Skip to main content

Chris Lattner & Jeremy Howard: Building Lasting Software in the AI Era

[HPP] Chris LattnerOctober 30, 202558 min
34 connections·40 entities in this video

The Craft of Software Engineering

  • 💡 Chris Lattner (LLVM, Swift, Mojo) and Jeremy Howard (Fast AI, Solvent) emphasize building systems from fundamental understanding.
  • 🚀 They highlight the importance of well-designed architecture and a strong engineering culture for creating software that lasts, like LLVM's 25-year longevity.
  • 🛠️ Software craftsmanship is crucial for building scalable and maintainable products, contrasting with the current trend of rapid, AI-driven code generation.

AI Hype vs. Practical Application

  • ⚠️ The discussion draws parallels between the 2017 self-driving car hype and today's AGI predictions, noting that progress often follows S-curves rather than immediate exponential growth.
  • 🧠 Jeremy Howard explains that Large Language Models (LLMs) are powerful pattern predictors, not designed for AGI, and that fine-tuning is key to their utility.
  • 🚫 Chris Lattner cautions against making decisions based on fear of imminent AGI, advocating for a pragmatic approach to technology adoption.

AI Coding: Benefits and Pitfalls

  • ✅ AI coding offers 10-20% productivity gains for production code and 5-10x for prototypes, especially for discovery and learning new APIs.
  • 📉 Jeremy's team experienced a drop in productivity and morale when relying heavily on agentic AI workflows, finding AI-generated code often wrong or superficial.
  • ❌ A major pitfall is AI-generated unit tests becoming technical debt by testing details instead of core ideas, or encouraging developers to delegate understanding rather than learn.

Cultivating Mastery and Durable Value

  • 🎯 The speakers advocate for using AI as a senior expert advisor that enhances human understanding and problem-solving, rather than a tool for "vibe-coding" or learned helplessness.
  • 📈 Engineers should focus on mastery, continuous learning, and developing deep understanding to differentiate themselves in an AI-saturated market.
  • 🌱 Tight iteration loops and investing in robust tooling, as seen in Mojo's development and Solvent's approach, are essential for building products with durable value and maintainability.
Knowledge graph40 entities · 34 connections

How they connect

An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.

Hover · drag to explore
40 entities
Chapters20 moments

Key Moments

Transcript218 segments

Full Transcript

Topics15 themes

What’s Discussed

LLVMSwiftMojoAI CodingSoftware CraftsmanshipArtificial General Intelligence (AGI)Large Language Models (LLMs)Unit TestsTechnical DebtSoftware ArchitectureIteration LoopsJupyter NotebooksCareer DevelopmentProgramming LanguagesCompiler Design
Smart Objects40 · 34 links
People· 6
Products· 18
Concepts· 13
Company· 1
Events· 2