Amazon, Microsoft, and Meta are facing scrutiny from Democratic lawmakers regarding their use of “ghost work,” to drive the artificial intelligence market. This refers to the behind-the-scenes labour that involves tasks like data labelling and response rating, which play a crucial role in the advancement of artificial intelligence.
According to reports from Bloomberg, these invisible data workers are subjected to demanding tasks, constant surveillance, low wages, and a lack of benefits.
A group of lawmakers, led by Massachusetts Senator Ed Markey and Washington Representative Pramila Jayapal, expressed their concerns in a letter addressed to the CEOs of nine companies, including Amazon, Alphabet, Meta Platforms, Microsoft Corp, and International Business Machines Corp.
They pointed to the fact that these workers are responsible for screening potentially harmful chatbot responses, often without sufficient time to ensure safety. They also pointed out that inadequate training and supervision can introduce bias into AI systems.
“Data workers are often given scant training or supervision, which can result in the introduction of bias.”
In their letter, the lawmakers are seeking detailed information from the executives about their data workforces.
They want to know about policies related to breaks, procedures for appealing suspensions, and access to mental health resources for those exposed to distressing content. Their message is straightforward: tech companies must not exploit workers in their pursuit of AI innovation. This message is supported by Massachusetts Senator Elizabeth Warren and others.
In addition to well-established tech giants, the letter is also addressed to newer AI-focused companies like OpenAI Inc., Inflection AI, Scale AI Inc., and Anthropic.
Recent reports show that to develop AI products, U.S. companies rely heavily on subcontracted staff, often hired through external staffing agencies, both domestically and internationally. These workers typically do not receive the same benefits as the companies’ direct employees and are responsible for tasks such as content moderation and product quality assurance.
For generative AI tools, which generate responses to text prompts in various formats, thousands of contract workers are often employed to train, fix, and enhance algorithms. These algorithms are then presented to customers as technological marvels. Unfortunately, many of these workers report being underpaid, stressed, and overworked.
Some even suffer trauma from having to filter disturbing images. One such example was OpenAI paying workers in Kenya less than $2 per hour to filter harmful content from ChatGPT.
The lawmakers’ letter coincides with an AI summit that included executives from Tesla Inc., Meta, Microsoft, and Alphabet. Senate Majority Leader Chuck Schumer hosted the summit, although he did not sign the document. Senator Markey expressed his concern about tech companies’ unfair labour practices and called for accountability.