On January 24, 2024 Google announced the immediate termination of its contract with AI‑data specialist Appen Ltd. The move, framed as part of an efficiency review of supplier partnerships, ends a multi‑year relationship that supplied large‑scale data annotation and quality‑rating services for Bard, Google Search and other generative‑AI tools.
Why Google Ended Its Partnership with Appen
Google cited an ongoing effort to evaluate and adjust supplier relationships to ensure optimal vendor operations. The company emphasized a smooth transition for existing projects while seeking more efficient or internal solutions for data‑labeling needs.
What Appen Provided for Google’s AI Products
Appen recruited and managed a global crowd of contractors who performed essential AI‑training tasks, including:
- Rating search result relevance
- Labeling images and multimedia content
- Assessing the quality of model outputs for large language models
These activities formed the backbone of the data‑annotation pipeline that refined Bard, Search and other generative‑AI services.
Impact on Bard and Google’s AI Roadmap
The loss of Appen’s annotation capacity could slow iteration cycles for Bard and related AI products unless Google has already built alternative mechanisms. The company has not disclosed a replacement vendor, suggesting a possible shift toward internal data‑labeling or synthetic‑data techniques.
Financial Implications for Appen
Google accounted for roughly 30 % of Appen’s 2023 revenue, contributing about US$82.8 million. The contract termination removes a major income stream, pressuring Appen to diversify its client base and accelerate its own AI initiatives.
Regulatory and Market Context
The abrupt termination, coupled with Appen’s claim of no prior notice, may attract scrutiny from regulators monitoring fair contracting practices in the AI supply chain. While no formal complaints have been filed, the change has been disclosed to shareholders through an ASX filing.
Future Outlook for Google’s Data Annotation Strategy
Google has not announced a timeline for re‑establishing a data‑annotation partnership. The emphasis on “efficiency” indicates a strategic shift that could involve greater internalization of labeling functions or the adoption of emerging synthetic‑data solutions, reshaping the AI‑data ecosystem.
