The Hidden Cost of the AI Boom: How Corporations Are Using Job Seekers to Train the Models That Replaced Them

The Layoffs Are Real. So Is What Comes After.

Over the past three years, some of the most profitable companies in the world have laid off hundreds of thousands of workers. The stated reason, more often than not: AI and automation will handle it from here.

And yet, if you've been job hunting recently, you may have noticed something strange. You're being asked to complete "skills assessments" before interviews. You're submitting writing samples to make it through an applicant tracking system. You're answering multi-step scenario questions as part of an online "pre-screening." You're doing unpaid "take-home projects" just to get a first call.

Here's the question no one is asking loudly enough: Where does all that data go?

The Pipeline No One Talks About

Large language models — the AI systems behind tools like ChatGPT, Copilot, and countless enterprise automation platforms — are trained on human-generated data. The more domain-specific and high-quality that data is, the better the model performs.

Writing samples. Problem-solving walkthroughs. Customer service role-plays. Sales call simulations. Code challenges. Marketing briefs.

These are exactly the kinds of tasks that skilled professionals do every day — and exactly the kinds of tasks that corporations are now asking job applicants to complete, for free, under the guise of "evaluation."

You are not just being assessed. In many cases, you are doing real work. And that work — your thinking, your writing, your professional judgment — may be going directly into training pipelines for the very systems designed to make your role obsolete.

This Is Not Paranoia. This Is a Pattern.

Consider the sequence of events:

1. A corporation announces record profits, then lays off 10–20% of its workforce citing "AI efficiency gains."
2. The same corporation continues hiring — but at lower levels, with more screening, and with longer, more involved application processes.
3. Job seekers, desperate for income, comply with every request: assessments, assignments, simulations, samples.
4. The corporation collects this data at scale, across thousands of applicants, for every open role.
5. That data is used to fine-tune AI systems that will handle the work those roles once covered.

None of this requires a corporate conspiracy. It just requires incentives — and right now, the incentives are perfectly aligned against the job seeker.

The Social Engineering Angle

What makes this particularly insidious is how it's been normalized.

The language of the modern hiring process has been carefully shaped to make compliance feel reasonable. "We just want to see how you think." "This is a standard part of our process." "It should only take 2–3 hours."

Two to three hours of professional-grade work. Unpaid. Repeated across dozens of applications. For a job you may never get — but whose application process just trained a model that might handle that job next year.

This is social engineering at scale. It exploits the power imbalance between a desperate job seeker and a gatekeeper corporation. It normalizes the extraction of labor under the guise of evaluation. And it has been remarkably effective, because the alternative — refusing — often means disqualification.

What You Can Do

Knowing this doesn't make the job market easier. But it can make you a more informed participant in it.

Be selective about what you submit. If a pre-interview assignment looks suspiciously like real deliverable work — a full marketing strategy, a working code module, a detailed operational plan — that's worth noting. You can ask directly: "How is this used beyond the evaluation process?"

Time-box your effort. A genuine skills assessment rarely needs more than 30–60 minutes of your time. If it's asking for more than that, recalibrate what you're willing to give.

Watermark or limit specificity. Submit enough to demonstrate your thinking without providing a fully polished, production-ready deliverable. Show your process, not a finished product.

Talk about it. The more job seekers share these experiences publicly — on forums, on LinkedIn, in communities — the harder it becomes for corporations to normalize the practice quietly.

Value your expertise. The skills that make you a strong candidate are exactly the skills that make your outputs valuable training data. That's not a coincidence. It's the point.

The Bigger Picture

We are living through one of the largest transfers of economic value in modern history — from workers to capital, mediated by AI. The workers being displaced aren't just losing jobs. In many cases, they're also being asked to provide the raw material that makes their displacement permanent.

That's worth being angry about. It's worth being clear-eyed about. And it's worth organizing around — whether through policy, through professional norms, or simply through the choices you make about what you're willing to give away for free.

Your knowledge, your judgment, and your professional craft are not free. Don't let the hiring process convince you otherwise.