EasyStarter job$250–$550/mo

Preference Ranking

License pairwise preference: given the same task, two agent rollouts, and your lived judgment, pick the better one and write the reason. This is the dataset Scale, Surge, Mercor, and frontier labs still cannot synthesize.

Palace Wing: Preference Ranking

This venue is a Wing in your Human architecture. Every exercise creates Rooms organized by Hall type. Data quality is tracked via Knowledge Graph triples.

Halls populated: Facts, Events, Discoveries, Preferences, Advice
Tunnels form when you work across multiple venue Wings

Readiness Requirements

Meet all milestones to unlock buyer engagement. Hybrid threshold — both content volume AND time consistency required.

40

Entries

21

Active Days

21

Min Days

Why Robotics & AI Companies Pay Premium

Agentic companies train reward models on human preference, not on more tokens. A ranked pair with a written rationale is worth more than a thumbs-up. You already do this when you review work.

How to Create This Content

3–5 hours initial, then 20 min/week · 4 steps · Each step builds your architecture

Output Specification

Final Deliverable Format

Preference pairs with rationale (JSON: prompt, A, B, winner, reason, stakes)

Quality Checklist
  • Minimum 40 ranked pairs from real work you did, not invented tasks
  • Each pair includes task, rollout A, rollout B, winner, and why
  • Stakes tagged (low / medium / high) and domain tagged
  • No confidential customer data in the traces
  • Ties allowed only with a written reason
AAAK Buyer Preview

When readiness is achieved, an AAAK-compressed summary of your data is generated for buyer evaluation — privacy-preserving, compact, and instantly readable by any AI system.

Active Buyers (4)

Scale AI

AI Company
High Demand

Largest buyer of human preference and RLHF data for frontier and agent models.

Looking for: Paired agent trajectories with written rationale
Accepts: JSON
Min. volume: 40+ pairs per bundle
Visit

Surge AI

AI Company
High Demand

Expert-only preference and critique data for labs that will not accept cheap click-work.

Looking for: Domain-expert pairwise rankings
Accepts: JSON
Min. volume: 30+ pairs
Visit

Mercor

AI Company

Matches domain experts to AI labs that need preference and eval work.

Looking for: Professional judgment on agent outputs
Accepts: JSON
Min. volume: 25+ pairs
Visit

Anthropic

AI Company
High Demand

Constitutional AI and Claude agent products consume human preference and critique at scale.

Looking for: Nuanced preference with safety-aware rationale
Accepts: JSON
Min. volume: 40+ pairs
Visit

Ready to start building your Preference Ranking Wing?

Open the Data Builder to complete guided exercises. Each entry builds Rooms, Halls, and Knowledge Triples in your Human architecture — reaching buyer readiness in 30–60 days.