DEV Community: Lorenzo Battilocchi

[Boost]

Lorenzo Battilocchi — Mon, 04 May 2026 10:54:08 +0000

Lorenzo Battilocchi

May 4

Technical debt and AI: is it gone?

#ai #programming #management #technology

Comments

3 min read

Technical debt and AI: is it gone?

Lorenzo Battilocchi — Mon, 04 May 2026 10:51:44 +0000

Technical debt hasn't disappeared; it has simply changed its "interest rate." While AI can rewrite a function in seconds, the definition of technical debt has shifted from "code that is hard to change" to "systems that are hard to understand."

In 2026, we are seeing a transition from traditional "Human Debt" to what many call "AI-Generated Rot." Here is why the concept is more relevant now than ever:

1. The "Black Box" Problem

Traditional tech debt was usually a "known-unknown"—you knew a specific module was a mess, but you understood why it was messy. AI-generated debt creates semantic debt: code that looks clean and passes tests, but no human on the team actually understands the logic.

The Risk: When a critical bug occurs at 3:00 AM, an engineer can’t "just ask the AI" to explain it if the AI doesn't have the full real-time system context.
The Maintenance Cost: Debugging code you didn't write (or that no human wrote) is estimated to take 3–5x longer than debugging authored code.

2. Bloat and Superfluous Logic

AI models are trained to be helpful, which often leads to "hallucinated complexity." An AI might include three different libraries or patterns to solve a problem that a human would have solved with five lines of native code.

Code Proliferation: Because generating code is "free," developers are more likely to accept 100 lines of AI code when 20 would do.
Dependencies: AI often suggests outdated or "overkill" dependencies, increasing the security attack surface and build times.

3. The Feedback Loop of Rot

This is a new phenomenon unique to the AI era. If your codebase is messy, the AI's "context window" (the code it "reads" to give you suggestions) becomes polluted with bad patterns.

Traditional Debt: Stays in one corner of the app.
AI Debt: Spreads. If your AI sees "bad" code in your repo, it will use that bad code as a template for everything else it writes for you, creating a self-reinforcing cycle of low-quality architecture.

Comparison: Traditional vs. AI Technical Debt

Feature	Traditional Tech Debt	AI-Generated Tech Debt
Origin	Human shortcuts/tight deadlines	Over-reliance on AI generation
Visibility	Obvious (messy formatting, "todo" tags)	Hidden (looks clean, but is logic-heavy)
Growth Speed	Linear (matches human output)	Exponential (matches AI output)
Primary Fix	Refactoring by senior devs	Agentic "sweeps" and strict linting

4. Where AI actually helps

It’s not all bad news. AI is excellent at "Debt Liquidation"—the grunt work of upgrading versions. For example, Amazon recently saved thousands of developer-years by using AI to upgrade legacy Java applications.

The Nuance: AI is great at migrating (moving from A to B) but still struggles with architecting (deciding if B is actually a good idea).

The Verdict

Technical debt makes more sense now because the "principal" is being borrowed faster. If you treat AI as a "magic wand" that eliminates the need for good design, you aren't getting rid of debt—you're just taking out a high-interest payday loan on your codebase's future.

The most successful teams in 2026 aren't using AI to replace refactoring; they are using it to automate the cleanup while humans focus on keeping the system's "mental map" clear.

Meet JImg2PDF: Your Simple, Private, and Offline Image-to-PDF Tool

Lorenzo Battilocchi — Sun, 15 Feb 2026 16:47:10 +0000

If you're anything like me, you've probably accumulated a massive collection of screenshot notes—from articles, whiteboards, or quick sketches—that you'd love to organize and annotate, perhaps on an iPad. The obvious solution is to combine them into a single PDF. But the reality of finding a good tool to do that? Often frustrating.

Many online "converters" are cluttered with ads, surrounded by distracting content, or, worse, hide basic functionality behind aggressive paywalls. You've likely encountered the dreaded "X MB limit—upgrade to Pro" message when all you needed was to merge a handful of images. Sure, you could open a document processor and insert each image one by one, but that process is slow, cumbersome, and wastes precious time.

I built JImg2PDF to solve this exact problem. It's a tool designed with one purpose in mind: to combine your images into a PDF as simply and quickly as possible, with complete peace of mind.

The Core Philosophy: Simple, Local, and Free

JImg2PDF is built on three key principles:

100% Local Processing: Your files never leave your device. There are no uploads to a remote server, ensuring your notes and data remain private and secure. This also means you can use it anywhere, with or without an internet connection.
No Artificial Limits: Forget about file size restrictions or tiered pricing. JImg2PDF is free to use. Whether you're combining two images or fifty, the process is the same. It's built for your workflow, not against it.
Effortless to Use: The interface is intentionally minimal. You simply import the images you want, arrange them if needed, and tap a button. In seconds, you have a single, consolidated PDF document, ready for organization, sharing, or annotation on your iPad or any other device.

Why Waste Time?

JImg2PDF strips away all the complexity. It's a focused, reliable tool for a common task. No sign-ups, no subscriptions, no distractions. Just a clean, fast way to get the job done.

If you're looking for a straightforward way to transform your scattered screenshot notes into a manageable PDF, give JImg2PDF a try. It's the tool I wished existed, and now it's here for everyone.

Find the project on GitHub