Why Disability Inclusion in AI Matters for Recruitment, Decision Making, and more.

Artificial intelligence is rapidly transforming the way we work, offering new opportunities to streamline operations, enhance recruitment, and support decision making across organisations. But as we embrace these technologies, it’s essential to ensure that AI implementation is inclusive, particularly for disabled people. If we’re not careful, AI systems can inadvertently perpetuate bias and create new barriers, undermining the very progress we seek.

Recruitment: Rethinking the Use of AI

AI-powered recruitment tools are increasingly being promoted as efficient and objective solutions for shortlisting candidates and streamlining hiring. However, these systems often carry forward biases present in historical data or their design. For disabled candidates, this can mean unfair exclusion if the AI isn’t built with genuine inclusivity in mind.

Given these risks, organisations in New Zealand should take a cautious approach and consider avoiding reliance on AI-driven recruitment tools until there is greater confidence that these systems are truly inclusive and robust safeguards are in place.

This includes waiting for independent testing and certification for accessibility, and clear evidence that such tools do not introduce or perpetuate discrimination. In the meantime, traditional and human-led recruitment processes, designed thoughtfully with accessibility in mind and in accordance with legal requirements and codes of practice, remain the best way to ensure every candidate has a fair chance.

In NZ, this aligns with obligations under the New Zealand Human Rights Act 1993, which prohibits discrimination on the grounds of disability, and the Health and Safety at Work Act 2015, which requires workplaces to be safe and inclusive for all employees. Furthermore, the New Zealand Disability Strategy and the Code of Good Practice for the Employment of Disabled People provide guidance on ensuring accessible employment practices and fair treatment.

Decision Making Support: Lifting the Standard Across All Areas

AI is increasingly woven into the fabric of our workplaces, supporting a range of decisions. From internal operations to customer interactions. Each context brings its own set of challenges and responsibilities when it comes to inclusivity.

Internal Decision Making

• Performance Reviews & Promotions: Make sure AI tools consider reasonable adjustments and recognise diverse contributions and work styles. Regularly review AI-generated assessments for bias and ensure disabled staff have avenues for feedback and challenge.
• Resource Allocation: When using AI to allocate projects, shifts, or equipment, include criteria that accommodate individual needs and adjustments. Avoid “one size fits all” algorithms that can overlook required supports for disabled employees.
• Policy Enforcement: Use AI with human oversight to monitor compliance or predict issues, but always review edge cases with a person involved, especially where outcomes could impact someone's employment or wellbeing.

Customer-Related Decision Making

• Actively Consider Impacts When Designing/Reviewing Products: Where you are using AI to design or review products, ensure that you have safeguards in place to measure it against your corporate responsibilities.
• Customer Service Chatbots: Ensure all automated service channels are accessible to people with a wide range of needs. Test with real users from the disabled community and provide alternatives, such as phone or in-person service, where necessary.
• Eligibility & Application Assessments: If AI is used to decide on customer eligibility for services, loans, or benefits, ensure criteria do not inadvertently disadvantage disabled customers. Open these processes to independent review and allow human appeals.
• Personalisation & Recommendations: When using AI to tailor content or experiences, avoid making assumptions based on incomplete data. Allow users to control their preferences and to opt out of automated personalisation if it doesn’t serve their needs.

Practical Steps for Inclusive Decision Making

Establish clear policies that require human oversight for significant or sensitive decisions, especially those affecting employment status, benefits, or service access.

• Involve disabled people in the design, testing, and ongoing review of decision-making tools, gathering feedback and adjusting as required.
• Train staff to understand the limitations and appropriate use of AI in decision support and empower them to intervene when needed.
• Regularly audit algorithms for bias, document findings, and be transparent about how decisions are made and challenged.

Embedding Safeguards: What Good Looks Like

Inclusive AI isn’t just about avoiding harm, it’s about proactively creating systems that welcome and support everyone. This means building accessibility into every stage of AI deployment, from design and testing through to implementation and ongoing review. Clear policies, regular training, and a commitment to transparency are key.

• Audit algorithms for bias and discriminatory patterns
• Ensure accessibility in all user interfaces and assessments
• Involve disabled people in design, review, and feedback processes
• Provide alternative formats and reasonable adjustments for all AI-driven interactions
• Maintain human oversight for critical decisions affecting people’s lives and careers

The Path Forward

AI has the potential to make our workplaces more dynamic and inclusive, but only if we’re deliberate about how we design and deploy it. By embedding safeguards, prioritising accessibility, and listening to disabled voices, we not only create technology that works for everyone, but also support our legal and corporate responsibility obligations.

When we commit to these principles, we build a future where technology lifts barriers, rather than creating new ones, ensuring our organisations foster fairness and accountability.