{"id":8304,"date":"2021-04-21T08:00:00","date_gmt":"2021-04-21T08:00:00","guid":{"rendered":"http:\/\/ncjolt.org\/?p=8304"},"modified":"2021-04-24T02:34:32","modified_gmt":"2021-04-24T02:34:32","slug":"could-ai-technologies-for-worker-health-and-safety-be-discriminatory","status":"publish","type":"post","link":"https:\/\/journals.law.unc.edu\/ncjolt\/blogs\/could-ai-technologies-for-worker-health-and-safety-be-discriminatory\/","title":{"rendered":"Could AI Technologies for Worker Health and Safety Be Discriminatory?"},"content":{"rendered":"\n

As we near the end of the COVID-19 pandemic, employers are grappling with how to transition employees back into the office while ensuring workplace safety. Many employers are relying on artificial intelligence (\u201cAI\u201d) and other sophisticated technologies to implement screening measures (e.g., body temperature checks) and other public health-recommended workplace protocols (e.g., social distancing) per the Centers for Disease Control and Prevention\u2019s (\u201cCDC\u201d) COVID-19 employer guidelines<\/a>. These AI technologies could potentially enable businesses to return to pre-pandemic normalcy.<\/p>\n\n\n\n

However, since the use of AI technologies have been shown to return discriminatory results in the workplace<\/a>, there remains the danger that AI technologies employed for workplace safety could (inadvertently) compromise the ideal of equal employment opportunity for all. For example, an employer might set a rigid threshold for the use of automated thermometers: \u201cNo employee with a temperature above 100.0 degrees Fahrenheit may work in the office.\u201d But, what if an employee is naturally predisposed to having a higher body temperature, and thus \u201cfails\u201d the screening protocol more often than not? Although fever (and the accompanying high body temperature) can be a symptom of COVID-19 infection, there are also various other identified conditions<\/a> that could skew temperature reading results.<\/p>\n\n\n\n

Being compelled to work from home because an employer implements workplace safety protocols using AI technology, with metrics that fail to consider unique and individual circumstances, may have a disproportionately negative impact on certain populations.<\/p><\/blockquote>\n\n\n\n

One alternative for those who fail their employer\u2019s screening protocol is simply to \u201cwork from home,\u201d but this alternative could be disadvantageous for certain employees. For instance, employees with children may find it difficult to work from home if their children are also in the house. Other employees may have multiple family members at home who simultaneously need to use the family\u2019s internet connection, thereby reducing its speed. Whatever the reason, being compelled to work from home because an employer implements workplace safety protocols using AI technology, with metrics that fail to consider unique and individual circumstances, may have a disproportionately negative impact on certain populations.<\/p>\n\n\n\n

Typically, the Americans with Disabilities Act (\u201cADA\u201d)\u2014which attempts to mitigate workplace discrimination\u2014could have served as a safeguard for employees in the above circumstance, as taking an employee\u2019s temperature reading could constitute a prohibited \u201cmedical examination\u201d according to the Equal Employment Opportunity Commission\u2019s (\u201cEEOC\u201d) guidelines<\/a>. However, there are two notable exceptions under the ADA where an employer can engage in a \u201cmedical examination\u201d as defined by the EEOC: “(1) the employer can show that the inquiry or exam is job-related and consistent with business necessity, or (2) the employer has a reasonable belief that the employee poses a \u201cdirect threat\u201d to the health or safety of the individual or others that cannot otherwise be eliminated or reduced by reasonable accommodation.”<\/p>\n\n\n\n

Moreover, per the EEOC\u2019s guidance on pandemic preparedness in t<\/a>he workplace, a \u201cdirect threat is an important ADA concept during an influenza pandemic,\u201d and COVID-19 \u201cmeets the direct threat standard.\u201d Thus, this principle permits an employer to send an employee home with potential COVID-19 symptoms, such as a temperature reading above 100.0 degrees Fahrenheit.<\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n

Notably, the EEOC has clearly acknowledged<\/a> that the \u201cnon-pandemic\u201d ADA rules still apply, thereby mandating that employers administer temperature checks and other public health-recommended procedures \u201cin a uniform and nondiscriminatory manner, and keep all information about employees obtained through disability-related inquiries or medical examinations confidential.\u201d Meaning, so long as employers do not target certain employees due to their predisposed characteristics, employers do not run afoul of the ADA. Thus, employers utilizing AI technology to guard against possible COVID-19 exposure risk (by taking employees\u2019 temperatures), are not per se <\/em>engaging in discriminatory practices. However, even when an employer requires all employees to participate in a screening procedure, there still remains the potential for the employer to deploy the results of the screening in a discriminatory manner against certain groups of employees.<\/p>\n\n\n\n

The greatest potential for employee discrimination lies in the discretion used by employers when analyzing the data obtained via these AI technologies, because a clear and reliable set of rules governing the implementation of AI technology in the workplace to ensure employee safety amidst COVID-19 does not exist. Currently, employers generally use the CDC\u2019s recommendations as a guidepost when deciding how to utilize AI-generated data to determine whether employees can come to work. But, as noted above, and as expressed by former EEOC Chair, Janet Dhillon<\/a>, there is a concern that employers could use the data for discriminatory purposes, particularly against those of Asian descent. Furthermore, given many unknowns about the manifestation of COVID-19 infection and also the various conditions that can skew body temperature readings, these CDC-approved metrics (e.g., a body temperature above 100 degrees Fahrenheit) may not serve as the best \u201cyard-stick\u201d to determine whether an \u201cemployee poses a \u2018direct threat\u2019 to the health or safety\u201d of those in the workplace.  <\/p>\n\n\n\n

Thus, to ensure that AI technologies in the workplace can effectively assess the risk to COVID-19 (and potentially other public health-related risks in the future), without their metrics either (1) disproportionately negatively impacting certain employees or (2) effectively serving as a justification for employers to engage in unlawful discriminatory practices, the EEOC should issue new guidelines that directly govern the use AI technologies for worker health and safety. Promulgating standards will ensure that we do not sacrifice the equal employment opportunity of workers in the quest for public health safety.<\/p>\n\n\n\n

Meredith Doswell<\/strong><\/p>\n","protected":false},"excerpt":{"rendered":"

As we near the end of the COVID-19 pandemic, employers are grappling with how to transition employees back into the office while ensuring workplace safety. Many employers are relying on artificial intelligence (\u201cAI\u201d) and other sophisticated technologies to implement screening measures (e.g., body temperature checks) and other public health-recommended workplace protocols (e.g., social distancing) per …<\/a><\/p>\n","protected":false},"author":4,"featured_media":8352,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[51],"tags":[],"_links":{"self":[{"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/posts\/8304"}],"collection":[{"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/comments?post=8304"}],"version-history":[{"count":3,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/posts\/8304\/revisions"}],"predecessor-version":[{"id":8358,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/posts\/8304\/revisions\/8358"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/media\/8352"}],"wp:attachment":[{"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/media?parent=8304"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/categories?post=8304"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/journals.law.unc.edu\/ncjolt\/wp-json\/wp\/v2\/tags?post=8304"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}