I Was Afraid of AI

I admit it. I felt a very deep and ever-present gnawing fear that the entire bottom of my industry would collapse. That the process by which I had come up in it was the only way for a person to gain the kind of experience they need to become a professional software engineer or architect. Here I am, nearly forty years old thinking in the very same way I once looked down upon.

Of course there are other ways for people to go from curiosity about software development to being expereinced in the craft. They don't need to slave in services industry jobs working sixty hour work-weeks to meet somebody's deadline that was never reasonable. Some of the best software engineers I've met in my twenty-year career have come into the craft from another place. Often sciences, engineering, or even psychology.

The ramp for entry level developers starts very smoothly now that they can learn while doing with AI. I've seen junior developers and mid-level developers pick up Copilot or Claude Code and skyrocket in producivity. I've seen their understanding increase rapidly and their questions become much more complex and insightful over a matter of days and weeks where it would have been weeks to months before. That said, the ramp steepens rather quickly now. Only those who are able to quickly synthesize new information are going to excel in this fast paced environment.

I have spent the better part of this year building a complex, multi-cloud commerce platform for a business that deals with customizing other products on behalf of businesses, teams, associations, and individuals. I've been using AI to do it. In small ways at first, but now — in our tenth month of development — I have become reliant upon this new tool.

Not only have I come to understand, trust for some things, and depend on AI to deliver value for my client; I have also begun thinking about digital products and how agents need to be able to read, write, transact, and relate with systems I build and the data they maintain. I think about my application's consumer, the person, yes but also their agent. I build UIs for humans and MCPs for machines. The APIs bubble out of both.

Nine months ago I was afraid of how my industry would crumble. Yesterday, that same AI that I was afraid of did actually cause a tremendous amount of damage. The PHP vulnerability, the poorly created certificates (4), and NodeJS all put an immediate halt on every team's roadmap for 48 hours. Some portion of the web went dark to prevent further damage. In my team, I was able to remediate my personal machine by cleaning up 4 years of NodeJS installed tech debt on my system and also patch our entire application stack in an hour. My team did the same on their PCs and moved on in about the same amount of time.

Sustainability

Our Earth is reeling from the energy demands we are placing upon it. Captialism refuses to allow us to slow down. So I expect to see — largely out of necessity — solar-powered home appliance compute nodes in our near future. Especially with public backlash against massive scale datacenters. Perhaps they'll become a bit like blade server architectures. It's already happening in the form of mini-pcs on the compute side. In cities we're seeing adoption of railing-mounted panels designed for parital-off-grid apartment use. The ai compute mini-pc running on that would be an ideal carbon neutral way of responsibly using AI.

No doubt the future can be bright if we as a society agree to use AI responsibly. Business has continued to show us that it typically cannot be trusted to do the right thing. Shareholder value is the only priority. When that wins, we all lose. So we must celebrate the advancements in biomed research, in engineering, and in materials science. We must use those advancements to de-risk our future. So there is some hope. However, I still fear that the circular debts of Microsoft, Google, Nvidia, OpenAI, Anthropic, and more are all hastening a massive collapse. Hopefully I'm wrong.