A SaaS outage can turn a workday unproductive — whether it’s a company chat app, a hosting service, or network connection.
Most programming jobs today require online tools to do real work. Yet writing code itself continues to be largely an offline task — if you understand the project and have the source locally, you can work with minimal online interaction.
Then things changed with AI SaaS tools like Claude, and most of us started using them to write less code, review it more, and pump out lines of code faster.
Until Claude had an outage, which made me realize how much I had built a reliance on it to the point I felt some of my programming skills were atrophying.
I value self-reliance in every aspect of life. In programming, I want to be able to tackle any task with as little assistance as possible. However, I inadvertently gave that away to an AI tool. Blinded by the perceived increase in productivity and distracted by the convenience, I relaxed and stopped exercising the discipline and rigor that made me a good programmer to begin with.
With the service down, I found myself using more energy to remember what to do instead of typing “do this, now find that” to an AI prompt. I became a little less independent as a programmer.
<h3>Before AI assistance, the work was something like this:</h3>
<ol>
<li>Understand the problem</li>
<li>Read the code and find the relevant areas</li>
<li>Plan a solution, maybe run some tests</li>
<li>Use my current knowledge/experience to apply the solution</li>
<li>Write tests</li>
<li>Review</li>
<li>Submit the solution</li>
</ol>
<h3>With AI assistance, I found myself doing:</h3>
<ol>
<li>Trying to understand the problem less and less</li>
<li>Let AI handle 2-5</li>
<li>Do a final review, check tests (barely) and submit</li>
</ol>
That’s a lot of important skills not being used.
I’m confident we’ll have local AI services which will largely mitigate outages. But does that justify losing (or never fully developing) those skills that make good programmers? No.
If we aren’t intentional about how we use AI, we will eventually become managers of fast, tireless “interns” that never forget anything. For some, that’s the intended change in their career.
But if my goal is to actually get better at writing and reading code, I must ensure my skills are enhanced by AI tools — not replaced by them. Only then will I remain self-reliant as a coder, capable of programming even when programming tools are not available.

Smiley Face

Isaac de Andrade

A programmer reflects on how over-reliance on Claude and AI coding tools led to skill atrophy, noticing the gap only when a service outage forced them to work independently. The post contrasts the traditional programming workflow with an AI-assisted one where critical thinking, code reading, and problem-solving steps are increasingly delegated to AI. The author argues that unless developers are intentional about how they use AI tools, they risk becoming mere managers of AI output rather than skilled programmers, and advocates for using AI to enhance rather than replace core coding skills.

The intentional, self-reliant programmer

We do not know the impact of using AI/LLMs on our minds and cognition, and its perfectly acceptable, I’d say wise even, to be apprehensive of using them too much.
We didn’t understand the impact of Social Media before it was rolled out to the masses rapidly, and now we’re dealing with incredibly impactful negative consequences, from misinformation to attention deficit.
There is already ample evidence now that points to the cognitive impact and atrophy that results from relying on these models and outsourcing our thinking. The rapid integration of AI is very much an experiment with unknown consequences.
I am starting <a href="http://confident-coding.com" target="_blank" rel="noopener nofollow">confident-coding.com</a> with that need in mind, as well. There’s an infinite number of courses to teach people how to code, but virtually none that teach people how to debug and utilize their critical and systems thinking. We desperately need to retain a focus on these concepts as it becomes easier and more accessible to reach for the infinite answer machine.

That feeling is valid. I’ve recently read an article too about a research that AI makes software engineers dumb. I think that’s scary. Programs can be exploited faster now due to weaker code security and an engineer who don’t even know why the breach happened.
Because of this, I am gearing towards true upskilling by using AI as a mentor when I get stuck. It is slow but the progress will be real. Leveraging myself this way will surely be a good investment in the future. Good engineers will be more valuable than ever.

If you outsource thinking part to AI you will start depending to much on assistant tools. If you use it to assist you with any task but you are the one who thinks and AI is there to help you when you get stuck you will improve.

I think that all of us should be aware of the enormous power of the AI tools and that with great power comes great responsibility. Not in a sense that important stuff gets built with it and it needs to be managed with care (that as well), but a great responsibility that we’re esentially outsourcing away our cognitive thinking efforts to the AI instead of thinking about the problem/solution ourselves, and that slowly makes us worse overall in our profession in my opinion.

almost an addiction, what’s life without internet?