0:00
HR Party of One is brought to you by BerniePortal.
0:03
The use of Artificial Intelligence (AI) in the workplace has sparked significant
0:09
concerns about increased discrimination. And with AI becoming increasingly difficult to
0:14
avoid in the workplace, I know I’ve definitely been thinking
0:17
about how it can impact workplace diversity, equity, and inclusion.
0:22
But will AI actually lead to more workplace discrimination? The answer isn’t straightforward.
0:28
If employers rely too heavily or carelessly on AI, they could risk unintentional discrimination and
0:35
face serious legal consequences. Recognizing the growing impact of AI on hiring practices and the
0:41
need for oversight, the EEOC appointed its first Chief AI Officer on June 10,
0:48
2024. This move highlights the EEOC’s commitment to ensuring that AI tools are
0:54
used fairly and ethically, protecting workers’ rights and promoting a more inclusive workplace.
1:00
A notable example of AI-induced discrimination occurred in 2023 when iTutorGroup,
1:06
an online learning platform based in China, used AI recruitment software programmed to
1:12
automatically reject all female applicants over the age of 55 and male candidates over the age of
1:18
60. This case marked the first U.S. discrimination lawsuit involving AI hiring tools, and highlighted
1:25
the potential risks of unregulated AI use. Keep watching to learn what the consequences were.
1:31
It’s clear that AI is already making a significant impact on the workplace, and policymakers are
1:37
actively responding to these changes. In this episode, we’ll cover the pros and cons of AI use
1:43
in the workplace and EEOC Guidelines. Let’s get started!
Pros of AI Use in the Workplace
1:47
The Pros and Cons of AI Use in the Workplace.
1:50
AI in the workplace has its pros and cons, particularly in our day-to-day
1:55
HR processes. Let’s start with the pros. As a busy HR Party of One, there’s no doubt
2:00
that AI can save you time and energy as you complete your day-to-day HR processes. Today,
2:06
many HR professionals are regularly using AI to help with the following workplace tasks:
2:12
Recruitment and hiring: to analyze resources, screen applicants and resumes. If you have
2:18
thousands of applications for one role, using AI can be helpful. According to Sage Group,
2:24
at least 24% of companies already use AI to hire employees. Using AI in recruitment and
2:30
hiring can also reduce costs and ensure the best candidates are seen quickly.
2:35
Employee Onboarding: to personalize learning using virtual reality systems
2:41
Task and Meeting Management: to optimize scheduling, reallocate resources,
2:45
and track meeting notes and action items, saving HR time and energy.
2:50
Softwares like FirefliesAI can even send recap emails to meeting attendees for you!
2:55
HR Administration and Compliance: to automate payroll and benefits administration processes,
3:01
reducing the risk of human error.
Cons of AI Use in the Workplace
3:03
Those are some pretty attractive reasons to use AI for recruitment,
3:06
so why wouldn’t you dive right in? Because while AI can be a valuable tool in the HR toolkit,
3:13
the cons are significant, especially when used in recruitment and hiring. AI use in recruitment can
3:19
lead to unintended biases and violations of the ADA, ADEA, the Civil Rights Act of 1964,
3:27
and other significant discrimination laws. Let’s dive into how this can happen:
3:33
Historical Data Bias: One major issue with using AI as a screening tool is
3:38
that it relies on the data you provide. For example, if you detail the application
3:43
requirements and describe previous successful employees, the AI will use this information
3:49
to find candidates it thinks you want to hire, and this can lead to problems.
3:54
Imagine your company has been around for 50 years and historically hired more men
3:58
than women for certain roles. If you supply the AI with this historical data, it might
4:04
limit its search to men, perpetuating the pattern of gender bias. In fact,
4:09
Amazon had to scrap an AI hiring tool because it was doing exactly this. The fact of the matter is,
4:16
the technology isn’t quite there yet to avoid unintended discriminatory practices.
4:22
Exclusionary Filters: AI might unintentionally use filters that exclude qualified candidates from
4:29
diverse backgrounds. For example, it might require certain keywords or experiences that are less
4:34
accessible to underrepresented groups, leading to a lack of diversity in your candidate pool.
