Meta's Chief AI Officer Alexandr Wang: 13-Year-Olds Should Spend All Their Time on Ambient Programming as a Lifestyle
Meta's Chief AI Officer Alexandr Wang stated on the TBPN podcast that 13-year-olds should spend all their time on ambient programming and treat it as a lifestyle. He mentioned that investing 10,000 hours in AI coding tools would create a significant advantage.
Ambient programming refers to giving instructions in plain English for tools to generate code. Tools he mentioned include Replit and Cursor. In projects by Microsoft and Google, AI has written up to 30% of the code. Google CEO Sundar Pichai also mentioned in June that he created a webpage using this method.
Wang said that the ability of AI to write code has "radicalized" him. He believes the role of engineers is fundamentally different from before and predicts that in the next five years, models will be able to write a large amount of code, including all the code he has written.
Born around 1997, he dropped out of MIT as a freshman and founded the data labeling company Scale AI in 2016, serving as CEO for nearly a decade. Forbes listed him as the youngest self-made billionaire with a net worth of about $2 billion. At the time of the article's publication, he was 28 years old.
In June, he announced his departure from Scale to join Meta. Meta invested $14.3 billion for a 49% stake and talent, including him, increasing Scale's valuation from $14 billion to $29 billion. NVIDIA CEO Jensen Huang stated that natural language is the new programming language, and not having learned to code is no longer a barrier.
The buyers are teenagers using natural language to code, non-technical founders, and large engineering teams, while the sellers are outsourcing firms charging for manual coding and courses that only teach syntax. The event was driven by podcast discussions. The budget is shifting from computer degrees and line-by-line coding to prompt-based tools, benefiting Replit, Cursor, and those who can turn English into runnable products, while putting pressure on training that only sells syntax without teaching model scheduling capabilities.
Source: Public Information
ABAB AI Insight
In 2016, Alexandr Wang used Scale AI to provide data labeling needed for model training, serving clients who required manual conversion of raw data into training sets. He did not become a billionaire by completing a foundational model himself, but by supplying annotations for models. In June 2025, Meta acquired a 49% stake for $14.3 billion, with the valuation rising from $14 billion to $29 billion, contingent on his transition to Chief AI Officer. Control of the data company remains with the original shareholders, while he is bought into a large firm.
The capital path is equity for personnel. A 49% stake is not a full acquisition; Scale remains independent, and Meta gains annotation capabilities and Wang's schedule. In the podcast, he redefined engineers as those who schedule models, aligning with Scale's old business: humans no longer produce line by line but are responsible for giving instructions to models and accepting results. The 10,000 hours he mentioned as an advantage are aimed at 13-year-olds, not working engineers.
This can be compared to Jensen Huang calling natural language the new programming language, and Pichai personally using prompts to create a webpage. The current position is not during Scale's annotation expansion phase, but rather after Wang's transition to Meta, shifting coding rights from professional training to English instructions. The fact that models write up to 30% of the code in Microsoft and Google's projects indicates that replacement has begun, but is not yet complete.
Structurally, this represents technological replacement. Pricing power shifts from those who know syntax to those who can give instructions and accept results. If models can write all of his code within five years, the scarcity of computer degrees will decrease, while the fees for prompt-based tools will increase. If acceptance and system responsibilities still must be borne by humans, what is replaced is typing, not signing.
ABAB News · Cognitive Laws
Syntax can be outsourced to models, but acceptance remains with humans.
If 10,000 hours are spent on old skills, the advantage will expire first.
A 49% stake buys a company, but personnel must be written into the terms separately.