I Used AI to Build an AI Tool. Then I Used AI to Write This Post About It.
Keep Your Job by Giving It to AI
Let’s get meta for a second.
I recently built an AI-assisted royalty contract reader—a lightweight script that can ingest a PDF royalty agreement and answer questions about it in plain language. It’s built using OpenAI’s API, and a simple embedding-based search function. You can view here:
It was part curiosity, part necessity. I’ve been working on a project in the commodity space and I figured it would be easier to ask a contract questions than to read it. If AI can summarize Kafka and debug code, it can probably handle some royalty clauses too. I turned a boring task into an easy and interactive one.
This project came on the heels of taking Wharton's AI for Business course. I wanted a solid foundation and a bit of resume padding. But what the course really drove home for me was clarity. AI isn’t going to destroy the job market, it is going to reshuffle it.
The Job Market Isn’t Dying. It’s Reformatting.
Here’s how I see it:
High-prestige, high-knowledge jobs might shrink. Think radiologists, accountants, paralegals. Not gone—just less needed at scale. And definitely using AI (or replaced by one who does)
Support roles and applied AI jobs will surge. More prompt engineers, data custodians, data pipeline plumbers, compliance watchdogs, energy system builders, data centre architects and yes—more people like me with specific domain knowledge stringing together duct-tape GPT projects that actually do useful things.
It’s not a collapse. It’s a reconfiguration.
Humans are control freaks. Most AI workflows will keep a “human-in-the-loop” model for decades. It’s the same instincts that make us think vinyl records sound better, we just like to have a say when things change.
History’s Playbook Of Generative & Disruptive Tech
There’s a technology from every era that futurists claim is going to crash the job market, but we always just end up with new jobs.
This pattern isn’t new. A few examples:
The printing press cost some monks their scroll-copying gigs. But it created publishers, journalists, editors, bloggers, and “thought leaders”(place me here please)
Robotics didn’t erase factory work; it transformed it. Operators became technicians.
I legitimately thought blockchain would gut institutions. Then BlackRock showed up and became the largest Bitcoin ETF issuer.
Tech kills the status quo, not the economy. We panic in the short term and adapt in the long term. Same cycle every time.
We also tend to overestimate the productivity impact of new tech in the short-term and underestimate it when looking out over the long-term. So consider that the hype you’re hearing today is likely just early chatter that will be nothing when compared the disruption we’ll see towards the end of the decade.
Evolve or evaporate.
Right now, the best thing you can do is dive in. Build something. Break it. Learn what you’re automating and what you’re not. If AI can do it, let it. Focus on the tasks it can’t.
My royalty reader is clunky, but it works—and it’s just the beginning.
…brb. Going to panic-learn SQL and RAG.
<3
Derek
Solid take. One stat I always come back to:
“63% of tasks currently done by humans could be automated by 2045.”
(Source: McKinsey Global Institute, 2023)
But here’s the catch tasks ≠ jobs.
Most jobs are a mix of 20+ micro-tasks. AI replaces pieces, not people.
At least not all at once.
It’s like replacing ingredients in a recipe you need to know which ones hold the dish together.