Bard workers are under time pressure to review AI answers, reports Bloomberg
Bloomberg’s report reveals that Google’s AI chatbot, Bard, relies on underpaid contractors who face strict time constraints and minimal training, as Google races to keep up with OpenAI’s ChatGPT and maintain its position in the AI market.
Bloomberg reports that Google’s AI chatbot, Bard, depends on a large workforce of contractors who face the expectation of reviewing its generated responses within extremely tight deadlines, as little as a three-minute timeframe. These contractors, working for companies like Appen and Accenture, receive minimal training and earn as little as $14 per hour.
Google developed Bard in reaction to the success of OpenAI’s ChatGPT, which presented a competitive challenge to Google’s search business. While both Bard and ChatGPT utilise large language models, human reviewers play a crucial role in ensuring the reliability and accuracy of the responses. The contractors’ tasks often involve rating responses based on their helpfulness and up-to-dateness. However, the workload for these reviewers has become increasingly demanding and complex.
Google stated that it prioritises building AI products responsibly through rigorous testing, training, and feedback processes, including human evaluation. Google’s spokesperson emphasised the involvement of specialised teams across the company in continuously enhancing the quality and accuracy of its AI products.