Should AI Be a "Senior Intern" or a Lead Engineer in Software Development?

A recent dev.to post outlines a "manifesto" for human primacy in software engineering, sparking a broader discussion on AI's role in design, architecture, and core technical responsibilities.

Where It Happened

The "LogiFlow Engineering Manifesto" was published on dev.to by @turacthethinker in a blog post titled "Back to Code | Ep 15: The New Manifesto — Master and Apprentice (Season Finale)" on May 26, 2026. This article served as the finale to a 15-week simulated journey of a company "waking up from the illusion created by artificial intelligence and returning to real engineering," drawing significant attention within the developer community.

Side A — Steelman: Human Primacy in Engineering

Proponents like @turacthethinker argue that while AI excels at generating code, the critical aspects of software engineering (design, architecture, intent, and system-level concerns) must remain firmly in human hands. The "LogiFlow Engineering Manifesto" asserts that "Architecture Belongs to Humans," specifying that AI "cannot determine Hexagonal boundaries, Domain rules, or Database schemas" because these are "insurances taken against future uncertainty." Similarly, "Intent (TDD) Belongs to Humans," with humans writing the "Red tests" and AI providing the "Green implementation," ensuring "Control and design must always remain with the human." Furthermore, "Physics and Security Belong to Humans," as AI cannot be held responsible for "Concurrency, Memory, Network, and Security audits," which require "Thinking at the machine level." This perspective positions AI as a "Senior Intern" that "must be given context, and its output must be reviewed with skepticism," serving as "an apprentice" rather than an autonomous decision-maker. The author concludes that "Coding became a commodity. But thinking... thinking is more expensive than ever."

Side B — Steelman: AI Autonomy and Advanced Roles

Conversely, proponents of more advanced AI roles in software engineering contend that AI's capabilities extend beyond mere code generation, increasingly encompassing higher-level design and architectural decisions. While not explicitly named in the source, this perspective is implied by the manifesto's opposition to an "illusion created by artificial intelligence." Advocates for this view suggest that with further advancements, AI could analyze complex system requirements, propose optimal architectural patterns, and even identify security vulnerabilities or performance bottlenecks with greater efficiency and accuracy than human engineers. They might argue that AI's ability to process vast amounts of data and learn from existing codebases makes it uniquely suited to define "Hexagonal boundaries" or "Domain rules" by identifying emergent patterns and best practices. In this view, AI could evolve beyond an "intern" to become a co-pilot or even a lead architect, handling much of the "design, the context, the system, the failure modes, and the architecture," thereby potentially reducing the need for human involvement in these areas. The goal would be for AI to eventually "put us out of work" from routine or even complex engineering tasks, freeing humans for more abstract problem-solving or product vision.

What's Underneath

The underlying tension in this debate centers on the evolving definition of "engineering" itself in an AI-augmented world. Both sides implicitly agree that something valuable remains, but they diverge on where the line between "coding" (automatable) and "engineering" (human-centric) is drawn, and how permeable that line is. The debate reflects a deeper anxiety about agency and responsibility: who holds the ultimate accountability when an AI-generated system fails, and what unique human capacities (like intuition about "future uncertainty" or "craft") are truly irreplicable by machine intelligence? It's less about if AI can write code, and more about if it can truly think like an engineer.

Pull quote: “Coding became a commodity. But thinking... thinking is more expensive than ever.”