Hiring and promotion using AI: Prepare for the legal (r)evolution

The use of artificial intelligence (AI) has the potential to revolutionize the way companies recruit and select candidates for job openings and promotions. However, as with any new technology, there are concerns about how it may be used and the potential for unintended consequences.

AI contributes to the recruitment and selection of candidates for hire or promotion in several ways:

• Screening resumes: AI-powered tools can automatically scan resumes and identify candidates who match the qualifications and skills required for a given job. Natural language processing (NLP) algorithms can analyze resumes and cover letters to identify relevant skills and experience, making it easier for recruiters to identify qualified candidates and save time on manual resume screening. This can also help to reduce the chances of unconscious bias, as the AI algorithms process resumes based on pre-defined criteria, rather than human prejudices.
• Answering questions: Chatbots and other virtual assistants can be used to schedule interviews and answer candidate questions. This can help to streamline the hiring process, making it more convenient for candidates and reducing the workload for recruiters.
• Interviewing: AI-powered interviewing tools can conduct initial interviews with candidates, using natural language processing to understand their answers and identify key qualifications and skills. AI facial recognition and voice analysis can be used to assess a candidate’s non-verbal communication and tone of voice. This can provide valuable insights into a candidate’s personality and communication style, which can be especially useful for roles that require strong communication skills.
• Predictive analysis: AI-powered tools can analyze data from previous hires and identify patterns that are indicative of success in a given role, helping to identify the most suitable candidates for a job. AI can also be used to assess which skills a candidate has and which they need, furthering opportunities to upskill and reskill candidates who may previously have been overlooked.
• Fairness and bias mitigation: AI-powered tools can help to reduce bias in the hiring process by eliminating subjective factors and focusing on objective criteria when evaluating candidates.

While the use of AI in hiring has the potential to make the process more efficient and objective, it is important to take steps to mitigate the risk that it does not perpetuate existing inequities or lead to discrimination against certain groups of people. Proposed and recently enacted laws and regulations aim to ensure that the use of AI is transparent and that job candidates are treated fairly and have the right to appeal decisions made by AI systems.

EEOC guidance

The U.S. Equal Employment Opportunity Commission (EEOC) recently identified discrimination caused by AI tools as a key enforcement priority as part of its proposed 2023 Strategic Enforcement Plan. The EEOC plans to focus on instances where the use of technology contributes to discrimination against individuals on the basis of their race, ethnicity, age, sex, disability, or other protected classification. The EEOC specifically noted risks found in the “use of automated recruitment, selection, or production and performance management tools; or other existing or emerging technological tools used in employment decisions.”

Discrimination can take the form of intentional discrimination, such as the online tutoring company sued by the EEOC, which allegedly programmed its online software to automatically reject female applicants age 55 or older and male applicants age 60 or older. Or discrimination can be inadvertent. The EEOC recently published guidance regarding the risks of disability discrimination in this context, warning employers to have safeguards in place to prevent such outcomes.

Guideposts from New York City’s new law

In the first U.S. law of its kind, New York City Local Law 144 of 2021 regulates the use of “automated employment decision tools” in the screening of candidates for hire or promotion within New York City. Enforcement of the law starts on April 15, 2023. While the currently proposed interpretive rules add some clarity, it is possible more changes will occur before the law’s effective date.

New York City’s law is likely to provide a roadmap to similar laws across the U.S. Key components of the law include:

