AUTOMATED VIDEO INTERVIEW SCORING SYSTEMS: EMPLOYMENT DISCRIMINATION, ALGORITHMIC BIAS AND LEGAL ACCOUNTABILITY

1. INTRODUCTION

Automated video interview scoring systems use artificial intelligence (AI), machine learning, facial analysis, speech recognition and behavioural analytics to evaluate candidates during recruitment. These systems may assess spoken responses, vocabulary, facial expressions, eye movements, voice characteristics, communication patterns and other recorded behavioural indicators to generate candidate scores or rankings.

Although these technologies can improve recruitment efficiency, they raise significant legal concerns involving discrimination, privacy violations, disability accommodation, transparency, procedural fairness and unlawful employment decision-making. An algorithm may disadvantage candidates because of their gender, age, disability, accent, ethnicity or socioeconomic background, even when these characteristics have no legitimate relationship with their professional competence.

The central legal question is whether employers can lawfully rely on automated interview assessments when the underlying evaluation criteria are inaccurate, discriminatory or insufficiently transparent.

2. LEGAL FRAMEWORK GOVERNING AUTOMATED INTERVIEW SCORING

In India, automated recruitment systems must operate consistently with constitutional principles, applicable employment legislation and data protection requirements.

A. CONSTITUTIONAL PROTECTION

Article 14 guarantees equality before the law and protection against arbitrary state action.

Article 15 prohibits discrimination on specified grounds in the constitutional contexts covered by the provision.

Article 16 protects equality of opportunity in public employment.

Article 21 protects life and personal liberty, including recognised privacy interests.

Article 16 is particularly relevant to algorithmic recruitment conducted by government employers and public authorities. Private-sector recruitment is principally governed by applicable statutory employment protections rather than a direct general application of Article 16.

B. DATA PROTECTION REQUIREMENTS

The Digital Personal Data Protection Act, 2023 establishes a framework governing the processing of digital personal data. Video interviews may involve personal data such as recordings, voice information and identifiable facial images.

Employers must consider applicable requirements concerning lawful processing, notice, consent where relied upon, security safeguards and individual rights. Biometric or facial-analysis information may require particularly careful treatment depending on the nature and purpose of processing.

C. EMPLOYMENT AND DISABILITY PROTECTION

The Rights of Persons with Disabilities Act, 2016 is relevant where automated assessments disadvantage candidates with disabilities. Section 3 prohibits discrimination against persons with disabilities, subject to the statutory framework, and requires equality and non-discrimination.

Employers should provide reasonable accommodation and avoid treating disability-related speech, movement or facial differences as indicators of poor professional suitability.

3. MAJOR LEGAL RISKS ASSOCIATED WITH AUTOMATED VIDEO INTERVIEW SCORING

A. ALGORITHMIC DISCRIMINATION

An automated system may assign lower scores to candidates whose accents, speech patterns, facial movements or communication styles differ from the training data.

For example, a candidate with a regional Indian accent may receive a lower communication score despite providing technically correct and professionally appropriate answers.

B. DISABILITY-RELATED BIAS

Candidates with autism, facial paralysis, hearing impairments, speech disabilities or other disabilities may be unfairly evaluated through facial-expression recognition or speech-analysis tools.

A system that interprets limited eye contact as insufficient confidence may disadvantage neurodivergent candidates.

C. GENDER AND AGE DISCRIMINATION

AI models trained on historically unrepresentative recruitment data may reproduce gender and age-related stereotypes. Facial characteristics, vocal pitch and perceived enthusiasm may influence scores without demonstrating actual job competence.

D. LACK OF TRANSPARENCY

Candidates may not know which characteristics are being assessed, how scores are calculated or whether an automated ranking determines their rejection.

E. PRIVACY AND DATA MISUSE

Continuous recording, facial analysis, voice profiling and indefinite retention may exceed what is reasonably necessary for recruitment. Secondary use of interview recordings for unrelated AI training may create additional compliance concerns.

4. RELEVANT CASE LAWS

The following judicial decisions provide direct or analogous legal principles. They do not all concern AI-based video interviews specifically.

1. National Legal Services Authority v. Union of India (2014) – DIRECTLY ANALOGOUS AUTHORITY

The Supreme Court recognised the constitutional rights and dignity of transgender persons and emphasised equality and non-discrimination.

Relevance: Automated recruitment systems must not reproduce discriminatory assumptions concerning gender identity or gender expression.

2. Jeeja Ghosh v. Union of India (2016) – DIRECTLY ANALOGOUS AUTHORITY

The Supreme Court recognised the dignity, autonomy and equal treatment of persons with disabilities.

