World's Fair 2026
Should AI Engineers Still Read Code in 2026? The Z/L Continuum — Alex Volkov, ThursdAI
Overview
The talk explores the evolving role of AI engineers in 2026, specifically addressing whether they still need to read code generated by AI agents. It presents two opposing viewpoints: one advocating that code is now "free" and human intervention is unnecessary, and another emphasizing the critical need to read every line of code. The speaker proposes that the true question is not *if* engineers should read code, but rather *what level of proof* is required for specific tasks, suggesting a continuum based on task criticality.
Who should watch
- AI Engineers
- Product Managers
- Builders of AI systems
- Those concerned with code quality and AI agent output
- Individuals navigating the rapid changes in AI development
Key takeaways
- AI-generated code is rapidly increasing, with platforms like GitHub seeing a 14x surge in commits, largely driven by AI assistance.
- A significant shift is occurring where engineers are supervising or babysitting agents rather than handcrafting most code.
- Two contrasting perspectives exist: one that code is cheap and humans should focus on prompts and guardrails, and another that critical code must be meticulously reviewed.
- The "Zeal Continuum" is reframed from a people-centric to a task-centric model, where different tasks require different levels of scrutiny.
- Critical code related to authentication, money movement, permissions, and irreversible data requires thorough human review.
- AI agents are adept at decomposing code and can be leveraged for tasks like generating tests and identifying critical code sections.
- The emergence of "loops" in AI systems, where agents grade their own work, represents a shift in how AI systems are built and verified, but does not eliminate the need for human judgment.
- Flexibility and continuous learning are essential for AI engineers as model capabilities advance at an unprecedented pace.
Notable quotes
*Code is free. It's free for you to produce, free to refactor, and it is not a thing to get hung up on anymore.* — Ryan LeFebvre
*Critical code, read every line.* — Mario Zechner
*It's never felt so tempting to stop looking at code at all. But don't do this in production.* — Andrej Karpathy
Unofficial community note. Prefer the recording for nuance.