Ratifi Español Sign in
Ratifi

Federal bills from Congress.gov. State bills from Open States.

A 3855Session 221

What changed between versions

99 paragraphs changed+728 words added−1552 removed

Highlighted text was added in the newer version. Struck-through text was removed.

ASSEMBLY COMMITTEE SUBSTITUTE FOR

ASSEMBLY, No. 3855

STATE OF NEW JERSEY

221st LEGISLATURE

INTRODUCEDADOPTED FEBRUARYMAY 2216, 2024

Sponsored by:

Assemblyman HERBBALVIR CONAWAY, JR.SINGH

District 7 (Burlington)

SYNOPSIS

Creates standards for independent bias auditing of automated employment decision tools.

CURRENT VERSION OF TEXT

As introduced.

Substitute as adopted by the Assembly Science, Innovation and Technology Committee.

An Act concerning automated employment decision tools and supplementing Title 34 of the Revised Statutes.

Be It Enacted by the Senate and General Assembly of the State of New Jersey:

1. As used in this act:

“Automated employment decision tool” or “AEDT” means any system the function of which is governed by statistical theory, or systems the parameters of which are defined by systems, including inferential methodologies, linear regression, neural networks, decision trees, random forests, and other learning algorithms, which automatically filter candidates or prospective candidates for hire or for any term, condition, or privilege of employment in a way that establishes a preferred candidate or candidates.

“Bias audit” means an impartial evaluation, including but not limited to testing, of an automated employment decision tool to assess its predicted compliance with the provisions of the “Law Against Discrimination,” P.L.1945, c.169 (C.10:5-1 et seq.), and any other applicable law relating to discrimination in employment.

1. As used in P.L. , c. (C. ) (pending before the Legislature as this bill):

“Automated employment decision tool” means a machine-based system that can, for a set of human-defined objectives provided by an employer or an individual acting on behalf of an employer, make predictions, recommendations, or decisions influencing recruitment, workforce, or employment decisions.

“Bias audit” means an impartial evaluation conducted by an independent auditor, including but not limited to:

a. rigorous assessment of an automated employment decision tool to determine its impact on persons of any category;

b. identification and documentation of any biases, risks, or potential discriminatory outcomes that arise from the automated employment decision tool’s design, implementation, or use; and

c. clear actionable recommendations to avoid, manage, or mitigate, identified biases and risks, and to ensure the sage, secure and trustworthy use of the automated employment decision tool in employment decisions.

“Candidate for employment” means a person who has applied for a specific employment position by submitting the necessary information or items in the format required by the employer or employment agency.

“Category” means any component 1 category required to be reported by employers pursuant to subsection (c) of 42 U.S.C. s.2000e-8 as specified in 29 C.F.R. s.1602.7, as designated on the Equal Employment Opportunity Commission Employer Information Report EEO-1.

“Distribution date” means the date the employer or employment agency began using a specific AEDT.

“Category” means race, color, national origin, ethnicity, sex, gender identity, sexual orientation, age, religion, marital or familial status, disability, and deriving income from any public assistance program.

“Covered individual” means a candidate for employment or current employee being assessed by an automated employment decision tool to make an employment decision.

"Employer" includes any individual, partnership, association, corporation, and the State and any county, municipality, or school district in the State, or any agency, authority, department, bureau, or instrumentality thereof, employing any person.

“Employment decision” means to screen a candidate for employment or otherwise to help decide compensation or any other terms, conditions, or privileges of employment.

“Employment agency” means the same as that term is defined in section 1 of P.L.1989, c.331 (C.34:8-43).

"Employer" includes any individual, partnership, association, corporation, and the State and any county, municipality, or school district in the State, or any agency, authority, department, bureau, or instrumentality thereof, employing any person.

“Historical data” means data collected during an employer or employment agency's use of an AEDT to assess candidates for employment or employees for promotion.

