Google Ends AI Data Firm Contract Integral to Bard's Training

Google has terminated its contract with Appen, an Australian data company crucial for training its large language model AI tools, including those used in Bard, Search, and other products. This decision comes amid increasing competition in the development of generative AI tools. Google spokesperson Courtenay Mencini stated that the move was part of ongoing efforts to evaluate and optimize supplier partnerships across Alphabet for enhanced efficiency.

Also Read: Elon Musk Refutes Claims of His AI Company Securing $500 Million, Targets $1 Billion Funding Goal

Appen, unaware of Google’s decision, informed the Australian Securities Exchange through a filing. Human workers, often handling less desirable aspects of AI training, are the often-overlooked backbone of the industry. Appen’s contractors play a role in rating data quality and AI model responses. The company faced scrutiny when some Alphabet Workers Union members petitioned for higher wages, reaching $15 from $10 per hour. Despite winning wage increases, layoffs occurred, citing business conditions.

According to CNBC, Appen has contributed to training AI models for Microsoft, Meta, and Amazon. The termination of the Google contract significantly impacted Appen’s revenue, with $82.8 million generated from Google alone in the fiscal year 2023. Appen reported total revenue of $273 million last year. Google is collaborating with Appen to ensure a smooth transition, according to Mencini.

The workforce challenges related to AI training extend beyond Google, with Accenture employees voting to join the Alphabet Workers Union. This union represents Google contractors and gained support after workers refused to handle objectionable prompts for the Bard chatbot in November. The broader issue of low wages for content moderators in Kenya came to light when Sama, a data-labeling firm, faced a lawsuit from its employees against Meta for paying $2.20 an hour for viewing disturbing content.