Prolific competing for personality data

Prolific is trying to turn worker identity into model behavior data. Mercor and similar marketplaces win when labs need a cardiologist or a tax lawyer to grade answers. Prolific wins when labs need a person who sounds patient, skeptical, funny, culturally local, or emotionally steady across many interactions. Its long running participant records, prescreening system, and representative sampling make it better suited to collect those traits at scale.

• Prolific already sells access based on self reported demographics, attitudes, and curated participant groups, and it supports representative US and UK samples. That is the raw machinery needed for companion, education, and support model training, where the goal is not expert correctness alone, but responses that feel like they came from a specific kind of person.
  2 sacra 3 prolific 4 prolific
• This is a real wedge against Mercor. Mercor has been built around pre vetted domain experts for RLHF, with revenue scaling around high skill labor for law, medicine, and other expert tasks. Prolific is pushing toward the opposite end of the market, where the scarce input is human style and social grounding rather than formal credentials.
  1 sacra 5 sacra 6 sacra
• The end markets make that distinction valuable. AI companion products already compete on conversation quality and persona consistency, and regulators and researchers are increasingly focused on safety in therapy like and youth companion use cases. That raises the value of training and evaluation data from known, recontactable humans with measured background traits, not anonymous clickworkers.
  7 sacra 8 esafety 9 arxiv

As AI moves from solving bounded expert tasks to spending more time in relationships and coaching loops, the best data marketplaces will look less like staffing firms and more like controlled panels of humans. Prolific is positioned to become core infrastructure for that shift, because the next bottleneck is collecting repeatable signals about how different kinds of people actually talk, react, and build trust.