Will AI Make Our Work Easier?
Since last year, every company of any size has been pushing hard to weave AI into daily work — programmers especially. Whether you’re on the front end or the back end, “using AI to boost efficiency” has become a hard metric. From measuring who uses AI the most, to who uses it best, to handing work over to AI entirely — it all came a little too fast. The rules change every week and every month. But as an employee, is your actual, effective output really several times greater than before? That remains to be proven. Has AI really made things easier? Hard to say!
AI boosts efficiency by 20%, yet we programmers work overtime harder than ever: is the ruler bosses use to measure productivity crooked? — after reading this piece on InfoQ, I felt it deeply.
“I don’t understand why the progress of AI isn’t letting us clock off at 5, but instead letting more people get laid off while the rest keep working until 10.”
In enterprise-level R&D, the time actually spent writing code is usually only 20%–30%. The bulk of the rest goes into communication, alignment, reviews, testing, and all sorts of ad hoc tasks. What AI mainly improves is that 20%–30% slice, so naturally it’s hard for it to move the whole picture. “Local improvements, I think, are easy to achieve,” said Ru Bingsheng. “The hard part is truly threading efficiency through the entire end-to-end process. That demands a lot from whoever is driving the change: you have to see clearly where the bottleneck is, dare to touch the process, and be capable of making trade-offs.”
In the future, as the token meter keeps ticking and context costs keep snowballing, will the human brain — which “seems slower” — instead become a high-end luxury? An engineer can spend 5 hours carefully studying a complex architecture and thinking a problem through deeply, without racking up the staggering compute bills the way AI does. In a corporate budget, this “slow human brain” might instead become the ultimate fixed-cost asset. Human value lies precisely in the places where tokens can’t be counted.
What the article describes is exactly what I’m living through right now — AI being woven into work, and AI-driven efficiency gains.
My own attitude toward AI has gone from curiosity, to astonishment, to habit, and finally to a bit of resignation. What happened along the way? AI tools have been iterating fast, and the sense of accomplishment that used to come from solving problems at work has gradually been eroded by how quickly AI solves them.
Think back to before GPT-3.5 came out. The path to solving a problem was nothing more than search engines, communities, forums, and tech blogs — opening dozens of tabs, searching high and low.
But after large language models and AI tools became widespread, a single chat box and a few agents solve problems that used to take a long time. You really can’t help but marvel at how powerful AI is!
As powerful as it is, though, has it really made our work easier? From my own experience, if it’s writing code, then yes, it really is easier — you no longer have to program the old-fashioned way, hand-typing every line. In short, the speed from development to implementation really has gone up. Exactly how much faster depends on what product you’re building. For simple, off-the-shelf CRUD-type things, the speed isn’t even in the same league. But for things with heavy business coupling, or things that are underrepresented in AI’s training data, the efficiency gains may not be as obvious — first you have to give the AI clear context so it understands your business, and only then can you build.
All in all, since first touching large language models around late 2023 / early 2024, and then the emergence of all kinds of AI coding tools in 2025, my work and life have gained a lot of convenience. The number of times I use Baidu and Google has dropped noticeably — I can basically solve things just from the AI’s answers. But in terms of workload, AI’s arrival has actually increased it, because no matter the company or the boss, none of them will let their employees take a paycheck out of their pocket so easily. They’ll squeeze out as much of an employee’s value as they can.
So, has AI’s arrival actually made our work easier? As it stands, no — if anything, the workload has grown larger and larger, because when efficiency per unit of time goes up, maintaining the original ratio of output requires raising the total volume of work.
And in the future, will AI make our lives easier? Optimistically, that’s surely inevitable, because a lot of what people do can be handed off to AI, so naturally there’s less for humans to do. Pessimistically, it may not be easier, because AI’s arrival will shrink the number of jobs in certain industries, which means the corresponding competition will grow. AI will force everyone to change — as for what kind of change, everyone is probably still figuring it out, myself included.
But one thing is certain: you need to prepare in advance. If your job is displaced because AI steps in, what else can you do? Worth thinking hard about!