Relevance: Employers must ensure that automated interview assessments do not disadvantage candidates because of disability-related characteristics.

3. Vikash Kumar v. Union Public Service Commission (2021) – DIRECT AUTHORITY ON REASONABLE ACCOMMODATION

The Supreme Court explained that reasonable accommodation is an essential component of substantive equality under disability law.

Relevance: Recruitment systems must accommodate candidates who require alternative interview formats, additional time or other disability-related adjustments.

4. K.S. Puttaswamy v. Union of India (2017) – DIRECT AUTHORITY ON PRIVACY

The Supreme Court recognised privacy as a fundamental right under Article 21.

Relevance: Extensive video recording, facial analysis, voice profiling and unnecessary retention of candidate information raise constitutional privacy concerns, particularly in public-sector recruitment.

5. Anuj Garg v. Hotel Association of India (2008) – DIRECTLY ANALOGOUS AUTHORITY

The Supreme Court rejected discriminatory employment restrictions based on gender stereotypes and emphasised substantive equality.

Relevance: Automated scoring systems must not rely on stereotypical assumptions about how men and women should communicate, behave or demonstrate confidence.

6. Lt. Col. Nitisha v. Union of India (2021) – DIRECT AUTHORITY ON INDIRECT DISCRIMINATION

The Supreme Court recognised indirect discrimination and explained that apparently neutral requirements may produce discriminatory outcomes.

Relevance: An interview-scoring algorithm may appear neutral but disproportionately disadvantage particular groups through facial, vocal or behavioural criteria.

7. Maneka Gandhi v. Union of India (1978) – ANALOGOUS AUTHORITY ON PROCEDURAL FAIRNESS

The Supreme Court established that procedures affecting rights and liberty must satisfy standards of fairness, reasonableness and non-arbitrariness.

Relevance: Public recruitment authorities should not rely on arbitrary automated assessments without appropriate procedural safeguards.

8. State of Kerala v. N.M. Thomas (1976) – DIRECT AUTHORITY ON SUBSTANTIVE EQUALITY

The Supreme Court developed the principle that equality requires attention to substantive disadvantage rather than merely formal identical treatment.

Relevance: Identical automated interview criteria may produce unequal outcomes where candidates have different disabilities, linguistic backgrounds or accessibility requirements.

5. EMPLOYER LIABILITY AND LEGAL ACCOUNTABILITY

Employers cannot necessarily avoid responsibility by arguing that discriminatory recruitment decisions were generated by third-party AI software.

Potential legal consequences include:

Discrimination claims under applicable employment and disability legislation.

Constitutional challenges involving arbitrary public recruitment decisions.

Privacy-related complaints concerning unlawful or excessive personal-data processing.

Recruitment disputes involving unreasonable or undisclosed selection criteria.

Contractual disputes concerning defective AI recruitment services.

Regulatory scrutiny where automated processing violates applicable statutory obligations.

The precise liability depends on the employer's conduct, the applicable legislation, the nature of the decision and the evidence establishing discrimination or unlawful processing.

6. PRACTICAL COMPLIANCE REQUIREMENTS

Employers implementing automated video interview scoring should adopt the following safeguards:

Conduct independent algorithmic bias assessments before deployment.

Validate scoring criteria against actual job-related competencies.

Avoid using facial expressions, eye contact or vocal characteristics as unsupported proxies for professional ability.

Provide reasonable accommodations for candidates with disabilities.

Inform candidates about automated assessment and relevant data-processing practices.

Establish meaningful human review before adverse recruitment decisions.

Provide an appropriate mechanism for candidates to challenge inaccurate assessments.

Maintain audit records of scoring decisions and algorithmic changes.

Restrict access to interview recordings and establish proportionate retention periods.

Periodically examine selection outcomes for disproportionate adverse effects.

Human review should be meaningful rather than merely approving the algorithm's recommendation without independent assessment.

7. CONCLUSION

Automated video interview scoring systems present significant challenges to employment equality, privacy, disability rights and fair recruitment. Although AI-based recruitment tools can improve administrative efficiency, technological convenience cannot justify discriminatory or arbitrary candidate evaluation.

Indian constitutional jurisprudence, particularly the decisions in Puttaswamy, Jeeja Ghosh, Vikash Kumar and Nitisha, provides important principles for evaluating the legality of automated recruitment practices. Employers should ensure that AI systems assess genuine professional competencies rather than irrelevant facial, vocal or behavioural characteristics.

Ultimately, automated recruitment must remain accountable, transparent, accessible and consistent with equality and non-discrimination principles. The responsibility for fair employment decisions cannot simply be transferred from employers to algorithms or technology vendors.