Alexandr Wang, Meta’s chief AI officer, has a blunt piece of advice for teenagers: spend your time vibe coding. In remarks in September 2025, reported by Fortune and CNBC, the Scale AI co-founder said that a 13-year-old who devotes the next few years to building software with AI tools will be in the position Bill Gates and Mark Zuckerberg occupied when personal computers and the web were young.
Wang drew a direct comparison with the computing boom of the late 1970s and 1980s, when a generation of young programmers learned to code on early machines and went on to found the companies that defined the industry. He argued that AI has created a similar opening, with one difference: the programming language is now ordinary speech.
What vibe coding means
Vibe coding describes building software by telling an AI assistant what you want in natural language and letting it write the code. A user might ask for a website with a blue button, a logo at the top and a contact form at the bottom, and a tool such as Cursor, Replit or ChatGPT produces a working first version. The developer then steers the result through conversation, testing what comes back and asking for changes, rather than writing every line by hand.
The term was coined in early 2025 by Andrej Karpathy, a co-founder of OpenAI and former head of AI at Tesla, who used it to describe a loose, conversational way of working with AI coding assistants. It spread quickly. Collins Dictionary named vibe coding its word of the year for 2025, and Google chief executive Sundar Pichai has said publicly that he has used tools such as Cursor and Replit himself.
Practice over curriculum
Wang’s view is that fluency with these tools comes from hours of use rather than from a course. He has said that someone who spends ten thousand hours experimenting with AI coding tools and learns to get more out of them than other people will have a significant advantage. The point is less about memorizing syntax and more about developing judgment: knowing what to ask for, how to evaluate what the model produces and when it is wrong.
Wang became the world’s youngest self-made billionaire through Scale AI, the data labeling company he co-founded as a teenager. In June 2025 Meta invested about 14.3 billion dollars in Scale and hired Wang to lead its new Meta Superintelligence Labs. In October the company cut roughly 600 positions across its AI units, a move Wang framed internally as a way to reduce layers and speed up decisions, even as Meta continues to spend tens of billions of dollars a year on AI infrastructure.
A starting plan
For a student or anyone else curious about the approach, the barrier to entry is low. A practical routine looks like this:
- Set aside regular time, even 30 minutes a day, in a tool such as Cursor, Replit or ChatGPT. Several offer free tiers.
- Start with a small, concrete project: a calculator, a personal home page or a simple game. If you are stuck for ideas, ask the assistant for some.
- Describe what you want in plain language, then read what the AI produces and try to follow its reasoning instead of pasting blindly.
- Iterate. Refine your prompts, fix what breaks and ask the assistant to explain any code you do not understand.
- Develop your own working style. Experienced users build up a repertoire of instructions and checks that suit the kind of software they want to make.
Critics of vibe coding point out that code produced this way can hide bugs and security flaws that a non-programmer will not spot, and that professional software still needs people who understand what the machine wrote. Wang’s argument is not that those skills disappear, but that the people most likely to acquire them are the ones who start experimenting with the tools now.
