answer. When it ran the first, I wrote the second, and when the second ran, I looked back at the results of the first, and I finished the third, and it was a few hours later. It's true that AI works. Last week, I had over 40 complaints of client service, which I classified as a problem, and 10 minutes had a form. And we've got the minutes. We just got to work. I've got the minutes.
But AI. couldn't get anywhere, 30 seconds and an hour. I tried to take advantage of the job, turn it halfway through, cut it back, cut it back, and often forget what it was. Then you brush your cell phone and you look at the progress. It was clear to me that the focus was falling, the screen cut and cut, and the day was tired, and I couldn't tell where. I had to classify the job, do it with the A, do it with the A, do it with the A, and do it with a little work like this.
Recently, the Department began to report to each individual on “AI use”. I took a look at the summary files distributed by the leaders in the group, and I said it was a synthesis of your minds. It's all AI's summary.
The Demo I wrote was useless, but the leadership was particularly satisfied.
Li Moo-chul, 27 years old, Beijing.
I'm the product manager of a big Beijing Internet factory. I'm 27 years old, and I've worked for two years.
At the beginning of the year, the company started asking for full access to AI and Skill. The first of these was the research and development colleagues who later spread to their posts. In fact, I have used AI intermittently before, writing and sorting the case, and it's in the "seeing" phase.
About March and April of this year, our team leader came to me and said that the PM would soon be eliminated if he didn't write the code, asking us to learn to write with the Ai, to solve some simple needs as much as we can, and not to look for research and development.
I did write some little demo after that, and I did some little UI design. To tell you the truth, it's not really working, and it's going to have to be done again for our research and development colleagues, and it's going to be like I'm writing it up again, and I'm going to end up in the background. But the leadership is satisfied. It was clear to me that it was easier to cross reports than it was before, and it was the second time that something could land.
Since June, our group has changed considerably, with adjustments at the front end and my reporting lines changed. The evaluation of AI outputs and efficiency by the new Leader is much more detailed, as was done earlier, and it can no longer be done on its own.
So I changed the way "AI works." I've written the steps of demand feedback, market research, product framework, and so on, it's a simple page reading and sorting, and it's every link for AI to handle, and it's basically AI to deliver. There's really not much value in it, and the performance is more important.
Many of these jobs do not require AI at all, and I can't take much time to do them on my own; second, there is a lot of demand that AI has given to colleagues who are not actually moving forward, who often find a logical error, and we have to meet again. On the face of the process, the actual efficiency is not necessarily higher.
But it's good to do it, because Leader thinks we're busy and AZworkflow is on board, and it's good to have good data and good reporting.
My own judgment in this matter is that AI can really be effective, especially in writing codes and documents, and it's really time-saving, but it's not always effective, but it's complicated. And I personally think the biggest problem is that when Leader thinks that you're more efficient, he'll want you to do other people's jobs, and the dividends don't come to me, and that's the basis for that.
I still think it's better if we do what we're good at. Otherwise, it's just spreading AI anxiety, wasting resources and forcing everyone to act.
Less than 70% of AI usage is down, and we've got a bunch of Skylls that nobody uses.
Nam, after 95, Shenzhen, planning the operation.
I'm planning a business in Shenzhen. Our boss is very keen on AI, hoping to embed it in every segment, asking you to use AI as much as possible to complement the work, and encouraging us to build our own AIworkflow.
The use of AI was then even included in the performance appraisal: less than 70 per cent was used, with a 20 per cent reduction in performance.
At first, it felt like nothing, after all, AI could do something for us. A programme of campaign planning, previously less than three days. Now with AI, the first draft can be completed in one day. There's a lot of petty work like competitive analysis and PPT production, and AI can help. It really saves time.
But slowly, things started to get weird. Within the company, there was an AI competition that covered more work scenes, built more workflow and produced more Skill than anyone who could use it. If it can be proved that you have reached out to a particular work session using AI, it can be counted as a result.
