AI vs Jobs

Hypothesis: If I'm an AI Prompt Expert, My Job Will Be Protected

Discussions about generative AI’s invasive assault on jobs are happening everywhere you go. Starbucks. Linkedin. X. It’s a pervasive concern.

Will AI take our jobs?

My take? None of it matters. Worry is the interest on a debt that may never need to be repaid¹.

StableDiffusionX: AI vs Jobs

AI doesn’t give a crap about your job. It has no idea it is in a subliminal war to displace you and millions of workers like you. Sure, you might train it to know this, but what would that achieve except to bias its output with cynicism?

I think our time, however limited as influencers in the workforce, is better spent pondering defensive tactics. And you know what a good defense means? Having a great offense.

Being the company's best generative AI prompting expert doesn't ensure anything except that you might see the second buzzsaw coming sooner than your other co-workers.

Imagine GPT-5² is released in about ten more AI minutes³, and it has new magical properties that transform the rigid and often frustrating aspects of today's prompt requirements such that they vanish entirely. Even a slight improvement could render all we know about prompt engineering irrelevant in the blink of an eye.

I hope all the newfound prompting expertise is not erased so quickly, but I peg the probability at even odds by mid-2025. Enjoy the run.

The “smart” workers — so it goes — have quickly learned how to use generative AI for greater productivity, faster output, and an overall desire to be more valuable to an organization or clients. AI undeniably helps you achieve this. As a proficient generative AI worker, you may have heightened your personal brand, emerging as the “AI guy”.

There’s plenty of encouragement on the Interwebs to become that AI guy and guarantee your job well into the next decade. I think it’s bad advice.

With a firm grasp of GPT-4, you are considered a God. Praise, stature, and wealth adorned your every work moment. You are draped in a glow of AI mastery, gleaming like Nefertiti at the Spring Solstice concert. But when GPT-5 ships, there’s an even-odds risk you will be "reclassified" as just another peasant by a vector embedding algorithm.

The irony is that you worked your butt off mastering AI, demonstrating to your company that it was powerful enough to reimagine how to do business at a faster pace and with less effort, all while making the demise of less skilled coworkers arrive sooner.

You are standing on a rock feeding the alligators, hoping they’ll eat you last.

This is a dark, cynical picture I literally "painted" with the technology that stands patiently awaiting the perfect moment to reveal my own demise, which I eagerly lean toward irrationally, hoping that something better lay ahead. Unlike the multitudes of AI hucksters and prompt “experts” claiming AI is our future, I’m under no illusion that what lay ahead could be nothing. Not even a gold watch and a party.

The few skills I possess could be at or near the end of economic demand.

To all generative AI prompting "experts": you're standing in the swamp with a fixed amount of meat (your AI-challenged coworkers) feeding the alligators, hoping they will eat you last.

Are We Wrappers?

We learned quickly that generative AI startups based fundamentally on a “wrapper”⁴, are being wiped out. Millions in VC money and thousands of jobs - gone in an AI minute as fast as they emerged seemingly out of thin air. These businesses lack a technological foundation based on innovation and software architectures that add value to some form of generative AI process. Sometimes, wrappers tiptoe across the line and discover a moat.

Almost all gen AI wrappers are moat-less. Anyone can duplicate them. Custom GPTs in the new GPT Store are vastly prompt wrappers made by well-meaning GPT developers with deep domain expertise. They’re learning why wrappers are mostly a waste of time.

Linkedin - “GPT Store is a disaster”.

Today's generative AI skills -- especially prompting skills - are simply another form of wrapper. You think you are building a personal moat to defend against job obsolescence. However, proficient AI workers may be unwittingly setting the table for workplace disruption and, possibly, more rapid disruption of their own jobs.

You must ask yourself - is my use of AI exposing the weaknesses that workers have in the new context of generative AI?

We see this sentimental delusional thinking a lot.

At the heart of work are people who will always be required to reason, understand customers, and formulate innovative ideas that make business great.

It is often a valid argument until circumstances arise, showing that it isn’t. That’s the hidden side of automation. And AI is simply a form of intellectual automation.

While humans are needed to identify weaknesses and automation patterns, once they’re successful, they become obsolete. Automation that ultimately displaces the worker AND the automator is not uncommon. It’s why businesses tend to hire external consultants to build automation systems.

One can argue -

There needs to be an automation support and maintenance role.

Agreed. However, these new AI-inspired roles are a small fraction of the roles displaced by automation. Net-net, the headcount is lower.

Using AI to increase personal productivity is undeniably a good strategy today. The premise that there are sustainable new roles that AI will create may be flawed or temporary attributes of the reimagining of work that's beginning to take place.

I've tried to imagine future AI roles; in every case, AI itself will eventually threaten that role. Feel free to help me envision a list of roles conjured by AI that AI will not soon replace.

Another fallacy…

AI won’t take your job; someone else effectively using AI will take your job.

By extension, then, you don't need to outrun the bear. You only need to outrun your hiking partner.

Strategy outcomes that are determined by relative pace are self-limiting. In my view, co-workers are not the competitive threat we should be concerned with. Instead, it's AI itself. This is the real threat.

Workers are not magically moved into a protected class because they know how to write a prompt better than other co-workers. The overwhelming trend on Linkedin is that all the AI prognosticators advise us to …

Learn AI. Become proficient with it. Focus all your energies on AI because it is the future!

A better defense strategy is to fine-tune your skills that AI cannot touch. Or, reshape existing skills such that you are differentiated in a manner that makes you impervious to AI’s likely propensity to displace workers. Learning new skills adjacent to, but perhaps very relevant to, generative AI’s future seems like better advice.

Learning and using AI in your current skill set context may accelerate your demise. It may achieve precisely the opposite of what you want.

Engaging with AI in the work you have deep experience in may ultimately reveal the displacement pattern for management.

During WWII, bombers returning from France were riddled with holes. Armorers in England were kept busy adding bullet-proof shields where holes were pervasive until a data scientist identified the bias that made this senseless. Armor was needed in all the areas that prevented bombers from returning. This is the survivorship bias.

The “Survivorship Bias”

A similar bias probably exists for workers who aggressively lean into generative AI. As an AI user, you may be engaging in your own personal survivorship bias.

Best Defense is a Good Offense

Surviving AI’s assault on the workforce will play out in two dimensions… (i) expanding your skills that AI cannot easily displace and (ii) combining adjacent skills with AI to create new value.

At the root of this challenge is knowing that generative AI is a fast-moving freight train, and you need to step off the tracks. But where do the tracks run? This is a difficult question. Knowing where they’re not likely to run is equally beneficial.

Let’s explore a few possibilities.