Did AI Just Cost You a Job? How Algorithmic Bias Is Hurting California Workers

Artificial intelligence (AI) and machine learning algorithms are increasingly being used to make high-stakes decisions in employment, including screening resumes, evaluating job performance, and even determining pay.

While AI has the potential to reduce human biases, a growing body of research shows these systems can also perpetuate or even amplify discrimination.

For California employees, algorithmic bias is emerging as a major concern and a new frontier in the fight against workplace discrimination.

What Is Algorithmic Bias?

Algorithmic bias occurs when a computer system reflects the implicit values, prejudices, or stereotypes of the humans involved in coding, collecting, selecting, or using data to train the algorithm.

Since machine learning models depend on training data to find patterns and make decisions, they can “learn” and perpetuate human biases baked into that input data.

Some common ways algorithmic bias creeps into AI-powered employment tools:

• Biased training data: If the resumes or performance data used to train a model underrepresents certain groups (e.g., by race, gender, age), the resulting predictions will likely disadvantage those groups.
• Faulty feature selection: Developers may inadvertently embed prejudiced assumptions into a model’s decisions – e.g., prioritizing an applicant’s school over their skills.
• Feedback loops: Some algorithms mine historical data to predict future outcomes, replicating past discriminatory hiring patterns.
• Proxy discrimination: Models may consider factors like zip code that closely correlate with protected categories like race, causing disparate impact.

Machine learning algorithms can drift and degrade in accuracy over time. If not carefully monitored, they may develop biases unnoticed.

Real-World Examples of AI Bias Impacting Workers

Unfortunately, algorithmic discrimination isn’t just hypothetical – it’s already negatively impacting job seekers and employees here in California and beyond:

• Gender bias in recruiting: In 2018, Amazon scrapped a recruiting algorithm that was systematically discriminating against female applicants, likely because it was trained on historically male-dominated resumé data.
• Racial bias in facial analysis: AI-powered facial recognition tools used in video interviews have shown higher error rates for people of color, especially those with darker skin tones.
• Ableist bias in personality tests: Some employers use AI pre-employment personality tests that may screen out qualified candidates with mental health disabilities.
• Age bias in targeted job ads: Automated processes that determine which users see certain job postings have been accused of disproportionately excluding older candidates.

California Laws on Algorithmic Bias

While there are currently no U.S. laws specifically governing the use of AI in employment, several existing California statutes provide some related protections:

• FEHA: The Fair Employment and Housing Act prohibits discrimination based on protected characteristics in hiring, firing, and terms of employment. FEHA applies to employers with 5+ employees (compared to 15+ under federal law) and may cover AI practices that produce disparate impact. (Cal. Gov. Code § 12940 et seq.)
• CCPA: The California Consumer Privacy Act grants rights to access, delete, and opt out of the sale of one’s personal data. Employees have some CCPA protections for data used in automated decision-making. (Cal. Civ. Code § 1798.100 et seq.)
• Salary history ban: California employers cannot rely on an applicant’s salary history or use race/ethnicity as a factor in setting pay, which may check AI-powered compensation discrimination. (Cal. Lab. Code § 432.3, § 1197.5)

However, AI moves fast and regulations struggle to keep up. That’s why California lawmakers have proposed the Workplace Technology Accountability Act (AB-1651), which would:

• Require employers to give notice of automated decision tools used in hiring, monitoring, and firing
• Mandate “impact assessments” to audit such systems for bias based on protected characteristics
• Create a private right of action for individuals harmed by biased AI with penalties of up to $20,000 per violation

If passed, AB-1651 would be the first U.S. law to comprehensively regulate AI in employment decisions. The bill reflects growing momentum around the need for AI oversight in the workplace.

What to Do If You Suspect AI Bias at Work

If you believe an algorithm unfairly screened you out for a job, denied a promotion, or determined your pay, you have options to fight back.

Here are some key steps:

1. Document everything. Keep records of the suspected AI bias incident, including any data, metrics, or communications provided to you. The more details, the better.
2. File an agency complaint. If you choose, you can file a discrimination charge against your employer with the California Civil Rights Department (CRD). CRD can investigate potential FEHA violations relating to unfair AI practices.
3. Consult an attorney. Speaking with an experienced employment lawyer can help you understand your full range of options and build the strongest case. Look for firms that understand AI bias issues.
4. Consider legal action. Depending on the details of your situation, you may have grounds for a lawsuit against your employer or the AI vendor for algorithmic discrimination. An attorney can advise you on the merits and strategy.

As AI rapidly evolves, individual cases will be crucial in clarifying how current and future laws apply in this new domain. Staying informed and taking action is key to keeping employers accountable.

Promoting AI Fairness in the Future of Work

Ultimately, mitigating AI harms will require a multi-pronged approach from employers, workers, developers, and regulators alike. Recommendations from experts include:

• Inclusive data and teams. Employers should strive to use representative data and involve diverse perspectives in AI design and oversight.
• Rigorous auditing. Companies must proactively monitor and test their AI for discriminatory impact, not wait for complaints.
• Transparency for workers. Employees deserve clear notice about any AI systems used to make decisions about them.
• Forward-looking regulations. Policymakers need to craft flexible frameworks that keep up with the unique challenges posed by AI discrimination.

With the proper safeguards, AI could actually help boost diversity, equity, and inclusion in the workplace. But it will take ongoing research, advocacy, and vigilance to make that vision a reality.

The TONG LAW team is closely tracking developments in AI fairness and how they affect workers’ rights in California. If you believe you’ve experienced algorithmic bias on the job, you’re not alone – and we’re here to help. Contact us for a free, confidential consultation with an attorney well-versed in this cutting-edge legal issue.

References

1. Dastin, J. (2018, October 9). Amazon scraps secret AI recruiting tool that showed bias against women. Reuters. https://www.reuters.com/article/us-amazon-com-jobs-automation-insight/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK08G
2. Buolamwini, J. (2019). Facial recognition technologies: A primer. Medium. https://medium.com/@Joy.Buolamwini/facial-recognition-technologies-a-primer-27c7e34ea5bf
3. Harwell, D. (2019). Rights group files federal complaint against AI-hiring firm HireVue, citing ‘unfair and deceptive’ practices. Washington Post. https://www.washingtonpost.com/technology/2019/11/06/prominent-rights-group-files-federal-complaint-against-ai-hiring-firm-hirevue-citing-unfair-deceptive-practices/
4. Dwoskin, E. (2019). Facebook is letting job advertisers target only men. ProPublica. https://www.propublica.org/article/facebook-is-letting-job-advertisers-target-only-men
5. A.B. 1651, 2023-2024 Reg. Sess. (Cal. 2023). https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=202320240AB1651