At the outset, what you did was useful, such as Skill, which was associated with the planning of activities, and could reduce some repetitive and trivial work.
But, after all, the scenes of value are limited, and in order to complete the task, you have begun to roll in numbers. Some were looking for less important work scenarios, making Skill, and packaging it into a name that sounded great; others were using AI to do a small program to show what they had solved and how much time they had saved.
The boss saw so many new tools and felt very motivated to use AI. In practice, however, there are things that are of little help to the job and that have been made to no avail.
I feel like you've come to a delicate agreement. It's more important than whether AI is helping or not to produce tangible results. After all, it is easy to “mask” how much of the Skill, how much of the work is covered, but how much time has been saved and how many practical problems have been solved, which is not easy to measure.
In addition to these AI results, which are specifically designed to showcase, a phenomenon is beginning to emerge in daily work: AI is creating new workloads.
A colleague quickly produced a program of activities with AI, with a fine PPT, to report to the boss. The boss was satisfied and felt particularly efficient and immediately allowed the people below to follow the programme. As a result, it was only at the implementation stage that a number of gaps in the programme were identified and many practical issues were not taken into account. Finally, it has to be supplemented and adjusted by a little bit.
What is even more overwhelming is that the existence of AI has also increased the number of programme changes. Once a previous plan had been completed, the leadership had looked at it and made some comments, which could be taken forward. Now the leader will see for himself, and I'll have it reviewed by AI. AI is often able to propose a whole range of changes, some of which are of real value, but some of which do not match reality and we have to examine, interpret and modify each article.
I think it's quite contradictory. The boss wanted AI to save time, but we had to spend extra time in the competition to prove we were actually using AI. It has not been easy to use AI to improve efficiency and to undertake additional audits, modifications and assignments. Thus, the use of AI becomes an uneconomical one.
We're doing the AIC.T.C.T.T.
Wang Min, 95 years later, Shanxi Software Development Company sold it.
I've been in this company for five months, and I've come to figure out what it means to be "Ai."
The main business of the company is software development, which has been in place for more than a decade and has approximately 100 employees. The owner is in his 60s, and the knowledge of AI comes largely from marketing numbers. Last year, he proposed an AI transformation, which was upgraded directly into an annual strategy early this year. For most of the year, some of the operations doubled their post-AI output, but more often than not they were out of proportion.
First, staff performance appraisals have added AI participation and performance ratings are based on AI usage.
And so, you're all like, "Ai." For example, the day before yesterday a skill was done, yesterday a data cleansing exercise was done, and today clients are planned for a full life cycle from 0 to 1, and AI's share is rolled from 30 to 80 per cent. It doesn't matter if AI does or doesn't do anything. Just "participated."
I used to write weekly newspapers for the boss, and now it's a loop, and I send ideas to AI, I write weekly newspapers, I check them and I send them to the boss, I send them to AI, I send them to AI, and I send them to AI. What I do most with AI every week is for AI to make up for me.
That was not enough, and the boss had introduced an AI monitoring system.
He felt that the daily job of a staff member was to touch fish by typing to a computer, often saying that typing failed to perform, and to make more phone calls and communicate. As a result, the AI client management system was put on duty, an automated recording was made, AI listened, summarized, advised, and the production of daily newspapers was closed.
Now the employees are calling at the company like they're looking at the mirror. It's a bad note from the back office, it's not good for AI to advise, and the customers don't talk about auto-red. When the boss is all right, he replays the APP communication records and hears what's not right and right. Every time we see a crowd of news pops out, we get cold back.
And then we learned how to show it to AI, for example, to talk to the pronunciation, to smile, to say thanks before we hang up. The better it is, the higher the AI scores.
I don't know how much it worked for the company, but one time, AI's assistant exposed the company's payroll.
One day, the boss had a meeting to build an in-house AI robot, with the staff at the centre of it, to adapt it to address information sharing and improve efficiency. Following the online curriculum, the personnel colleagues carried the attendance, performance and salary scales to the site and finished the day. It was then unknown who had asked the company about the wages and AI had dumped the details.