4:41
Automated Rejections: Another issue is when AI systems automatically reject applicants who,
4:47
due to a disability (for example), cannot interview in person. This denies reasonable
4:53
accommodations that could enable these candidates to participate fully in the
4:58
hiring process. Additionally, AI might ask candidates questions about disabilities or
5:04
medical conditions before a job offer, which can be discriminatory. As a general rule,
5:09
screening out an applicant due to their disability is considered a violation of the ADA if the
5:16
applicant can perform the essential functions of the job, with or without reasonable accommodation.
5:22
These are only a few of the ways that AI use in
5:24
the workplace can go wrong– but the possibilities are endless.
5:28
Keep in mind that there are legal ways to screen candidates efficiently! Knock-out
5:33
questions like “Are you authorized to work in the United States?” or “Do you have X years
5:38
of experience in this field?” or “Are you willing to relocate?” to narrow down your candidate pool.
5:45
Using BerniePortal, I can customize these questions and have full control
5:49
over how they are used. That way, I can rule out candidates who would not
5:54
ultimately be interested in a role at our company, without risking discrimination.
The AI Bill of Rights
5:59
It’s clear that policymakers are quickly responding to the prevalence of AI in
6:04
the workplace. Just a year before the iTutor Group lawsuit, in October 2022,
6:10
the EEOC addressed the increasing use of AI in society by developing
6:15
a blueprint for an AI Bill of Rights highlighting the following worker rights:
6:21
You should be protected from unsafe or ineffective systems.
6:25
You should not face discrimination by algorithms and systems should
6:29
be used and designed in an equitable way. You should be protected from abusive data
6:34
practices via built-in protections and you should have agency over how data about you is used.
6:42
You should know that an automated system is being used and understand how and why
6:47
it contributes to outcomes that impact you; and You should be able to opt out, where appropriate,
6:53
and have access to a person who can quickly consider and remedy problems you encounter.
AI Disparate Impact Guidance
6:59
By 2023, the EEOC released a document called the AI Disparate Impact Guidance
7:06
which clarified that the EEOC treats employer use of algorithmic decision-making tools as an
7:12
employment “selection procedure” under Title VII of the Civil Rights Act of 1964. This
7:18
notified employers that they must comply with anti-discrimination laws as technology evolves.
7:25
The document focuses on preventing workplace discrimination, and asks
7:29
the following primary questions: Does the employer use a particular
7:33
employment practice that has a disparate impact on the basis of race, color,
7:39
religion, sex, or national origin? If so, can the employer show that
7:44
the selection procedure is job-related and consistent with business necessity?
7:49
Is the selection procedure associated with the skills needed to perform the job successfully? And
7:55
Is there a less discriminatory alternative available?
The Four-Fifths Rule
7:58
It can be difficult for HR to determine if a certain algorithmic selection process used in
8:03
recruitment or hiring is inherently biased or discriminatory. The “Four-Fifths rule,”
8:09
published by the EEOC in 1978, states that one rate is substantially different from another
8:15
if their ratio is less than four-fifths (or 80%). This rule of thumb can help employers
8:22
determine if their processes result in discrimination or involve bias.
8:27
The technical assistance document uses the following example to explain the rule:
8:32
An employer administers a personality test to 80 White applicants and 40 Black applicants.
8:38
48 of the White applicants and 12 of the Black applicants advance to the next step.
8:44
The selection rate is thus 48/80= 60% for White applicants and 12/40 or 30%
8:53
for black applicants. The ratio of the two rates is 30/60=50%, which is less than 80%,
9:01
thus failing to meet the requirements of the four-fifths rule. The employer may be able to
9:07
conclude that the personality test administered discriminates against Black applicants.
9:13
Keep in mind that the rule is just a guideline and isn’t applicable in all
9:17
circumstances. An AI-related employment practice that follows this rule isn’t necessarily lawful.
9:24
There has actually been a push in recent years to ditch the four-fifths rule altogether.
