(If today’s AI is like a well-behaved intern helping you organize files, then the 2026 workplace might no longer need people who can’t even utilize an intern.)
Today is May 15th. I glanced at the old almanac on my desk, and it says it’s the “International Day of Families.”
But if you look at the newly released “Maimai 2026 Spring Recruitment Workplace Insight Report,” you’d probably feel that many people are experiencing a kind of “running away from home” in the workplace—that familiar, stable work environment seems to have been reset overnight. The data clearly shows a 22.6% increase in job postings, and everyone is shouting about a market recovery.
So why is it that when you ask people around you, they still can’t find a job?
Ice and Fire: The True Cross-Section of K-Shaped Divergence
Guess what.
On one side, scientists working on AI pre-training are easily breaking the 1.6 million RMB annual salary mark, and for high-performance computing engineers, there are 4 open positions fighting for 1 candidate. On the other side? Nearly 70% of people have sent out dozens of resumes, only to end up waiting right where they started.
This is just absurd. As a corporate worker who has been cultivating myself in code and documentation, I’ve been pondering this too. No exaggeration, the tearing sensation of this year’s spring recruitment is more intense than ever before. This is no longer a simple warm spring breeze; these are two completely non-intersecting parallel universes.
There’s a highly counter-intuitive statistic in the report: people who submitted their resumes 1 to 10 times actually had the highest onboarding rate, reaching 36.6%.
And those who mass applied over 50 times? It dropped to 26.4%.
Here comes the question. Why is it that the more you apply, the less likely you are to get an offer?
My superficial understanding is that companies today are no longer casting a wide net for hiring; instead, they are precisely targeting core skill sets. Basic positions requiring less than 1 year of experience have directly shrunk by 12%. Interestingly, even people who haven’t submitted a single resume have a high rate of receiving offers. How did they get them? Either through internal referrals from acquaintances or by being spotted early on by headhunters on professional platforms.
It’s not that companies aren’t hiring anymore.
Rather, they really don’t want to spend the resources to “nurture” newcomers who lack immediate combat readiness.
(Looking at the setup of this lab, embodied AI is no longer just drawing pie in the sky on blueprints; it’s fighting real battles for practical application.)
The Secretly Evolving “Shrimp Farmers”
Regarding the impact of AI, there’s a set of data in the report that sent a chill down my spine.
Nearly half (43.62%) of AI users purchase these tools out of their own pockets. That means they are using their own salaries to buy AI to do the company’s work.
Honestly, this is like bringing your own rations and shovel to farm for the landlord.
Big companies have deep pockets, providing AI tools for free to 70.3% of their employees. For small companies with under 500 people, this drops to a pitiful 42.5%. When the company doesn’t issue weapons, employees have to buy their own. Why is everyone trying so hard?
Because there is a term going around now called “farming shrimp.”
Put simply, it means using autonomous execution Agents like OpenClaw, throwing tedious documents and reports to the AI, and letting it run on its own. In data and product roles, one in four people is doing this.
This is quite fascinating.
The biggest impact brought by AI is not eliminating programmers, but forcing everyone to become one. Waiting for leadership to arrange unified training? It’ll be too late. 38.51% of our peers say their companies have already incorporated AI proficiency into performance reviews. Everyone is paying out of pocket to boost efficiency, evolving in secret.
From programmers to all other roles in the rat race, how much longer do you think it will take?
(The Agent on the desktop can now directly take over the computer to work; it feels somewhat like hiring an invisible assistant who doesn’t need social security or housing fund benefits.)
Don’t Rush to Change Careers; Even the Trend has its Tiers
Since AI is so hot, can’t I just impulsively switch careers?
It’s not that simple.
Among newly posted jobs, the share of AI roles jumped from less than 3% last year directly to 22% this year. The speed of this foundation shift is indeed fierce. But you need to realize that within the AI sector itself, it’s also a tale of ice and fire.
The ones truly eating meat and taking high salaries are the Class A technical talents doing foundational pre-training and inference optimization.
Today’s large models easily mobilize tens of billions of parameters just to run a simple text summary. Bluntly speaking, it’s like buying the entire supermarket just to cook one dish. This crude parameter redundancy must rely on these low-level tech giants for inference optimization.
In contrast, the salaries of the massive crowd pouring into Class B application development and Class C usage-based roles have barely budged, and have even slightly decreased.
By the way, the craziest thing in this spring recruitment is embodied AI.
The hiring index skyrocketed by 15 times. A fellow Shandong native working in embedded systems posted on Maimai that he hopped to a robotics company and his annual package directly doubled. During the interview, they didn’t even try to lowball him; they just asked how much he wanted to write down.
But the problem is that the threshold is absurdly high. SLAM algorithms, planning and control, embedded hardware development… It’s really not something you can bluff your way into just by reading a couple of Python introductory books.
This practically touches my knowledge blind spot. I am slowly catching up on these lessons too. Companies are frantically stockpiling talent, and the industry is racing from the lab to the mass production line. But this bowl of rice is destined only for those who have prepared their hardcore skill sets in advance.
(People who understand algorithms use AI to double their output, while those who don’t understand the principles will only be completely drowned by AI-generated nonsense code.)
If Time is No Longer a Moat
There is another very anomalous phenomenon in the report.
Roles with “No Experience Required” surged by 54%.
I sometimes wonder, have companies abandoned filtering people by years of experience truly because bosses have become more lenient? Perhaps I’m overthinking it. But today, your routine experience accumulated from writing CRUD (Create, Read, Update, Delete) every day for the past three years might truly pale in comparison to a rookie who is adept at playing with various Agents.
Think about it.
If a fresh graduate, relying on a few handy AI tools, can handle the modules you stayed up all night working overtime to write.
Then how much can your three years of “experience” really sell for in the market?
If years of experience no longer matter, will we see massive numbers of senior workers being silently replaced by cheaper, more “feral” workplace newcomers in the future?
That is truly heart-wrenching.
(The increase in “no experience required” jobs actually shows that companies now value your ability to harness these new tools, rather than how many years you’ve ground away in a specific position.)
The warm spring is indeed here. But the warming sunshine has certainly not fallen evenly on every worker.
Epilogue
Closes notes
Having talked about so much, I actually don’t have any particular conclusion. I just glanced out the window; the morning rain seems to have stopped, and it’s time to water the succulent on the windowsill.
References:
- Reference Data from Maimai 2026 Spring Recruitment Workplace Insight Report
- Chinese Almanac Dos, Don’ts and Festivals for May 15, 2026
- AI Desktop Agent Organizes Photos Better Than Expected
- The 5 Best AI Agents for Your Desktop in 2026
—— Lyra Celest @ Turbulence τ.
