AI Won't Take Your Job. Someone Using AI Might.
I've been using AI tools daily for a while now. ChatGPT, Claude, Copilot, various image generators - the whole ecosystem. And I've noticed something interesting: the people getting the most value from these tools aren't the ones you'd expect.
It's not the early adopters who spend hours crafting elaborate prompts. It's not the skeptics who refuse to engage. It's the people who already had solid fundamentals and just added AI as another tool in the belt.
So let's talk about what actually matters now. Not in some theoretical future where AGI does everything, but right now, today, in the work you're doing this week.
Yes, Seriously, Learn to Type Properly
I know. It feels like advice from 1995. But hear me out.
Every interaction you have with AI is text-based. Your prompts, your follow-ups, your refinements - all typed. The faster you can get thoughts out of your head and into the interface, the faster you can iterate. And iteration is everything with these tools.
I watched a colleague spend three minutes hunting for keys while trying to describe a bug to an AI assistant. By the time he finished typing, he'd lost half his original thought. Meanwhile, someone who types fluently can have a rapid conversation - try this, no that's not quite right, what about this approach - and get to a useful answer in the same three minutes.
Touch typing is throughput. It's boring, I get it. But if you're still looking at your keyboard, you're leaving value on the table.
You Need to Know What You're Asking For
Here's where things get real. AI can write code, draft documents, analyze data, and produce content. What it cannot do is figure out what you actually need.
Ask an AI to "make a website" and you'll get something generic. Ask it to "create a responsive landing page with a hero section, three feature cards, a testimonial carousel, and a contact form that validates email input" and you'll get something you can actually use.
The difference isn't prompt magic. It's that the second person understands web development well enough to specify what they want. They know what components exist, what's feasible, what matters.
This is true in every domain. The financial analyst who understands DCF modeling gets better AI assistance than someone who just asks for "valuation help." The marketer who understands conversion funnels gets better copy suggestions than someone who asks for "good marketing words."
You don't need to memorize syntax or recall every function name - AI handles that beautifully. But you absolutely need to understand the concepts well enough to direct the work and evaluate the output.
Breaking Big Problems into Small Ones
AI is really good at specific tasks. It's really bad at vague ones.
"Help me build a business" - terrible prompt, useless output.
"Draft an executive summary for a managed IT services company targeting mid-size manufacturing firms, emphasizing security compliance and cost predictability" - specific prompt, useful output.
The skill of taking a big ambiguous goal and breaking it into concrete, actionable pieces has always been valuable. Now it's the entire ballgame. Every complex project becomes a series of smaller AI-assisted tasks. Your job is figuring out what those tasks are and how they fit together.
Incidentally, this is also what makes someone good at managing people or leading projects. Decomposition is a universal skill. It just has a new application.
Developing Your BS Detector
Here's something that took me a while to internalize: AI lies. Confidently. Regularly.
It will cite academic papers that don't exist. It will produce code that looks perfect but breaks on edge cases. It will make claims that sound authoritative but are completely fabricated. This isn't going away anytime soon - it's fundamental to how these systems work.
Your job is to know when to trust and when to verify.
For low-stakes tasks - drafting an email, brainstorming ideas, getting a first pass at something - you can often use AI output with minimal review. For anything important - code going into production, facts going into a report, advice you're giving to clients - you need to check the work.
Over time, you develop an intuition. You start recognizing the types of tasks where AI is reliable versus the types where it hallucinates frequently. You learn to spot the telltale signs of confident BS. This intuition is valuable and takes time to build.
The alternative is getting burned, which also teaches the lesson, just more painfully.
Just Being Clear About What You Want
All the "prompt engineering" hype boils down to this: can you clearly articulate what you want?
If you can write a good project brief for a human contractor, you can write a good prompt for an AI. If you can explain a problem clearly in a meeting, you can explain it clearly to Claude. The skills transfer directly.
This means the people who were already good communicators have an advantage. They know how to specify requirements, provide relevant context, and flag constraints. They're used to thinking about what information the other party needs to do the job well.
If you struggle to explain what you want to colleagues, you'll struggle with AI too. The good news is that practicing with AI is a low-stakes way to get better at communication in general. It's infinitely patient and gives you immediate feedback on whether your instructions were clear enough.
Going Deep in Something
There's a temptation to become a "prompt generalist" - someone who knows how to use AI across many domains but isn't an expert in any of them.
I'd push back on that strategy.
AI amplifies existing expertise. A security professional who understands threat modeling, attack vectors, and compliance frameworks will get enormously more value from AI than someone who just knows how to ask questions about security. The expert knows what to ask, can evaluate the answers, and can integrate the output into real work.
The people I see thriving are the ones who already had deep domain knowledge and added AI as an accelerant. The ones struggling are the ones who thought AI proficiency could substitute for domain expertise.
Pick something. Get genuinely good at it. Then use AI to move faster within that expertise. That's the winning combination.
Knowing When to Put the AI Away
Sometimes AI makes things worse.
For quick tasks you already know how to do, the overhead of prompting and reviewing exceeds just doing it yourself. For highly creative work where you need to think through something, outsourcing the thinking defeats the purpose. For sensitive situations requiring judgment, AI's input might be more noise than signal.
The real skill is pattern recognition: this type of task benefits from AI, this type doesn't. This comes from experience. You have to try AI on various things and notice where it helps versus where it's a net negative.
The goal isn't to use AI constantly. It's to use it strategically, where it actually adds value.
The Bottom Line
None of this is revolutionary. Type fast, understand your domain, break down problems, think critically, communicate clearly. These have been valuable skills forever.
What's changed is the leverage. These fundamentals now multiply against an incredibly powerful set of tools. Getting 10% better at problem decomposition might yield 50% more value from AI-assisted work.
The people who invest in these basics will find AI transformative. The people who skip the fundamentals hoping AI will do the heavy lifting will be disappointed.
I know which bet I'm making.
Want to talk about how AI fits into your business operations? Ironwright works with organizations to integrate new capabilities without the hype - just practical solutions that actually work.