Maintain Your Technical Edge With AI

How did you build your skills to begin with? You probably read books, followed tutorials, built something simple, experimented, then built something more advanced, referenced technical documentation, maybe discussed with more senior engineers, and refined your skills until you were comfortable building “real” things.

Is there a reason you cannot follow this process with AI? I’d argue that with AI, each of the steps are easier with faster turn-around. Plus, they can now be specifically catered to your learning path and examples.

Worried that the AI does too much and you’ll lose your edge? You can control how much work the AI does for you, all the way down to: “Do not execute any commands or produce any code.” That leaves you in total control with a custom-built, on-demand tutor able to access nearly any technical topic at any level you want and deliver it in any format that works best for you.

If you do not think this is true, you have either not tried recent models, or you may need to refine your prompting methods.

This applies to junior and senior level software engineers alike. In fact, I’d argue that senior engineers are particularly well-suited for AI-assisted learning for 2 reasons:

There is a “No man’s land” for seniors that is hard to develop after a certain point in their career. Take any skill or topic that would benefit the senior. For sufficiently new skills or advanced topics, no worries. Seniors are meant to dig into these and have room to maneuver. But what if it isn’t a new skill (git)? What if it is a common tool (curl) or a well-worn path in day-to-day work (Jira, Github, Jenkins, Linux, OpenSSL, AWS)? These are all examples of assumed expertise and may be tough for a senior to develop “in the open”.
Seniors are more capable of digging deeper into what an Anthropic study called “conceptual inquiry”. Pushing back on a concept and asking questions of your agent may feel more natural to a senior. Just as important, having an agent ask questions in return was shown to improve retention. Seniors will recognize this as an AI-assisted rubber duck exercise.

With AI, the process is risk-free and on-demand. “No man’s land” is replaced by a private Socratic forum that you can choose to load up with whatever training you want whenever you want it. If you are unable to develop your skills with AI, then start by developing your AI skills.

Try learning something new about a tool you use everyday. You may not be aware of a gap in skills, but with AI, you can regularly ask about your typical workflows probing for alternate approaches. You may find an improvement or discover a new approach to an old problem. Either way, you are now more confident in your approach and understand the options and trade-offs.

Exercise

The next time you are working on a feature branch, discuss each git command you are about to issue. You don’t have to ask “How do I…” You can still be the expert talking to another expert and ask, for example, “I’m about to rebase onto main, I typically pass the following flags:…

Is there a way to streamline this flow? Any git config defaults I could use to reduce keystrokes? Any safety nets I should put in place to prevent issues?”

This is just one small example, but I’d be surprised if you don’t pick up a new trick or two regardless of how long you’ve been using git. It takes less than 5 minutes and there is no social risk.

Once you put this tactic to work, you will find yourself making a list of the most obvious, familiar skills to improve upon. For example, I have had 2 significant ramps in my Vim knowledge: 1) when I first learned Vim on Solaris, and 2) over the past few months asking AI to watch for areas I could improve.