In the era of AI sprinting forward, we are just followers

Written by: haotian

I’ve seen many friends show off new AI tools, share Prompt techniques, and flaunt dazzling workflows, but but but, I wonder if everyone has the same feeling: we, the so-called “smart people” standing at the forefront of the trend, are actually just behind, desperately catching up. AI development is moving so fast that even immersing ourselves in it overall, we still feel unable to keep up with the rhythm:

1. If we define a “half-life” for learning AI skills, it’s probably measured in weeks.

Originally, I was still learning how to better control Cursor, then Claude Code came out and became an instant hit. I was proud of the various Prompt engineering techniques I had been developing, but once Skills appeared, I suddenly felt those techniques were useless… In the past, learning a technology could last at least three to five years, but now it might be outdated in three to five months.

This is the harsh reality right now: the so-called skills and tricks we spend a lot of time developing may be overtaken by a new AI iteration. But gradually, you’ll realize that AI development ultimately levels the playing field, bringing everyone to the same starting line. Who uses tools more innovatively? Who crafts more exquisite Prompts? These differences will be smoothed out.

What does the final competition come down to? “Curiosity and learning ability.” While others still think AI tools are irrelevant to them, your repeated exploration, experience, and trial-and-error put you ahead.

2. Using AI has shifted from secretive experimentation to proud display.

Moreover, I’ve noticed an interesting phenomenon: half a year ago, everyone used AI to write code secretly, afraid of being caught and exposed as “your code is entirely AI-generated.” Now? My programmer friends are actively showing off projects made with AI. “Look at this Dashboard, see this small app I built? I had Claude handle it in 10 minutes this morning,” and their tone is full of pride.

Actually, this mindset shift is crucial. In the past, our value in the workplace was based on “what skills I have.” Now, it’s shifting to “what I can accomplish with AI.” Just like after the Industrial Revolution, no one would mock you for using machines instead of handmade work; AI is just a productivity tool.

Those who oppose AI will eventually find that it’s not AI eliminating you, but those who use AI that will surpass you. Speed itself is a barrier.

3. For friends working on AI, explore subjective initiative beyond AI’s boundaries.

Of course, this doesn’t mean we should blindly rely on AI. Many times, AI oversteps, jumping out of your intended scope and making random operations, wasting valuable time. This requires us to approach AI with cognitive logic, rather than being completely led by AI.

Remember, no matter how powerful AI is, it’s just a tool. It cannot provide you with the understanding of “what to do” and “why to do it.” For example, you just want AI to optimize a simple data query, but it ends up overdoing it and restructuring your database architecture.

AI’s execution involves significant conditional triggers and rule definitions. These are areas we should expand our capabilities into—thinking about what AI cannot do, especially within its path-dependent scope, and then leveraging human subjectivity on top of that.

Ultimately, mastering AI isn’t about chasing the speed of AI tool iterations. It’s about truly understanding what the “limitations” of AI thinking and execution are. Then, using the innate wisdom that animals possess, to fill in the gaps.