Why I picked this
Victor's instinct here is correct — this piece deserves to break through the paywall noise. It's a practitioner story that shows what most AI adoption discussions carefully avoid: the human supply chain behind the models. Katya (pseudonym) is a displaced content marketer now training AI for $45/hour through Mercor, with zero visibility into which model she's improving or how long the work will last. Her project evaporated in two days. The irony is structural, not accidental: white-collar workers automated out of stable roles are now the precarious labor force teaching models to do 'the worst version' of what they used to do well.
This matters for GTM leaders because it surfaces the labor dynamics you're inheriting when you deploy AI tools. Someone trained that model. Probably someone who used to do the job you're now automating. The opacity is deliberate — workers don't know 'the client,' can't assess what they're building, can't opt out of training their own replacements. It's gig economy mechanics applied to knowledge work, with none of the transparency.
The consulting relevance is immediate: if you're running AI pilots in content, copywriting, or customer support, you're participating in this system whether you know it or not. The ethical questions aren't abstract. They're embedded in your vendor contracts and your team's job security. This article is a mirror, not a warning. You're already here.
Three lenses
The opacity is a feature, not a bug — if workers knew they were training competitors to their own roles, the labor pool would collapse. But that fragility is your risk surface. Build assuming your training data pipeline has a half-life measured in months, not years.
If I'm deploying AI content tools across my org, I need to know: who trained this model, under what conditions, and what happens when that labor supply dries up? The $45/hour gig worker today is my vendor's existential risk tomorrow. Show me the sustainability model or I'm not buying.
Everyone celebrates AI efficiency gains. Nobody's pricing in the cost of burning through the knowledge worker class that makes training possible. When Katya and hundreds like her stop taking these jobs — and they will — your model quality degrades and you don't even know why. I've seen cost-cutting destroy vendor reliability before. This is that, but hidden three layers deep in the supply chain.
“My job is gone because of ChatGPT, and I was being invited to train the model to do the worst version of it imaginable”
Key takeaways
- White-collar workers displaced by AI (content marketing, copywriting) are being recruited to train the very models that replaced them, creating a cruel economic feedback loop
- AI training labor operates as precarious gig work with no job security - projects canceled with zero notice despite workers planning finances around the income ($45/hr but can disappear in 2 days)
- The AI training supply chain is deliberately opaque - workers don't know which AI they're training ('the client'), what it's for, or how their work fits into the larger system, preventing informed consent about contributing to further automation
People mentioned
- Katya, Freelance journalist/content marketer (pseudonym) @ Unemployed/Mercor contractor
- Melvin, AI interviewer @ Mercor
Companies
Key metrics
- •$45 per hour
- •hundreds of people
- •two days
- •several hours per task
Why this matters for operators: GTM leaders deploying AI content/copywriting tools need to understand the precarious labor dynamics and ethical implications embedded in their vendor supply chains; opacity in training pipelines creates sustainability and quality risks
I cover AI×GTM intelligence like this every Wednesday.
Get STEEPWORKS WeeklyMore picks
Humans will keep inventing new reasons why we must stay in the loop with agents
- Human resistance to full AI autonomy is not purely technical—it's psychological and organizational; companies will rationalize keeping humans in decision loops even when agents are capable
- The 'human-in-the-loop' requirement may become a self-perpetuating narrative rather than a genuine necessity, driven by organizational risk aversion and change resistance
- Product leaders at scale (Notion) are observing this pattern, suggesting it's a widespread phenomenon across enterprise AI adoption, not isolated to specific use cases
Comp plans for consumption pricing
- Consumption pricing fundamentally breaks traditional SaaS comp models—requires rethinking sales incentive structures around usage vs. contract value
- Four distinct contract structures exist (pay-as-you-go, uncommitted, committed, hybrid), each requiring different compensation mechanics and sales behaviors
- Enterprise consumption-based deals create tension: customers want flexibility, sales teams need predictability for quota attainment—comp design must bridge this gap
3 revenue motions your AI is only half wired into
- Model parity has arrived: OpenAI/Claude now trade evenly on core tasks, making 'better AI' a non-differentiator—the edge shifts to integration depth into existing revenue motions
- Waste is quantified: teams paying $17K-$37K/month for AI seats that never touch pipeline generation; real cost is opportunity cost of unused capacity, not subscription fees
- Lean teams have a structural advantage: cannot out-buy larger competitors on model access, but can out-embed them by wiring AI 1 revenue motion deep (pipeline → content → deals) with proprietary deal context competitors haven't seen
This analysis was produced using the STEEPWORKS system — the same agents, skills, and knowledge architecture available in the GrowthOS package.