As the influence of artificial intelligence (AI) extends through every part of music organizations, one area that has begun to see the influence of AI is in the realm of human resource management (HRM). Today, most organizations utilize some form of electronic HRM process, whether it is to review candidates for a position or to assess staff performance for promotion or termination. Many organizations justify the use of AI in HRM as part of cost saving measures, diminishment of human bias and error, and compliance with labor laws. However, like their human counterparts, the notion that AI systems are unbiased, especially in the realm of HRM, is now widely regarded as a fallacy. With this in mind, many federal and state agencies in the United States are creating legislation that manages AI or automated decision-making Systems (ADMS).
California is leading the way in developing legislation that addresses the use of ADMS in the hiring and promoting of employees. The “No Bosses Act,” as it has been euphemistically termed, is designed to regulate the use of AI without human oversight. Senate Bill 7 (SB 7) prohibits an employer from “discharging, threatening to discharge, demoting, suspending, or in any manner discriminating or retaliating against any worker for taking certain actions asserting their rights under the bill”. This legislation, if enacted, will allow a public prosecutor or employee to file a civil action against an employer who has utilized ADMS in making an HRM decision that violates current Californian labor law. Under current law, the Labor and Workforce Development Agency is responsible for ensuring the protection of workers’ rights and interests in California. This Act requires music organizations using ADMS in hiring decisions to provide written notice to applicants and employees of its use. It defines an ADMS as any computational process that uses machine learning, statistical modeling, data analytics, or artificial intelligence to create an outcome. These include classifications and recommendations that impact employees. Furthermore, the Act guarantees that an employee can appeal a decision made by ADMS and can identify errors in the inputs or output data used by ADMS. Finally, the ACT allows employees to request access to inputs or outputs created by the ADMS in an evaluation. The legislation also requires organizations to guarantee the privacy of this data; an issue not fully covered in this proposed legislation. However, the use of AI in determining hiring and work rights is a serious issue facing our digitally driven society. To comply with this new legislation, employers must ensure that organizations adhere to the legislation by requiring ADMS decisions to have meaningful human oversight. The legislation also requires employers to be transparent with employees regarding the use of ADMS, including the use of data to formulate reports on employees, including credit history, health records, immigration status, and other state and federally protected criteria. Achieving these goals requires organizations to train employees on AI, as well as integrate protocols into their bylaws.
ADMS are autonomous systems that utilize AI to perform a range of tasks. Financial reasons, as well as the efficiencies of time and accuracy, often drive their application in the hiring process. A single ADMS can review data within milliseconds, compared to human endeavors. One of the most controversial aspects of using ADMS is its application in decision-making processes. These systems use artificial autonomous agents (AAs) in place of humans. AA software responds to environments based on direct instructions from a human or utilize machine learning protocols autonomously of any human interaction. Transferring control of the decision-making process can be done at various levels based on the accountability associated with the task. However, fully automated systems have the potential to introduce bias into the sample, failing to understand subtleties in the task, or in extreme cases act unethically due to a programmer’s bias. In these scenarios, the ADMS implements autonomous decisions without any human intervention and can create serious issues, as a human agent has no control over the decisions made by an independently acting ADMS. Apart from the financial benefits of using this system, fully automated decision-making enables organizations to distance themselves from the decision and, as such, cannot be held liable for damages caused by it. Researchers of AI and ethical decision-making have highlighted several issues with fully autonomous ADMS, including a lack of transparency in the decision-making process, manipulation of behavior (e.g., through the decision options), bias, and discrimination in based on gender, race, ethnicity, and other demographic characteristics of applicants. Although these issues may be apparent in a fully automated systems, human-led or even human-AI systems may also introduce bias that hinders the unbiased management of human resources.
Given that we are on the precipice of an AI revolution, the California SB7 legislation is a welcome addition to the principles governing the use of AI systems in the workplace. To ensure that systems put into place not only adhere to labor standards, they must comply with human standards and ethics. Factors such as empathy, holistic thinking, and equity are central to human endeavors and should take precedence over financial and competitive gain.
The Analytical Musician 