“Impact ratio” means:

a. the ratio of the protected class that receives a favorable outcome and the proportion of the control class that receives a favorable outcome when the decision being made is binary, including but not limited to the decision to hire or not and the decision to promote or not; or

b. the ratio of the difference between the average protected class outcome and the average control class outcome to a measure of the standard deviation of the outcome across the overall population when the decision being made is not binary, including but not limited to the decision to increase base salary or compensation of an employee.

“Independent auditor” means a person or group that is capable of exercising objective and impartial judgment on all issues within the scope of a bias audit of an AEDTautomated employment decision tool. An auditor shall not be considered independent if the auditor:

a. Is or was involved in using, developing, or distributing the AEDTautomated employment decision tool;

b. At any point during the bias audit, has an employment relationship with an employer or employment agency that seeks to use or continue to use the AEDTautomated employment decision tool or with a vendor that developed or distributes the AEDTautomated employment decision tool; or

c. At any point during the bias audit, has a direct financial interest or a material indirect financial interest in an employer or employment agency that seeks to use or continue to use the AEDTautomated employment decision tool or in a vendor that developed or distributed the AEDTautomated employment decision tool.

“Impact ratio” means either the selection rate for a category divided by the selection rate of the most selected category; or the scoring rate for a category divided by the scoring rate for the highest scoring category.

“Machine learning, statistical modeling, data analytics, or artificial intelligence” means a group of rule-based, mathematical, computer-basedor computation techniques:

a. Thatthat generate a prediction, prescription, recommendation, or decision, meaning an expected outcome for an observation, such as an assessment of a candidate'scovered individual’s fit or likelihood of success, or that generate a classification, meaning an assignment of an observation to a group, such as categorizations based on skill sets or aptitude; and

b. Forfor which a computer program implementing the mathematical or computational technique, at least in part identifies the inputs, the relative importance placed on those inputs, and, if applicable, other parameters for the models in order to improve the accuracy of the predictiontask orperformed classificationby the technique.

“Scoring Raterate” means the rate at which individuals in a category receive a score above the sample's median score, where the score has been calculated by an AEDTautomated employment decision tool.

“Screen” means to make a favorable or unfavorable determination about whether a candidate being considered for employment or employee being considered for promotion, termination, or performance review should be selected or advanced in the hiring or promotion process.

“Selection rate” means the rate at which favorable or adverse reactions are taken regarding individuals in a category are either selected to move forward in the hiringemployment decision process or assigned a classification by an AEDTautomated employment decision tool. This rate may be calculated by dividing the number of individuals with favorable or unfavorable outcomes in the category moving forward or assigned a classification by the total number of individuals in the category who applied for a position or were considered for promotion.

“Simplified output” means a prediction or classification as specified in the definition for “machine learning, statistical modelling, data analytics, or artificial intelligence.” A simplified output may take the form of a:

a. Score , such as rating a candidate's estimated technical skills;

b. Tag or categorization, such as categorizing a candidate's resume based on key words, assigning a skill or trait to a candidate;

c. Recommendation, such as whether a candidate should be given an interview; or

d. Ranking, such as arranging a list of candidates based on how well their cover letters match the job description.

“Simplified output” does not refer to the output from analytical tools that translate or transcribe existing text, such as converting a resume from a PDF or transcribing a video or audio interview.

“Test data” means data used to conduct a bias audit that is not historicaltraining data.

“Training data” means data used in an employer or employment agency's use of an automated employment decision tool to assess candidates for employment, termination, compensation changes, performance improvement, or employees for promotion.

