AI Is Making Coding Cheap. Here's What Matters Now
Perfology Clips
526,837 views
AI is changing what it means to be an engineer.
Andrew Ng and Laurence Moroney offer two complementary perspectives on the future of AI — from how AI is changing software development and product discovery to the skills developers need to build valuable AI systems.
Andrew Ng argues that AI is making software construction significantly faster and cheaper. As the cost of building software falls, the bottleneck increasingly shifts toward something harder:
What should we build, and why?
That changes the role of the engineer.
The fastest-moving engineers may increasingly be those who can combine technical ability with product thinking, user empathy, experimentation, and a deep understanding of the problems customers actually need solved.
Laurence Moroney approaches the same shift from the career and engineering side. His advice is simple:
Don't become an AI shill. Become a trusted advisor.
When someone says:
“Let's build an AI agent.”
The first response shouldn't be choosing an agent framework.
It should be:
Why? What problem are we solving? What outcome do we want?
Moroney proposes a practical four-step workflow for building useful AI agents:
1. Understand intent
2. Plan
3. Execute using tools
4. Reflect and verify
This process helps separate real AI engineering from AI hype.
The discussion also explores a major split emerging in AI development:
Big AI vs. Small/Open-Weights AI.
Big AI focuses on increasingly powerful models and the race toward AGI.
Small AI focuses on efficient, self-hostable and open-weight models that developers can fine-tune and run closer to the user.
That second category could create major opportunities for engineers who understand model fine-tuning, deployment, optimization and local inference.
Laurence Moroney has recently highlighted this Small AI shift, including increasingly capable open-weight models that can run on phones, laptops and single-GPU systems.
Andrew Ng has similarly emphasized moving from AI experimentation toward AI products and workflow redesign rather than simply accumulating disconnected AI experiments.
The career message from both speakers is therefore surprisingly practical:
Don't just learn AI. Learn how to create value with AI.
That means developing skills across:
• AI and machine learning
• Software engineering
• Product thinking
• User empathy
• AI agents
• Model fine-tuning
• Open-weight models
• Small AI and local inference
• Problem solving
• Communication
• Measuring real business outcomes
They also emphasize the importance of hard work — but not simply counting hours. The goal is meaningful output and measurable progress.
And when evaluating a new AI technology, Moroney recommends making it as mundane as possible.
Strip away the marketing.
Understand the mechanics.
Know what the model is actually doing.
That is how engineers become trusted technical advisors instead of simply repeating the latest AI buzzwords.
In this video, we break down Andrew Ng and Laurence Moroney's perspectives on AI engineering, AI jobs, AI agents, product discovery, AI hype, Small AI, open-weight models, career strategy and the future of software development.
The biggest question is no longer:
“Can AI write the code?”
It's:
“Do we know what should be built?”
Copyright and Attribution Notice:
All original lecture content, slides, and audio rights belong exclusively to Stanford University and Andrew ng and Laurence Moroney. This summary and commentary are produced strictly for educational purposes.
🔔 Subscribe to our channel for more tech tips and tutorials:
👍 Like us on Facebook : Facebook: perfology
👍 Add us on Instagram: Instagram: perfologys
👍 Follow us on Linkedin: LinkedIn: perfology
#AndrewNg #LaurenceMoroney #AI #AIEngineering #AIAgents #AIJobs #SmallAI #ArtificialIntelligence #MachineLearning #SoftwareEngineering #Perfology
20 Comments
• AI and machine learning
• Software engineering
• Product thinking
• User empathy
• AI agents
• Model fine-tuning
• Open-weight models
• Small AI and local inference
• Problem solving
• Communication
• Measuring real business outcomes
it's already very good if you can be good in 2 to 3 areas in this list. it's a rather tall order to be a generalist across that many areas.
1. The Rapid Acceleration of AI Capabilities
Exponential Task Complexity Growth: Rather than evaluating AI on fixed accuracy benchmarks, task complexity (measured by human time to complete a task) is doubling rapidly—roughly every 7 months for general tasks and every 70 days for AI coding [02:03].
Golden Age for Building: Developers now have access to powerful AI building blocks (LLMs, RAG, agent workflows, voice AI) [03:02], enabling small teams or individuals to create software that was impossible for even top-tier engineering teams just a year ago [03:32].
2. The Shift to the "Product Management Bottleneck"
Code is Getting Cheap: As AI dramatically lowers the time and cost to convert a clear specification into functional code [05:40], the biggest bottleneck is no longer writing code [06:31].
Deciding What to Build: The primary constraint is now Product Management (PM)—knowing what to build, understanding user feedback, and refining clear product specifications [05:40].
