Title: Google Contract Workers Express Concerns Over Working Conditions for AI Chatbot Bard
In January of this year, reports emerged about Google’s concern over OpenAI’s ChatGPT, an AI chatbot that gained immense popularity soon after its launch in November 2022. As a response, Google introduced its own AI chatbot called Bard, which bears similarities to ChatGPT. However, it wasn’t until May of this year, after the Google I/O event, that Bard became widely accessible. The experiences of users with Google’s AI chatbot started surfacing on social media shortly thereafter. Recently, Bard received significant updates, indicating Google’s commitment to enhancing its capabilities as promised by CEO Sundar Pichai.
However, the contract workers responsible for training Bard have voiced their dissatisfaction with the working conditions. A group of contract workers approached Bloomberg to express their grievances, claiming to be overworked, underpaid, and stressed while reviewing Bard’s responses.
These contract workers play a crucial role in training generative AI tools like Bard, operating within the framework of Large Language Models (LLMs). Their primary responsibility is to provide feedback on the AI chatbot’s responses, identify biases, and flag any mistakes. Google has employed thousands of contract workers to handle this task and improve Bard’s performance. Unfortunately, the contract workers are facing challenges in their work environment.
According to Bloomberg’s report, six contract workers state that since Google entered the competitive race with OpenAI, the workload and complexity of their tasks have significantly increased. Without adequate training, these workers are expected to evaluate Bard’s answers on various subjects, ranging from medicine to law. The publication also uncovered documents containing convoluted instructions given to the workers. Moreover, the workers face extremely tight deadlines, sometimes as little as three minutes, to review Bard’s responses.
One of the contractors told the publication, As it stands right now, people are scared, stressed, underpaid, and don’t know what’s going on. They further added, And that culture of fear is not conducive to getting the quality and the teamwork that you want out of all of us.
In addition, a contract worker at Google warned Congress about the potential risks of Bard becoming a faulty and dangerous product due to the pressure on contractors to review content under tight deadlines.
The report disclosed that these contract workers are being paid as little as USD 14 per hour for their efforts.
A Google spokesperson responded to the concerns, stating that contract workers are just one aspect of the comprehensive testing process for Bard’s responses. They emphasized that the company employs multiple approaches to ensure the quality and accuracy of their AI products.
The spokesperson explained, Connecting people to high-quality information is core to our mission. We undertake extensive work to build our AI products responsibly, including rigorous testing, training, and feedback processes we’ve honed for years to emphasize factuality and reduce biases. They reassured that ratings from human evaluations do not directly impact the output of their models and that accuracy is promoted using various strategies.
Google’s statement aims to address the contract workers’ concerns while highlighting their commitment to testing and improving Bard’s performance. However, the challenges faced by these workers raise important questions about the treatment of those involved in training AI technologies and the potential impact on the final product.
As Google continues to enhance Bard’s capabilities, it is crucial for the company to ensure a supportive working environment for the contract workers involved. Striking a balance between innovation and fair treatment of workers will be key in maximizing the potential of AI chatbots for users worldwide.