9:29
According to an AI Bias panel, “The four-fifths rule is a proclamation
9:34
that practices that keep up to 20% more of a disadvantaged group from being hired should
9:41
be accepted without further scrutiny.” This raises an important question: Is it legal or
9:47
ethical to discriminate by a smaller margin? As an HR pro, you know that it is illegal for
Principles for AI Systems in the Workplace
9:53
even a single person to be discriminated against because of the AI tools you use.
9:58
That’s why it’s extremely important for HR leaders and hiring managers to analyze,
10:03
evaluate, and regularly monitor the processes that their job sites and recruitment platforms use. To
10:10
support this, on May 16, 2024, the White House released a fact sheet with a set of principles
10:16
employers can apply to the deployment of AI systems in the workplace. Let’s take a look:
10:23
Centering Worker Empowerment: Workers should have genuine input in the design,
10:28
development, testing, training, use, and oversight of AI systems.
10:32
Ethically Developing AI: AI systems should be designed,
10:36
developed, and trained in a way that protects workers.
10:40
Establishing AI Governance and Human Oversight:
10:43
Organizations should have clear governance systems,
10:46
procedures, human oversight, and evaluation processes for AI systems in the workplace.
10:52
Ensuring Transparency in AI Use: Employers should be transparent with both employees
10:59
AND job seekers about the AI systems that are being used in the workplace.
11:04
Protecting Labor and Employment Rights: AI systems should not violate or undermine workers’ rights.
11:11
Using AI to Enable Workers: AI systems should assist, complement,
11:17
and enable workers, and improve job quality.
11:21
Supporting Workers Impacted by AI:
11:23
Employers should support or upskill workers during job transitions related to AI.
11:29
Ensuring Responsible Use of Worker Data: Workers’ data collected, used,
11:34
or created by AI systems should only be used to support legitimate business aims. This means
11:40
that the responsibility falls on you to ensure any data collected or inputted by AI is kept secure.
11:47
The White House’s framework for AI in the workplace can help you guide
11:51
your business and make the most ethical decisions when using AI.
11:55
At our most recent Weekdays with Bernie conference, a representative from Ford
11:59
Harrison LLP, urged HR professionals to understand that the responsibility will
12:04
ultimately fall on them to make ethical choices when using AI, which brings us
12:09
to the concept of the “No-Third-Party Shield.” If you crash into another driver on the road,
12:15
does the other person involved get mad at you or at your vehicle? Under Title VII,
12:20
an employer is responsible for its use of algorithmic decision-making tools or AI
12:27
even if their selection process goes through a third-party entity or software. This means
12:32
you could still be held liable for what a platform does on behalf of your employer.
Best Practices for Protecting Your Business & Implementing AI
12:37
At Weekdays with Bernie, Meredith Box shared the following best practices
12:42
for protecting your business and implementing AI:
12:44
Bring the experts to the table: If you are really unsure about an employment practice
12:49
you are seeking to implement, reach out to an employment lawyer or AI expert
12:54
Identify AI tools and processes: Develop clearly defined policies and procedures for employment.
13:01
Regularly self-audit and monitor: Test the data and the product on a regular basis,
13:06
and retain records for evidence. Conduct audits and self analyses
13:10
on a regular basis: Determine if your employment processes have a large negative
13:16
impact on particular groups of people. Implement regular user training:
13:21
this can help ensure all parties involved are using AI ethically.
13:25
Know the purpose: if you’re going to use AI, have a clear reason for why you’re using it.
13:30
Be transparent: If you’re going to use AI in the recruitment or interviewing process,
13:36
inform candidates of that as early on as in the job descriptions.
Final Thoughts
13:42
Now that we’ve covered legal guidelines and best practices for AI in the workplace, I hope you
13:46
feel empowered to use AI as a tool rather than a crutch. Since the technology isn’t foolproof yet,
13:53
HR must carefully manage it and its use should always be balanced with human oversight.
13:59
Oh and what happened to iTutorGroup? The company was sued for violating
14:03
the Age Discrimination in Employment Act (ADEA) and paid a fine of $365,000 to more
14:10
than 200 job applicants! Remember—your role is as strategic as you make it!