Engineers as Product Thinkers: Engineers are encouraged to adopt more product design and user feedback loops to maintain fast execution velocity without waiting on dedicated product teams [09:00].
3. Career & Practical Advice
Stay on the Frontier of Tools: AI coding tools (e.g., Claude Code, OpenAI Codex, Gemini) evolve rapidly [04:05]. Remaining even half a generation behind can significantly reduce personal productivity [05:01].
Build and Iterate: The single best way to advance an AI career is to continuously build projects, ship them, and show them to others [05:13].
Team & Manager Selection: When choosing where to work, prioritize the specific project team and daily manager over general company brand prestige [12:40].
Work Ethic & Responsible AI: Build responsibly to avoid software that harms others [15:32], and put in the hard work required to capture the immense opportunities available right now [16:44].
4. Broader Impact of AI Access (Q&A Highlights)
Democratizing Hardware: Cloud tools (like Google Colab) have democratized AI development by giving researchers around the world free or low-cost access to compute GPUs [01:42:22].
Social Impact & Intent: AI itself is neutral; whether it promotes equality or inequality depends on human intent, education, and implementation [01:43:36]. Developers should "assume good intent, but prepare for bad intent" [01:44:05].
Amazing lecture! Thanks for the tips, clarifications, and insights into career paths in AI.
👏👏👏
AI Is Making Coding Cheap. Here's What Matters Now
Perfology Clips
526,837 views
AI is changing what it means to be an engineer.
Andrew Ng and Laurence Moroney offer two complementary perspectives on the future of AI — from how AI is changing software development and product discovery to the skills developers need to build valuable AI systems.
Andrew Ng argues that AI is making software construction significantly faster and cheaper. As the cost of building software falls, the bottleneck increasingly shifts toward something harder:
What should we build, and why?
That changes the role of the engineer.
The fastest-moving engineers may increasingly be those who can combine technical ability with product thinking, user empathy, experimentation, and a deep understanding of the problems customers actually need solved.
Laurence Moroney approaches the same shift from the career and engineering side. His advice is simple:
Don't become an AI shill. Become a trusted advisor.
When someone says:
“Let's build an AI agent.”
The first response shouldn't be choosing an agent framework.
It should be:
Why? What problem are we solving? What outcome do we want?
Moroney proposes a practical four-step workflow for building useful AI agents:
1. Understand intent
2. Plan
3. Execute using tools
4. Reflect and verify
This process helps separate real AI engineering from AI hype.
The discussion also explores a major split emerging in AI development:
Big AI vs. Small/Open-Weights AI.
Big AI focuses on increasingly powerful models and the race toward AGI.
Small AI focuses on efficient, self-hostable and open-weight models that developers can fine-tune and run closer to the user.
That second category could create major opportunities for engineers who understand model fine-tuning, deployment, optimization and local inference.
Laurence Moroney has recently highlighted this Small AI shift, including increasingly capable open-weight models that can run on phones, laptops and single-GPU systems.
Andrew Ng has similarly emphasized moving from AI experimentation toward AI products and workflow redesign rather than simply accumulating disconnected AI experiments.
The career message from both speakers is therefore surprisingly practical:
Don't just learn AI. Learn how to create value with AI.
That means developing skills across:
• AI and machine learning
• Software engineering
• Product thinking
• User empathy
• AI agents
• Model fine-tuning
• Open-weight models
• Small AI and local inference
• Problem solving
• Communication
• Measuring real business outcomes
They also emphasize the importance of hard work — but not simply counting hours. The goal is meaningful output and measurable progress.
And when evaluating a new AI technology, Moroney recommends making it as mundane as possible.
Strip away the marketing.
Understand the mechanics.
Know what the model is actually doing.
That is how engineers become trusted technical advisors instead of simply repeating the latest AI buzzwords.
In this video, we break down Andrew Ng and Laurence Moroney's perspectives on AI engineering, AI jobs, AI agents, product discovery, AI hype, Small AI, open-weight models, career strategy and the future of software development.
The biggest question is no longer:
“Can AI write the code?”
It's:
“Do we know what should be built?”
Copyright and Attribution Notice:
All original lecture content, slides, and audio rights belong exclusively to Stanford University and Andrew ng and Laurence Moroney. This summary and commentary are produced strictly for educational purposes.
🔔 Subscribe to our channel for more tech tips and tutorials:
👍 Like us on Facebook : Facebook: perfology
👍 Add us on Instagram: Instagram: perfologys
👍 Follow us on Linkedin: LinkedIn: perfology
#AndrewNg #LaurenceMoroney #AI #AIEngineering #AIAgents #AIJobs #SmallAI #ArtificialIntelligence #MachineLearning #SoftwareEngineering #Perfology