You Don’t Need AI Everywhere (And That’s a Good Thing)

⚠️ This blog post was created with the help of AI tools. Yes, I used a bit of magic from language models to organize my thoughts and automate the boring parts, but the geeky fun and the 🤖 in C# are 100% mine.

End-of-year reflection note
As the year wraps up and we all start thinking about what we built, what worked, and what didn’t, I’ve been reflecting on something I’ve seen a lot lately: AI being added everywhere… sometimes without a real reason.

AI is powerful. I use it daily. But that doesn’t mean it should be the default answer to every problem. And yes, I’m fine, not kidnapped and writing this 😊

TL;DR

You don’t need AI everywhere because:

✅ Some problems are already deterministic
💸 AI adds cost, latency, and operational complexity
🧩 AI can introduce hidden bugs and technical debt
🛠️ Classic programming is often simpler, faster, and more reliable
🎯 AI works best when the problem is fuzzy, not well-defined

I’ll explain each of these below — with a real example from my podcast episode editor workflow.

The Cost of “AI by Default”

We’re in a phase where “AI-powered” has become a feature by itself. But adding AI without a clear justification can actually make things worse.

Studies and real-world reports show that:

AI-generated code often introduces more defects than human-written code
Teams accumulate technical debt faster when AI output isn’t carefully reviewed
Automation bias can lead people to trust AI output even when it’s wrong
Many teams lack the operational maturity to run AI reliably in production

AI is powerful — but it’s not free, and it’s not magic.

Where AI Shines (And Where It Doesn’t)

AI is fantastic when:

The problem is ambiguous
You’re working with language, meaning, or creativity
You need pattern recognition at scale

But when the problem is:

Predictable
Rule-based
Fully deterministic

…then classic programming usually wins.

A Real Example: Podcast Editing

While working on my podcast workflow, I automated the editing phase:

remove silence
normalize audio
add intro/outro
mix background music

This is a perfectly deterministic pipeline.

I could have added AI.
But I didn’t — on purpose.

Instead, I built a small app using standard audio-processing techniques. The result:

predictable output
no inference errors
no API costs
easy to debug and evolve

AI wasn’t required to solve the problem well.

This Is Not “Anti-AI”

This is about intentional engineering.

AI should be used when it:

clearly adds value
reduces complexity instead of increasing it
solves a problem that rules alone can’t

Sometimes the most professional decision is saying:
“This doesn’t need AI.”

Final Thought

As we head into a new year, my personal rule is simple:

Use AI when the problem demands intelligence — not just because the tool exists.

Part 2 of this reflection will be more technical and aimed directly at developers, with a simple framework / step-by-step guide that I use to decide when AI makes sense.

📚 Resources

This reflection is not based only on my personal experience.
There is growing research, industry reporting, and analysis from very smart people showing that adding AI without a clear justification can introduce cost, risk, and complexity without proportional benefits.

AI-generated code contains more bugs and errors than human output
https://www.techradar.com/pro/security/ai-generated-code-contains-more-bugs-and-errors-than-human-output

Nearly half of all code generated by AI found to contain security flaws
https://www.techradar.com/pro/nearly-half-of-all-code-generated-by-ai-found-to-contain-security-flaws-even-big-llms-affected

AI slows down some experienced software developers, study finds
https://www.reuters.com/business/ai-slows-down-some-experienced-software-developers-study-finds-2025-07-10/

Developer Productivity in 2025: More AI, but Mixed Results
https://thenewstack.io/developer-productivity-in-2025-more-ai-but-mixed-results/

Why 95% of AI pilots fail — and what business leaders should do instead
https://www.forbes.com/sites/andreahill/2025/08/21/why-95-of-ai-pilots-fail-and-what-business-leaders-should-do-instead/

Bias in AI: Examples and 6 Ways to Fix It
https://research.aimultiple.com/ai-bias/

AI trust paradox
https://en.wikipedia.org/wiki/AI_trust_paradox

Echoes of AI: Investigating the Downstream Effects of AI Assistants on Software Maintainability
https://arxiv.org/abs/2507.00788

Replication crisis
https://en.wikipedia.org/wiki/Replication_crisis

Happy coding!

Greetings

El Bruno