Meta's new Chief AI Officer just told teenagers to abandon traditional coding. And Andrew Ng says that's "some of the worst career advice ever given."
Here's what's actually happening - and why both of them might be right.
What Alexandr Wang actually said:
Wang, now Meta's Chief AI Officer, went on the TBPN podcast and urged teenagers (around age 13) to spend as much time as possible on "vibe coding."
His definition: Using AI tools to generate code via natural language instructions rather than hand-writing every line.
His reasoning: In the next few years, AI could generate "all the code" he had ever written.
Learning to master prompt-driven code generation could become a major competitive edge.
Wang compared this moment to when early personal computing gave pioneers like Bill Gates a head start.
Andrew Ng's counterargument:
Ng argues this is actually the BEST time to learn to code - because AI amplifies the importance of knowing how to direct and guide these systems.
He's been vocal that discouraging people from learning programming due to AI is "some of the worst career advice ever given."
Ng also criticized the term "vibe coding" as misleading - coding with AI isn't casual or light work. It still requires deep thought and precision.
What both of them agree on (even if it's not obvious):
- Neither is saying traditional coding will vanish immediately.
- Wang is saying prompt-driven coding will take on more weight.
- Ng is warning not to abandon understanding how code actually works.
- The real debate: Where should your focus be?
Why Wang's advice matters:
Mastering AI-driven coding tools shifts the skill from hand-writing implementation to expressing intent, constraints, and higher-level structures.
The leverage moves up the stack.
Teens who get fluent in prompt-based code generation will likely outpace peers in: → Productivity → Rapid prototyping → Deploying AI-native solutions
Early adoption gaps may widen significantly.
Why Ng's pushback is critical:
Even if AI writes code, knowing how code works lets you:
This isn't "vibing" - it's a different form of engineering that still requires deep technical knowledge.
What's speculative in Wang's claim:
"AI might automate most coding jobs within five years" - This is controversial and unproven.
AI can generate a lot of code, but full automation of all software engineering is far from certain.
"Teens should dedicate ALL their time to one skill"
- This is extreme advice. People need balanced development, general thinking, breadth, and domain depth.
The skills that actually matter going forward:
Meta-skills are becoming differentiators:
→ Prompt design
→ System orchestration
→ AI tool literacy
→ Understanding when AI output is correct or flawed
→ Architecting solutions at a higher level
But these meta-skills still require understanding the fundamentals of what you're building.
.
.
Wang is right that AI-driven coding tools are changing the game. Early adopters will have an advantage.
Ng is right that abandoning coding fundamentals would be a mistake. You need to understand code to effectively work with AI that generates code.
The answer isn't "traditional coding" OR "vibe coding."
It's both.
Learn to code so you understand what's happening. Then learn to effectively use AI tools to amplify your output.
The practical advice for teenagers (or anyone learning to code):
Learn coding fundamentals. Understand how code works, what good code looks like, how to debug, how to architect systems.
Simultaneously, get fluent with AI coding tools. Learn to write effective prompts, orchestrate AI-generated code, and refine outputs.
Don't choose between traditional coding and AI-assisted coding. Master both.
Anik Singal
Meta's new Chief AI Officer just told teenagers to abandon traditional coding. And Andrew Ng says that's "some of the worst career advice ever given."
Here's what's actually happening - and why both of them might be right.
What Alexandr Wang actually said:
Wang, now Meta's Chief AI Officer, went on the TBPN podcast and urged teenagers (around age 13) to spend as much time as possible on "vibe coding."
His definition: Using AI tools to generate code via natural language instructions rather than hand-writing every line.
His reasoning: In the next few years, AI could generate "all the code" he had ever written.
Learning to master prompt-driven code generation could become a major competitive edge.
Wang compared this moment to when early personal computing gave pioneers like Bill Gates a head start.
Andrew Ng's counterargument:
Ng argues this is actually the BEST time to learn to code - because AI amplifies the importance of knowing how to direct and guide these systems.
He's been vocal that discouraging people from learning programming due to AI is "some of the worst career advice ever given."
Ng also criticized the term "vibe coding" as misleading - coding with AI isn't casual or light work. It still requires deep thought and precision.
What both of them agree on (even if it's not obvious):
- Neither is saying traditional coding will vanish immediately.
- Wang is saying prompt-driven coding will take on more weight.
- Ng is warning not to abandon understanding how code actually works.
- The real debate: Where should your focus be?
Why Wang's advice matters:
Mastering AI-driven coding tools shifts the skill from hand-writing implementation to expressing intent, constraints, and higher-level structures.
The leverage moves up the stack.
Teens who get fluent in prompt-based code generation will likely outpace peers in: → Productivity → Rapid prototyping → Deploying AI-native solutions
Early adoption gaps may widen significantly.
Why Ng's pushback is critical:
Even if AI writes code, knowing how code works lets you:
→ Critique the output
→ Correct errors
→ Debug problems
→ Refine solutions
Without coding literacy, you become a consumer of AI output instead of a creator.
You lose the ability to understand what the AI is actually doing.
What "vibe coding" actually involves (it's not casual):
The name is misleading. Here's what it really requires:
→ Designing effective prompts
→ Orchestrating multiple modules
→ Engineering pipelines
→ Debugging AI-generated code
→ Implementing safety constraints
→ Understanding system architecture
This isn't "vibing" - it's a different form of engineering that still requires deep technical knowledge.
What's speculative in Wang's claim:
"AI might automate most coding jobs within five years" - This is controversial and unproven.
AI can generate a lot of code, but full automation of all software engineering is far from certain.
"Teens should dedicate ALL their time to one skill"
- This is extreme advice. People need balanced development, general thinking, breadth, and domain depth.
The skills that actually matter going forward:
Meta-skills are becoming differentiators:
→ Prompt design
→ System orchestration
→ AI tool literacy
→ Understanding when AI output is correct or flawed
→ Architecting solutions at a higher level
But these meta-skills still require understanding the fundamentals of what you're building.
.
.
Wang is right that AI-driven coding tools are changing the game. Early adopters will have an advantage.
Ng is right that abandoning coding fundamentals would be a mistake. You need to understand code to effectively work with AI that generates code.
The answer isn't "traditional coding" OR "vibe coding."
It's both.
Learn to code so you understand what's happening. Then learn to effectively use AI tools to amplify your output.
The practical advice for teenagers (or anyone learning to code):
Learn coding fundamentals. Understand how code works, what good code looks like, how to debug, how to architect systems.
Simultaneously, get fluent with AI coding tools. Learn to write effective prompts, orchestrate AI-generated code, and refine outputs.
Don't choose between traditional coding and AI-assisted coding. Master both.
1 week ago | [YT] | 30