2. a. An employer or employment agency mayshall not use or continue to use an AEDTautomated employment decision tool if more than one year has passed since the most recent bias audit of the AEDTautomated employment decision tool.

b. Where an AEDT selects candidates for employment or employees being considered for promotion to move forward in the hiring process or classifies them into groups, a bias audit shall, at a minimum:

b. A bias audit shall, at a minimum:

(1) calculate the selection rate for each category or the scoring rate if the outcome is continuous for each category;

(2) calculate the impact ratio for each category;

(3) ensure that the calculations required in paragraphs (1) and (2) of this subsection separately calculate the impact of the AEDTautomated employment decision tool on:

(a) sex categories, such as the impact ratio for selection of male candidates vsversus female candidates;

(b) race, color, national origin, and ethnicity categories, such as the impact ratio for selection of Hispanic or Latino candidates versus Black or African American Non-Hispanic or Non-Latino candidates; and

(c) age categories, such as the impact ratio for selection of older covered individuals versus younger covered individuals;

(d) marital or familial status categories, such as the impact ratio for selection of married covered individuals versus unmarried covered individuals;

(e) disability categories, such as the impact ratio for selection of covered individuals with disabilities versus covered individuals without disabilities;

(f) religion categories; such as the impact ratio for selection of Hindu covered individuals versus Buddhist covered individuals;

(g) sexual orientation categories, such as the impact ratio for selection of heterosexual covered individuals versus covered individuals of other sexual orientations;

(h) gender identity categories, such as the impact ratio for selection of cisgender covered individuals versus transgender covered individuals;

(i) income source categories, such as the impact ratio for selection of covered individuals whose income is derived from any public assistance program versus covered individuals whose income is not derived from any public assistance program; and

(cj) intersectional categories of sex, ethnicity, and race, such as the impact ratio for selection of Hispanic or Latino male candidates versus Non-Hispanic or Non-Latino Black or African American female candidates;

(4) ensure that the calculations in paragraphs (1), (2), and (3) of this subdivisionsubsection are performed for each group, if an AEDT classifies candidates for employment or employees being considered for promotion into specified groups, such as leadership styles; and

(5) indicate the number of individuals the AEDTautomated employment decision tool assessed that are not included in the required calculations because they fall within an unknown category.

c. Where an AEDT scores candidates for employment or employees being considered for promotion, a bias audit shall, at a minimum:

(1) calculate the median score for the full sample of applicants;

(2) calculate the scoring rate for individuals in each category;

(3) calculate the impact ratio for each category;

(4) ensure that the calculations required in paragraphs (1), (2), and (3) of this subsection separately calculate the impact of the AEDT on:

(a) sex categories, meaning the impact ratio for selection of male candidates vs female candidates;

(b) race or ethnicity categories, meaning the impact ratio for selection of Hispanic or Latino candidates versus Black or African American Non-Hispanic or Non-Latino candidates; and

(c) intersectional categories of sex, ethnicity, and race, such as the impact ratio for selection of Hispanic or Latino male candidates versus Non-Hispanic or Non-Latino Black or African American female candidates); and

(5) indicate the number of individuals the AEDT assessed that are not included in the required calculations because they fall within an unknown category.

dc. Notwithstanding the requirements of paragraphs (2) and (3) of subsection b. and paragraphs (3) and (4) of subsectionthis c.section, an independent auditor may exclude a category that represents less than two percent of the data being used for the bias audit from the required calculations for impact ratio. Where such a category is excluded, the summary of results shall include the independent auditor's justification for the exclusion, as well as the number of applicants and scoring rate or selection rate for the excluded category.

3. a. Historical Data. A bias audit conducted pursuant to section 2 of P.L. , c. (C. ) (pending before the Legislature as this actbill) shall use historicaltraining data of the AEDTautomated employment decision tool. The historicaltraining data used to conduct a bias audit may be from one or more employers or employment agencies that use the AEDTautomated employment decision tool. However, an individual employer or employment agency may rely on a bias audit of an AEDTautomated employment decision tool that uses the historicaltraining data of other employers or employment agencies only in the following circumstances:

(1) if that employer or employment agency provided historicaltraining data from its own use of the AEDTautomated employment decision tool to the independent auditor conducting the bias audit; or

(2) if that employer or employment agency has never used the AEDTautomated employment decision tool.

b. Test Data. Notwithstanding the requirements of subsection a. of this section, an employer or employment agency may rely on a bias audit that uses test data if insufficient historicaltraining data is available to conduct a statistically significant bias audit. If a bias audit uses test data, the summary of results of the bias audit shall explain why historicaltraining data was not used and describe how the test data used was generated and obtained.

