老板们，脸疼吗？
亚特兰大联储的一项相关研究，做了一件很不厚道的事——它把过去几年企业高管们吹过的AI牛皮，挨个戳破了。
核心发现就一句话：约九成企业高管认为，AI还没有提高自己公司的生产力。
更扎心的是下面这句：2021年以来的生产力提升，主要来源于疫情期间远程办公带来的红利——跟AI没什么关系。
翻译翻译：你们以为的AI提效，其实是在家上班省出来的通勤时间。
裁员裁出了个寂寞
回想过去这两年，多少公司的操作是这样的：
第一步，高管在全员大会上激情演讲："AI时代来了，不拥抱AI的公司都会死！我们公司要全面AI化！"
第二步，宣布裁员："为了拥抱AI转型，我们要优化人员结构，减少冗余岗位。"翻译：老员工太贵，先砍一批。
第三步，斥巨资采购各种AI工具：订阅费、API调用费、企业版license，钱花了不少。
第四步，等效果。然后……就没有然后了。
生产力没上去，员工士气先崩了。剩下的人工作量翻倍，因为活儿还是那么多、人少了一半。AI工具倒是买了一堆，但真正能落地提效的没几个——大部分时候，员工还得花时间学习怎么用AI、怎么改AI生成的垃圾结果，工作量反而增加了。
最搞笑的是什么？裁掉的那些"老员工"，本来是公司里最懂业务、最知道怎么干活的人。你把懂行的人裁了，换成一堆AI工具，然后发现AI连公司最基本的业务流程都搞不清楚。
于是出现了很魔幻的一幕：很多公司裁员之后，又开始把老员工请回来——以顾问的名义，时薪比原来的工资还高。
合着折腾一圈，钱花了，人情也丢了，最后还得花更多的钱把人请回来。图啥呢？
AI不是万能钥匙，是放大器
关于AI提效这件事，很多老板理解错了一件事：AI不是用来替代人的，是用来放大优秀的人的。
一个本来就很能干的人，用了AI之后，效率可以翻两三倍。因为他知道该做什么、怎么做、怎么判断结果对不对——AI只是帮他省掉了重复劳动的部分。
但一个本来就不知道自己在干嘛的人，用了AI之后，只会更快地产出更多垃圾。因为他连判断AI输出对不对的能力都没有。
AI的本质是杠杆。支点对了，能撬动十倍效率。支点错了，能把自己的公司撬翻。
可惜的是，很多高管对AI的理解还停留在"AI能替代多少人"这个层面。他们把AI当成了裁员的借口，而不是成长的工具。
结果就是，人裁了，AI也上了，生产力没涨，公司还乱了。
实用提醒：别被AI焦虑绑架
说了这么多，给打工人几条实用提醒：
第一，别慌。天天看新闻说AI要替代这个替代那个，好像明天就要失业了。实际情况是：九成公司的AI都还没真正提效。AI确实会改变很多东西，但那个"奇点"比媒体渲染的要远得多。
第二，做那个会用AI的人，而不是被AI替代的人。AI会不会替代你，不取决于AI有多强，而取决于你会不会用AI。同样是写代码，会用AI辅助的程序员，效率是不会用的好几倍。学会用AI放大自己的能力，比担心AI抢饭碗有用得多。
第三，保持你的"不可替代性"。什么是不可替代性？不是某项具体技能——技能AI学得比你快。是你的业务理解、你的判断力、你的沟通能力、你对公司上下文的掌握、你解决模糊问题的能力。这些东西，AI短期之内学不会。
最后想说一句：那些天天喊着"AI要替代人"然后疯狂裁员的公司，最后大概率会发现——
替代得了员工，替代不了干活。省下了工资，省不下麻烦。
今天就槽到这里，明天继续。
Hey bosses - that sting on your face? That's reality.
A research study associated with the Atlanta Fed did something pretty unkind - it went around popping every AI bubble corporate executives have been blowing these past few years.
