Research map / Learning theory and algorithms / Prediction and decision losses
A Sustainable AI Economy Needs Data Deals That Work for Generators
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Position paper arguing the ML value chain has an "economic data processing inequality" that strips value from data generators, analyzes 73 public data deals, and proposes the EDVEX framework (task-data matching, auditable lineage, utility-based valuation).
Problem
Data generators receive little value from ML data deals: value accrues to aggregators and model monetizers, provenance is lost, bargaining is asymmetric, and pricing is static and opaque.
Approach
Frames the ML data value chain as an economic data processing inequality driven by three faults (invisible provenance, asymmetric bargaining power, inefficient price discovery) and proposes an Equitable Data-Value Exchange framework with task-data matching via sandboxed small-scale utility estimation, lineage tracking, and utility-driven pricing with Shapley-style revenue sharing in dynamic data u
Key results
- Of 73 analyzed data deals, 57 have undisclosed payment and only 6 include a generator revenue split, with only one publicly reporting actual revenue (~$2.5k).
- Disclosed revenue across the deal sample totals $677.3m, while creator royalties are described as rounding to almost zero.
- More than 40% of known transactions had OpenAI/Microsoft, Google, or Perplexity as the buyer.
Surprising findings
- Of 73 deals, 57 do not disclose payment amounts, indicating widespread opacity.
- Some deal sources report a revenue volume (e.g., Reddit $60m/yr from Google) while generators receive no share.
Limitations
- Deal dataset relies on public disclosures and likely undercounts private or NDA deals, skewed to large English-language US agreements.
- As a position paper, EDVEX is a conceptual blueprint with no implementation or pilot.
- Classification simplifies heterogeneous contracts and market conditions change rapidly.
Experimental setup
Datasets: 73 publicly disclosed data deals compiled from public sources
Metrics: count of deals by type, buyer, and payment category disclosed revenue volume
Builds on: Shapley values scaling laws data dignity
Closest papers by meaning
- Pricing and Competition for Generative AI (NeurIPS 2024 · similarity 0.84)
- Equilibrium of Data Markets with Externality (ICML 2024 · similarity 0.84)
- On the Existence and Complexity of Core-Stable Data Exchanges (NeurIPS 2025 · similarity 0.84)
- Fairshare Data Pricing via Data Valuation for Large Language Models (NeurIPS 2025 · similarity 0.83)
- An Instrumental Value for Data Production and its Application to Data Pricing (ICML 2025 · similarity 0.83)
- Decongestion by Representation: Learning to Improve Economic Welfare in Marketplaces (ICLR 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.