Do I need to level up? Navigating the evolution of AI skills

We’re witnessing a fascinating shift in how we work with artificial intelligence. If you’ve started incorporating AI into your workflow, congratulations! You’ve reached AI Level 1. But here’s the thing – new levels keep emerging, not quite as rapidly as the technology itself evolves, but quickly enough that we need to pay attention. Level 2 is already here, Level 3 is approaching, and Level 4… well, we’ll get to that later.

AI level 1: Basic tasks

At this foundational level, you’re using AI to handle everyday tasks – writing emails, generating code snippets, or summarizing those lengthy Slack threads that nobody has time to read. These are activities you’d do anyway, but now with AI assistance that makes you more efficient.

Sure, you can improve your Level 1 skills with specialized agents and better tools, but the fundamental approach remains the same: AI helps you do what you already do, just faster and often better.

AI level 2: Higher value tasks

At Level 2, you’re not just doing existing tasks better – you’re accomplishing things that previously seemed impractical or impossible. The mundane parts that would have consumed too much time manually (extensive research, creating boilerplate content, processing large datasets) are now handled efficiently by AI.

For software engineers, this might mean focusing more on architecture, code review, and refactoring – keeping code clean and extensible so your systems don’t collapse under their own weight. The AI handles the routine coding while you direct your attention to design decisions that have greater long-term impact.

The transition from Level 1 to Level 2 isn’t just about using more sophisticated tools; it’s about shifting your focus to areas where human judgment still adds the most value.

AI level 3: Reimagining your work

Perhaps the most exciting aspect of Level 3 is the democratization of expertise. Other departments in your organization are using many of the same AI tools you have access to. This creates unprecedented opportunities for cross-functional collaboration:

Imagine being a software engineer who can use AI to create a compelling product proposal written in your specific product manager’s style and vocabulary. You could generate a realistic schedule based on past project timelines, and even develop a presentation to sell your idea – all during your lunch break.

This isn’t about replacing your colleagues’ expertise but rather about bridging communication gaps and getting ideas flowing more freely across departmental boundaries.

The multi-agent workforce: A critical consideration

You might fantasize about having a team of 10 AI assistants at your command. What would you accomplish? How would you use that time, resource, and capability?

But here’s the crucial point that can’t be overemphasized: You still need to competently evaluate their work.

If you’re using an agent to generate code, you need enough programming knowledge to assess whether that code is well-designed and maintainable, does it have the right tests, and so on. You don’t need to be an expert in every language or capable of writing everything from scratch, but you must have the minimum competency to recognize quality and identify potential issues. You can’t afford to produce code that will become a bittle mess of dried out spaghetti that can’t be untangled after a year of updates.

This requirement applies across domains – whether you’re using AI for marketing content, financial analysis, or design work. The tools can dramatically extend your capabilities, but they can’t replace your judgment.

Time to level up

As promised, we’ll save Level 4 for later. For now, if you’re still at Level 1, it’s time to get moving and develop the skills needed for Level 2. And if you haven’t even reached Level 1 yet… what are you waiting for?