AI Risks and Ethical Dilemmas That Could Affect You

AI systems now decide who gets hired, who gets healthcare, and who gets flagged by police. These decisions happen without explanation.

We analyzed documented cases of AI harm across hiring, healthcare, and criminal justice. This research draws from court filings, academic studies, and firsthand accounts to show how artificial intelligence creates ethical challenges for ordinary people.

Here is what the data reveals about AI risks and ethical dilemmas that could affect you.

Key takeaways

• 83% of companies use AI algorithms to screen resumes, with 50% using AI exclusively for initial rejections
• Healthcare AI systems have shown 90% error rates while denying coverage and overriding medical professionals
• Machine learning models trained on historical data perpetuate discrimination in hiring, lending, and criminal justice
• Black boxes prevent affected individuals from understanding or challenging AI decisions about their lives
• Addressing these ethical issues requires transparency, human judgment in high stakes decisions, and accountability for AI errors

How AI systems affect your opportunities

Before exploring solutions, we need to understand the scope of the problem. AI development has outpaced oversight, creating situations where consequential decisions happen without human review.

Hiring decisions

Companies process job applications at unprecedented scale using AI. Research shows 83% of companies use AI algorithms to screen resumes. Of these, 50% use artificial intelligence exclusively for initial rejections. No human ever sees many applications. These workplace AI risks affect candidates who never learn why they were rejected.

The potential for AI bias became clear when Amazon discovered its recruiting tool learned to penalize resumes containing "women's" because it trained on male dominated historical data. The system downranked candidates who attended women's colleges or participated in women's organizations.

University of Washington research published October 2024 found language models preferred white associated names 85% of the time versus 9% for Black associated names. Black males faced the steepest disadvantage with only 15% preference rate. A parallel University of Chicago study in Nature found AI generates decisions associating speakers of African American English with lower prestige jobs.

These AI models process amounts of data but inherit biases from their training sets. The pattern recognition that makes machine learning powerful also makes it vulnerable to encoding historical discrimination.

Healthcare coverage

A class action lawsuit against UnitedHealth alleges their AI system called nH Predict has a 90% error rate yet continues denying elderly Medicare Advantage patients coverage for post acute care. The algorithm overrides recommendations from medical professionals to cut costs.

Plaintiffs include families of Gene Lokken, 91, who paid $150,000 out of pocket before his death, and Dale Tetzloff, 74, who paid $70,000 after coverage denial. The societal impact extends beyond individual cases. When AI systems deny coverage at scale, they shift costs onto patients and families while reducing accountability.

Analysis of 181 Reddit threads in medical subreddits found three dominant concerns among medical professionals: fear of replacement, tension in physician AI relationships, and trust gaps with patients. Research on AI in clinical settings confirms these patterns extend beyond social media discussions. One medical student noted growing concern that habitual AI consultation was undermining independent diagnostic thinking.

Criminal justice

Robert Williams spent 30 hours in jail after facial recognition falsely identified him as a robbery suspect. He was arrested in front of his family based solely on an AI match that was wrong. A 2024 settlement required Detroit Police to audit all facial recognition cases since 2017.

Porcha Woodruff's case generated particular outrage. The eight months pregnant Black woman was arrested for carjacking despite the actual perpetrator not being visibly pregnant. The AI system matched her face to surveillance footage without accounting for obvious physical differences.

Nearly every documented wrongful facial recognition arrest involves Black individuals. Documented cases from UNESCO and other organizations reveal how implementation of AI without proper testing creates discriminatory outcomes in high stakes contexts where human judgment should remain central.

Why these problems persist

The ethical challenge stems from three factors that compound each other.

Opacity

Most AI algorithms operate as black boxes. The system produces outputs but even developers cannot always explain why specific decisions were made. When machine learning models reject applications or deny claims, affected people receive no meaningful explanation. They cannot challenge what they cannot understand.

This opacity exists by design in many cases. Companies treat their AI models as proprietary. Revealing how decisions are made might expose the system to gaming. But this secrecy also prevents accountability and makes it impossible for affected individuals to identify errors or discrimination.

Scale

AI systems process millions of decisions daily. A 1% error rate sounds acceptable until you realize it affects thousands of people. When 83% of companies use AI for resume screening, even small bias rates translate to massive discrimination at population level.

The long term consequences include:

• Qualified candidates locked out of industries based on name or background
• Patients denied necessary care while appeals take months
• Innocent people flagged by law enforcement databases

Surveys confirm widespread anxiety. 89% of workers express concern about job security due to AI. 43% report knowing someone who lost a job to AI. The personal data these systems collect and process raises additional concerns about surveillance and privacy.

Speed

Organizations deploy AI faster than regulators evaluate them. A 2024 court ruling established that companies cannot disclaim responsibility for AI errors. Air Canada argued its chatbot was a separate legal entity before being ordered to pay damages.

This signals that accountability is possible when individuals push back.

What transparency looks like

Addressing these ethical dilemmas requires specific changes to how organizations deploy AI systems. The good news is that solutions exist and public pressure is producing results.

Disclosure requirements

Workers, patients, and applicants deserve to know when AI affects decisions about them. Several jurisdictions now require:

• Notice that AI was used in hiring decisions
• Explanation of factors the AI considered
• Human review process for adverse decisions
• Right to opt out of AI evaluation in some contexts

The 2024 Air Canada ruling established important precedent. Companies cannot disclaim responsibility for AI errors by calling systems separate legal entities. Courts will hold organizations accountable for the AI they deploy.

Bias audits

Organizations using AI for high stakes decisions should conduct regular audits. This means testing whether AI models produce different outcomes for different demographic groups and correcting disparities before deployment.

The Mobley v. Workday class action alleges AI hiring software discriminated based on race, age, and disability across hundreds of job rejections. Cases like this create financial incentive for companies to audit their systems proactively.

Appeal processes

When AI systems make errors, affected individuals need recourse. The current situation where people cannot challenge algorithmic decisions violates basic principles of fairness that apply to human decision makers.

71% of Americans support more AI regulation. Support among local policymakers rose from 55.7% to 73.7% in two years. The 2023 Hollywood strikes secured protections requiring consent for digital replicas, showing collective action works. Legal and governance frameworks continue evolving as public pressure produces change.

FAQ

Why do AI systems discriminate when designed to be neutral?

AI learns patterns from training data. Historical data reflects past discrimination. Without active intervention, AI algorithms perpetuate those patterns regardless of design intentions. Academic resources on AI ethics explain this dynamic in detail.

Can I find out if AI was used in a decision about me?

This depends on jurisdiction. Some regulations now mandate disclosure for hiring decisions. You can ask directly, though companies may not be required to answer in all contexts.

What recourse exists when AI makes an error?

Legal frameworks are developing. Some jurisdictions require human review processes. Document the error, request explanation, and consider whether the organization violated disclosure requirements.

Which decisions commonly involve AI?

Hiring, lending, healthcare coverage, insurance pricing, criminal sentencing, and housing applications all commonly involve AI components. Assume any large scale decision process may use AI.

Summary

AI risks and ethical dilemmas that could affect you center on one core problem: consequential decisions happening without transparency, explanation, or accountability.

The combination of black boxes, massive scale, and rapid deployment creates situations where AI systems deny jobs, healthcare, and freedom based on flawed data and biased training. Affected individuals often cannot learn why decisions were made or how to challenge them.

This is fixable. Disclosure requirements, bias audits, and appeal processes provide a roadmap. The 2024 Air Canada ruling shows courts will hold organizations responsible for AI errors when individuals push back.

Understanding how AI systems work is the first step toward demanding they work fairly.

‍

Schedule a free data strategy consultation