Technology

Employees accuse Meta of using AI in dismissals, but face a lack of evidence

Meta logo. Daniel Cole/Reuters An unprecedented lawsuit accusing Meta Platforms of using artificial intelligence tools in a discriminatory way to select employees for dismissal exposes the challenges faced by workers...

Share
Employees accuse Meta of using AI in dismissals, but face a lack of evidence
G1 Tecnologia

Meta logo. Daniel Cole/Reuters An unprecedented lawsuit accusing Meta Platforms of using artificial intelligence tools in a discriminatory way to select employees for dismissal exposes the challenges faced by workers trying to sue employers for using this technology, including the difficulty of proving how it was actually used. The case helps explain why the expected wave of labor lawsuits related to the use of artificial intelligence has not yet materialized. Experts say workers often know little about how AI systems are used in the workplace. Furthermore, many gave up the right to go to court by agreeing to resolve labor disputes through arbitration, a private procedure for resolving conflicts outside of court. In a decision released last week, in which he rejected Meta's request to complete the dismissals of 26 employees who filed a lawsuit, federal judge William pointed out a central obstacle for those alleging discrimination due to artificial intelligence: the workers did not participate in the discussions nor did they have access to the internal processes that led to the decisions.

This means that workers like Meta employees who claim they were selected for dismissal because of disabilities or because they took medical or family leave are often unable to gather enough evidence to demonstrate irregularities in the process. They face yet another obstacle: Like most American workers, they are bound by arbitration agreements, which means they cannot join a class-action lawsuit, present their case to a jury or press for a multimillion-dollar settlement in open court. Arbitration makes legal action difficult Companies generally prefer arbitration as they consider it a faster and cheaper alternative to courts. Workers' rights advocates say the model often favors employers and discourages employees from filing complaints. Additionally, the process is confidential, which can prevent unfavorable evidence discovered in a case from becoming known to other workers or the public. “Even if a certain system is proven to produce discriminatory results on a large scale, there is no way to share that information with other employees,” said Christine Webber, co-chair of the civil and employment rights practice at Cohen Milstein Sellers & Toll, which specializes in representing plaintiffs in lawsuits. Webber's office is not involved in the lawsuit against Meta. According to Webber and other lawyers who represent workers, these obstacles help explain why high-profile lawsuits involving the use of artificial intelligence by employers are still rare, even though the technology is increasingly present in the workplace. They claim that even the lawsuit filed against Meta, which seeks only injunctive relief, is an unusual case. One of the few cases that gained repercussion involves Workday, accused of using its human resources management software to illegally filter or exclude job candidates based on factors such as race, age and disability. In this case, arbitration is not a relevant factor because the company does not maintain contracts with candidates who apply for vacancies for its clients. Workday denies the allegations. Employees try to stop layoffs The contracts signed by Meta workers include a limited exception that allows them to request a court order to temporarily prevent one of the parties from taking measures capable of causing irreversible damage. However, this remedy is typically used in disputes involving alleged theft of trade secrets or grooming of customers and employees, rather than in cases of dismissal. Orrick denied the workers an injunction that would have prevented Meta from completing the layoffs. The judge has yet to decide whether to grant early relief, a broader provisional measure that could allow employees to return to their posts pending the conclusion of arbitration proceedings. According to him, this decision may be reviewed if the authors present evidence that demonstrates whether and how artificial intelligence was used inappropriately. A hearing is scheduled for August 24, and the losing party will be able to appeal Orrick's decision. Workers allege that, when defining which positions would be eliminated, Meta consulted artificial intelligence tools that monitored employees' productivity and use of AI resources. According to them, this system would have harmed people who needed to take time off work for health reasons or to care for family members. According to the process, Meta used several internal systems based on artificial intelligence. Among them were: "Metamate", a large language assistant; a tool described as a "second brain", powered by employees themselves and capable of tracking internal communications and documents; and a system that assigned productivity grades based on analysis of typing, content displayed on the screen, emails and browsing history. In court documents and statements presented last week, Meta stated that all decisions related to the almost 8 thousand layoffs announced at the beginning of the year were made by people, and not by automated systems. The company also denied using artificial intelligence to select employees for dismissal or to conduct performance reviews. A spokesperson told Reuters on Tuesday that the company would not comment on the case beyond what has already been presented to the Court. Orrick stated that at this stage in the process, he needed to consider the information presented by Meta, as the authors were unable to

Source: G1 Tecnologia

This story was originally published by G1 Tecnologia. Visit the original publication for further details.

Open original publication
More coverage

Related stories