4. a. Before the use of an AEDT, an employer or employment agency in the State shall make the following publicly available on the employment section of its website in a clear and conspicuous manner:

(1) the date of the most recent bias audit of the AEDT and a summary of the results, which shall include the source and explanation of the data used to conduct the bias audit, the number of individuals the AEDT assessed that fall within an unknown category, and the number of applicants or candidates, the selection or scoring rates, as applicable, and the impact ratios for all categories; and

(2) the distribution date of the AEDT.

b. The requirements of subsection a. of this section may be met with an active hyperlink to a website containing the required summary of results and distribution date, provided that the link is clearly identified as a link to results of the bias audit.

c. An employer or employment agency shall keep the summary of results and distribution date posted for at least six months after its latest use of the AEDT for an employment decision.

5. a. An employer that uses an AEDT for an employment decision shall provide notice to the candidates or applicants of the use of an AEDT that includes instructions for how a person may request an alternative selection process or a reasonable accommodation under other laws, if applicable. Nothing in this act requires an employer or employment agency to provide an alternative selection process.

b. To comply with subsection a. of this section, an employer or employment agency in the State may provide notice to a candidate for employment by doing any of the following:

(1) provide notice on the employment section of its website in a clear and conspicuous manner at least 10 business days before use of an AEDT;

(2) provide notice in a job posting at least 10 business days before use of an AEDT; or,

(3) provide notice to candidates for employment via U.S. mail or e-mail at least 10 business days before use of an AEDT.

c. To comply with the provisions of this act, an employer or employment agency may provide notice to an employee being considered for promotion who resides in the State by doing any of the following:

(1) provide notice in a written policy or procedure that is provided to employees at least 10 business days before use of an AEDT;

(2) provide notice in a job posting at least 10 business days before use of an AEDT; or,

(3) provide notice via U.S. mail or e-mail at least 10 business days before use of an AEDT.

d. To comply with the provisions of this act, an employer or employment agency shall:

(1) provide information on the employment section of its website in a clear and conspicuous manner about its AEDT data retention policy, the type of data collected for the AEDT, and the source of the data;

(2) post instructions on the employment section of its website in a clear and conspicuous manner for how to make a written request for such information, and if a written request is received, provide such information within 30 days; and

(3) provide an explanation to a candidate for employment or employee being considered for promotion why disclosure of the information would violate local, State, or federal law, or interfere with a law enforcement investigation.

64. This act shall take effect on the first day of the seventh month next following the date of enactment, except that the commissionerCommissioner of Labor and Workforce Development may take any anticipatory administrative action in advance as shall be necessary for the implementation of P.L. , c. (C. ) (pending before the Legislature as this actbill).

STATEMENT

This bill provides standards for the use of an independent bias audit if an employer elects to use an automated employment decision tool (AEDT) for an employment decision. The bill does not require an employer to use an AEDT in making an employment decision, but it does require that an employer utilize an independent bias auditing system of its AEDT if the employer elects to use an AEDT.

The bill defines “automated employment decision tool,” or “AEDT” to mean any system the function of which is governed by statistical theory, or systems the parameters of which are defined by systems, including inferential methodologies, linear regression, neural networks, decision trees, random forests, and other learning algorithms, which automatically filter candidates or prospective candidates for hire or for any term, condition, or privilege of employment in a way that establishes a preferred candidate or candidates.

The bill defines “bias audit” to mean an impartial evaluation, including but not limited to testing, of an automated employment decision tool to assess its predicted compliance with the provisions of the “Law Against Discrimination,” and any other applicable law relating to discrimination in employment.

The bill requires employers to post the result of bias audits on employer websites, and it requires employers to provide notice to employment candidates or applicants of the use of an AEDT in employment decisions.

This bill is modeled after regulations in the New York City Administrative Code addressing the use of AEDTs and bias auditing of those systems.

Back to the full text