After watching Wang Xiaochuan’s interview
Q: If you were a young person with 100,000 RMB to start a business today, what would you do? A: I would start by making videos on a video platform and become a KOL (Key Opinion Leader).
Twenty years ago was very different from now. After the dot-com bubble burst around the turn of the millennium, the frenzy subsided. Cisco’s stock only returned to its early 2000s highs after two decades. But beneath the surface of that bubble, many of today’s internet giants were just beginning to take shape. Between 1995 and 2005, Amazon, Google, Alibaba, Tencent, and Baidu were founded one after another.
That moment in history is quite like the present. OpenAI lit the first spark for large language models (LLMs), and in just three years, the technology has grown at an exponential rate.
It naturally raises questions: Are LLMs just a flash in the pan? Can they truly solve real problems?
To the first question, the answer is both yes and no. Before solving real-world problems, LLMs are merely cool and impressive technologies. If they can’t provide actual productivity or virtual value, the tide of LLM hype will eventually recede.
As for the second question — yes, LLMs have the potential to solve real problems. For example, programmer salaries are a major cost for big tech companies. If LLMs can reach the level of human programmers, an AI that never complains and works tirelessly could replace dozens, if not hundreds, of developers. Although there’s still a gap, it’s closing rapidly.
And that’s just coding. AI has already replaced most of the writing, consulting, and web development tasks on platforms like Fiverr. A single prompt can now generate and deploy a full website to a server for anyone to access. The number of websites is bound to grow exponentially.
Technological evolution doesn’t go from caterpillar to butterfly overnight. It evolves step by step — from pagers to feature phones to smartphones to foldables. You can always trace the past in each iteration.
Just like how TRPO and PG evolved into PPO, and now we have GRPO, DAPO, GSPO…
Get on the train early.
Preparing for Fall Job Hunting and Internships
Each generation faces its own choices, challenges, and anxieties — but we keep exploring, learning, and improving.