It blew up the pot. Some asked about the same pay, others found out that performance was down, others pulled the entire company's pay and found out the names of three pay-as-you-go, but never went to work. To everyone's surprise, instead of criticizing people in public at the wrap-up meeting the next day, the boss boasted that she was implementing efficiently. After the meeting, however, the owner contacted AI for technical restrictions of permission and the system was back online.
Although AI was a good thing, the more it went wrong, the less we expected it to work, the more it went. In the end, it was not AI that was exposed, the payroll that was leaked, the company's lack of data permission, the basic common sense. The employee shows up at work because the boss used AI as a new surveillance tool. AI amplifies the management itself, the process is smooth, it magnifies efficiency and guesses, and it magnifies internal consumption.
To figure out the floor, I'm relieved. Even without a test, I can't get out of AI, do a PPT, write a planning book, and sort a file, which really saves me a lot of effort. But I will not give it anything that requires a professional judgement.
The leader does not want efficiency gains. He looks like "all-in-one."
The canoe, after 0000, Guangzhou, software development project. Division
A few days ago, an unknown colleague came to me and asked me how publish's plug-in in the internal plug-in market worked. I'm happy. After all, this is something that I took my time to study, and it was used on my own initiative to show that it did solve real problems.
I've been assigned a lot of AI since I got on duty. Research tools, the development of plugins and the construction of workflow are being developed to automate duplicate links in daily operations. I'm actually willing to do these things, and I really have a sense of achievement.
But my +1 (direct superior) seems less satisfied.
His objective was to get the whole department to use AI, preferably to achieve end-to-end standardized processes. From personal use to full promotion, from one link to the establishment of a complete SOP, AI has gradually become a sectoral task to be accomplished.
I'm a little confused. The tools that we have found do work well, but why do we have to use the same process when everyone has different work content and habits? It is even more surprising that the older employees in the group who have worked for more than five years do not seem to be in a hurry. An old employee told me privately that he didn't really care.
Then I figured it out. The core competitiveness of older employees lies in business experience and expertise, and AI is only a supporting tool for them, how it can be easily used and not linked to performance. What really connects to AI is the +2 that calls for a transformation of the sector, the +1 that responds positively, and the new employees who are unfortunately assigned to specific tasks, like me.
This explains why +1 is so committed to creating SOOP.
Several new outsourcing colleagues have recently arrived in the Department. I'm afraid they'll probably leave when this version is developed. When the project starts, new people will come in. With such high mobility, the optimal way to manage it would naturally be to standardize the workflow route so that new recruits could be trained without much training. People can walk, skills left.
I figured that out, and I realized that I and the leadership might have a completely different understanding of AI.
I think that the direction of AI is a sexual and customized one. Used to be a human adaptation tool, it should now be an adaptation tool, and everyone can build workflow according to their needs. However, +1 seems to believe that AI can turn all business into SOP. It's hard to say. I think half his thoughts are wishes, half performances. A real increase in efficiency does not necessarily matter, but it is important to show that the entire sector is undergoing an AI transformation.
Of course, from a management point of view, this logic is perfectly reasonable. Reducing the cost of training through AI and reducing reliance on individual experience are normal demands of the organization.
It was only when I thought I was helping colleagues to reduce duplication of work that I realized that I might also be helping companies to reduce their dependency on employees. Even more subtle, it does not conflict with my personal interests, even in my favour. I can learn new technologies, build project experience and gain a sense of practical achievement.
I still like to do the AI work, but once I see what management might do, I have a mustard in mind. Now I understand that some say that moral sense is too high for the division. The hardest part is not being forced to do a bad thing, but to do what you like, but to find that it serves a logic that you do not agree with.
* At the request of the interviewer, the names Li Zhou, Li Moe, Nam, Wang Min, and Linzhou are aliased.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.