This Tech Exec Quit His Job to Fight Generative AI's Original Sin

Ed Newton-Rex, a former AI executive turned critic, launched Fairly Trained to address the ethical crisis in generative AI by introducing a "fair trade" certification for training data. This initiative aims to provide transparency and decision-making tools, encouraging companies to source data through explicit licensing, public domain status, or open licenses. By validating these practices, the nonprofit seeks to establish a standard for ethical data sourcing that respects creators’ rights, contrasting sharply with the industry’s prevalent reliance on unauthorized scraping. The program argues that building compliant AI models is both feasible and necessary, challenging dominant industry narratives that claim unlicensed data is indispensable. Early adopters, including music and image generation platforms, demonstrate that operating with proper licenses is achievable. Backed by significant cultural and publishing stakeholders, this movement positions ethical data usage not as a technical limitation but as a mandatory business standard, offering a sustainable alternative to the current model of mass data extraction. This effort is highly relevant to open data because it establishes a framework for verifying provenance and rights management within AI datasets. It highlights the growing intersection between open licensing and commercial AI development, suggesting that open data principles can coexist with corporate models if properly certified. The initiative encourages a shift toward greater transparency and accountability in how data is collected and used, promoting an ecosystem where open and licensed resources are explicitly tracked and respected.

Source: wired.com
Published on 2024-01-18

refik anadol's nature-focused open source AI model debuts at world economic forum »

Open Data

Related news