The core finding, in one sentence: roughly 90% of corporate executives believe AI has not yet improved their company's productivity.
And the even more brutal follow-up: productivity gains since 2021 mainly came from the dividend of pandemic-era remote work - they have basically nothing to do with AI.
Translation: all that "AI productivity" you thought you had? It was just the commute time people saved by working from home.
All Those Layoffs - For Nothing
Think back over the past two years. How many companies operated like this:
Step 1: CEO gives a fiery all-hands speech - "The AI era is here! Companies that don't embrace AI will die! We are going all-in on AI!"
Step 2: Announce layoffs - "To embrace the AI transformation, we're optimizing our workforce structure and reducing redundant roles." Translation: senior staff are expensive, let's cut a bunch first.
Step 3: Spend huge sums on AI tools - subscriptions, API fees, enterprise licenses. Real money goes out the door.
Step 4: Wait for results. And then... crickets.
Productivity didn't go up. Employee morale collapsed first. The people who are left have doubled workloads because there's still the same amount of work and half as many people. AI tools got purchased, sure - but few actually deliver efficiency gains. Most of the time, employees spend extra time learning how to use AI, then fixing all the garbage AI outputs - their workload actually increases.
The funniest part? Those "senior employees" who got laid off were the ones who actually understood the business and knew how to get things done. You cut the people who know what they're doing, replace them with AI tools, and then discover AI can't even figure out the company's most basic business processes.
Which leads to the most magical scene of all: many companies, after laying people off, start hiring the senior staff back - as consultants, at hourly rates higher than their old salaries.
So after all that折腾: money spent, relationships burned, and now you're paying more to get the same people back. What was the point, exactly?
AI Isn't a Magic Wand - It's an Amplifier
On the topic of AI productivity, many bosses have one thing fundamentally wrong: AI isn't for replacing people. It's for amplifying great people.
Someone who's already great at their job, equipped with AI, can become 2-3x more efficient. Because they know what to do, how to do it, how to judge whether the result is correct - AI just saves them from the repetitive parts.
But someone who doesn't know what they're doing, equipped with AI, will just produce more garbage, faster. Because they can't even tell whether the AI's output is right or wrong.
AI is essentially a lever. With the right fulcrum, it can multiply efficiency tenfold. With the wrong fulcrum, it can tip your whole company over.
Unfortunately, many executives' understanding of AI is still stuck at "how many people can AI replace." They treat AI as an excuse for layoffs, not a tool for growth.
Result: people get cut, AI gets deployed, productivity doesn't rise, and the company descends into chaos.
Practical Tips: Don't Let AI Anxiety Manipulate You
Okay, enough venting. Some practical tips for workers out there:
First, don't panic. Every day the news says AI will replace this, replace that, like we're all going to be unemployed tomorrow. Reality: 90% of companies haven't actually seen AI improve productivity yet. AI will change many things, but that "singularity" is much farther away than the media makes it sound.
Second, be the person who uses AI, not the person AI replaces. Whether AI replaces you doesn't depend on how strong AI is - it depends on whether you can use AI. Same coding job: a programmer who uses AI as a copilot is several times more efficient than one who doesn't. Learning to use AI to amplify your own abilities is way more useful than worrying about AI stealing your job.
Third, protect your "irreplaceability." What's irreplaceable? It's not some specific skill - AI learns skills faster than you. It's your business understanding, your judgment, your communication skills, your grasp of company context, your ability to solve ambiguous problems. These are things AI won't learn anytime soon.
One final thought: all those companies shouting "AI will replace humans" while frantically cutting staff will probably discover, in the end -
You can replace employees, but you can't replace the work. You save on salaries, but you can't save on headaches.
That's it for today's rant. Back at it tomorrow.
The Cao column draws from industry observations and public research, exploring absurd AI phenomena through a lighthearted lens. Disagree? Come argue about it.
