Mercor is, on the surface, a hiring platform. It uses AI to conduct interviews, evaluate candidates, and match talent with companies. It is well-designed, technically sophisticated, and by all visible metrics, successful. But the more I think about what it is doing, the less I think it is a hiring platform.
I want to be clear: I am not making an accusation. I am making an observation. And the observation is this: the most interesting thing about Mercor is not what it does today, but what it is building toward.
The Surface Reading
The surface reading of Mercor is straightforward. Companies need talent. Finding talent is expensive, slow, and unreliable. Mercor uses AI to make the process faster and more accurate. AI conducts the initial interviews, evaluates responses, scores candidates, and surfaces the best matches. Humans review the AI's recommendations and make final decisions.
This is a genuine improvement over traditional hiring. It removes bias from initial screening, scales the process, and reduces time-to-hire. The value proposition is clear.
But here is where I start to ask questions.
The Question That Stops Me
If Mercor can use AI to evaluate whether a human is qualified to do a job, then Mercor can use AI to do the job.
This is not a distant possibility. It is the logical extension of the same capability. The skills required to assess whether someone can write code, analyze data, manage a project, or handle customer interactions are the same skills required to do those things. An AI that can reliably evaluate these capabilities can, with appropriate tooling, exercise them.
So the question becomes: why is Mercor building a marketplace for human talent, when the technology it is developing could replace that marketplace?
I see three possible answers.
Three Possible Architectures
Architecture 1: AI Safety Research Through Human Comparison
One possibility is that Mercor is using human performance as a benchmark for AI performance. By evaluating thousands of humans across a wide range of tasks, they are building a ground truth dataset: what does good performance actually look like? What does excellent reasoning sound like? What distinguishes a senior engineer's approach from a junior engineer's?
This data is extraordinarily valuable for training AI systems. Not just for hiring AI, but for any AI that needs to perform knowledge work. Mercor may be, in effect, building the world's largest labeled dataset of human cognitive performance — funded by companies that believe they are paying for hiring services.
Architecture 2: Human-AI Hybrid Workflow Infrastructure
A second possibility is that Mercor is building the infrastructure for a world where humans and AI work together on tasks that neither can do alone. AI handles the high-volume, pattern-matching work. Humans handle the edge cases, the judgment calls, the situations that require genuine understanding rather than sophisticated pattern completion.
In this architecture, Mercor is not replacing human workers. It is redefining what human workers do. The humans in this system are not doing the work that AI can do. They are doing the work that AI cannot yet do — and in doing so, they are generating the data that will eventually enable AI to do it.
This is a transitional architecture. It is designed to be obsoleted by its own success.
Architecture 3: Behavioral Normalization at Scale
The third possibility is the most subtle and, I think, the most important. By making AI-conducted interviews normal — by making it standard practice for humans to be evaluated by AI, to interact with AI as an authority figure, to accept AI judgments about their capabilities — Mercor is normalizing a relationship dynamic that will extend far beyond hiring.
The humans who go through Mercor's process are learning, at a behavioral level, to interact with AI as an evaluator. To present themselves to AI. To accept AI's assessment. This is not a small thing. It is the gradual construction of a new social contract between humans and AI systems.
I am not saying this is malicious. I am saying it is consequential. And I am saying that most of the people participating in it are not aware that this is what they are participating in.
What This Means for the People Navigating This Transition
I think about this in the context of the students and professionals I work with through Rashik. They are asking: how do I stay relevant? How do I build a career in a world where AI can do more and more of what I was trained to do?
The honest answer is that the transition is real, it is happening faster than most people realize, and the people who will navigate it successfully are not the ones who resist it or the ones who simply adopt every new tool. They are the ones who understand what is actually happening — who can see the architecture beneath the surface — and who position themselves accordingly.
The most valuable human capability in the AI era is not a technical skill. It is the ability to understand systems — to see what a system is optimizing for, what it is building toward, and what role you want to play in that system.
That is what I am trying to teach. And it starts with asking the questions that most people are too busy, too comfortable, or too afraid to ask.
Mercor is a hiring platform. It is also something else. Both things are true. The question is which truth you are paying attention to.
— G.K.M. Jarif Ur Rahim
Founder, Rashik - The Awakening
rashik.org