• Covered AI broadly defined: “Automated employment decision tool” is defined as “any computational process, derived from machine learning, statistical modeling, data analysis, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making.” While the current proposed rules seek to further refine this definition, public comments focused heavily on the breadth of this definition, and it is possible additional changes may occur.
• Bias audit: An “independent auditor” must conduct an “impartial evaluation” of any automated decision tool to assess if its use has resulted in a disparate impact upon race, ethnicity, or sex, with a summary of results made “clearly and conspicuously” available on the employment section of the employer’s website. The audit must have occurred no more than a year prior to the date the tool is used. The current proposed rules detail the specifics of the audit, including more sophisticated statistical testing to address intersectionality (the proposed rules provide the following example: “impact ratio for selection of Hispanic or Latino male candidates vs. Not Hispanic or Latino Black or African American female candidates”).
• Notices: All New York City job candidates (external or internal) who will be screened for hire or promotion by an automated decision tool must be notified a minimum of ten days prior to its use. The notice must inform the candidate of the qualifications and characteristics the tool will use in its assessment, describe the applicable data retention policy, and offer the candidate an “alternative selection process or accommodation.” Notices must also be posted on the employment section of the employer’s website.
• Penalties: Violations are subject to a civil penalty of up to $500 for a first violation, with each additional violation subject to a penalty of $500-$1,500. Each failure to meet notice requirements constitutes a separate violation. Failure to meet the “bias audit” requirements results in separate, daily violations. Individual candidates can bring suit to enforce the law, in addition to potential government enforcement via New York City’s corporation counsel or the New York City Division of Human Rights.

Other legal trends

Regulation of AI continues to evolve, with the list of pending and proposed laws rapidly expanding (for example, the National Conference of State Legislatures tracks developments here).

Additional notable legal developments include:

• California’s Privacy Rights Act, effective Jan. 1, 2023, grants consumers and employees opt-out rights with respect to the use of “automated decision-making technology” and directs the California Privacy Protection Agency to issue regulations “governing access and opt-out rights with respect to businesses’ use of automated decision-making technology.” The California Civil Rights Department (formerly known as the Department of Fair Employment and Housing) proposed amendments to its employment regulations to require notices regarding the use of AI and to hold employers liable if they are unable to defend AI-assisted exclusion criteria.
• Illinois enacted the Artificial Intelligence Video Interview Act in 2020, which requires employers who use AI to analyze applicant videos to undertake certain actions with respect to notice, consent, confidentiality, and data destruction. Following a 2022 amendment, employers who rely solely on analysis of video interviews to determine whether an applicant will be selected for an in-person interview must collect and report additional demographic data.
• The federal Algorithmic Accountability Act, which was first introduced in the U.S. Congress in 2019, was re-introduced in 2022. The bill would require companies to conduct regular audits of their AI systems to ensure that they are not causing unintended harm, such as discrimination against certain groups of people. This includes ensuring that the data used to train the AI systems is not biased and that the algorithms used in the hiring process do not perpetuate existing inequalities.
• Some non-U.S. countries have already enacted regulations to limit the use of AI in hiring. For example, in the European Union, the General Data Protection Regulation (GDPR) requires companies to obtain consent from job applicants before using their personal data in the hiring process, which includes data used to train AI systems.

Practical steps for every employer

The practice of using AI to assist in critical decision-making is already ubiquitous and poised to expand as these tools become more readily available and less expensive to deploy. Companies embracing AI may obtain a significant competitive advantage over their peers, and in the near future, AI may become a necessary part of every employer’s hiring process to varying degrees.

Read more: Are AI hiring tools adding risks to your business?

Practical steps for every employer now include:

• Stay in tune with emerging trends for the use of AI within the company’s specific industry.
• Assess what software, algorithms, and AI are currently being used in the company’s hiring and other employment-related processes.
• Many vendors offer employment-related AI tools. Involve human resources and legal advisors early in the decision-making, including the selection process and the calibration of tools to the employer’s specific hiring criteria.
• Thoughtfully tailor automated screening criteria and be prepared to answer questions as to why the criteria was selected.
• Communicate openly with candidates about how the company uses AI in its screening and selection process.
• Periodically review statistical outcomes from AI-assisted employment practices, even if not legally required, to determine whether even inadvertent discrimination has occurred.
• Consider how the AI will accommodate disabled candidates and prepare to offer alternative reasonable accommodations, following the EEOC’s Guidance.

This publication is co-authored by law firm attorneys and artificial intelligence (AI). The human authors generated this article in part with GPT-3, OpenAI’s large-scale language-generation model. Upon generating draft language, the human authors supplemented it with additional legal trends and analysis and reviewed, edited, and revised the AI’s language to their own liking. The human authors take ultimate responsibility for the content of this publication.

Nicole Stover & Mischa Wheat, with contributions from GPT-3.