AI for Organizations of Persons with Disabilities : A Practical Guide

Before launching an AI-focused project for program participants, begin by strengthening your own AI knowledge and encouraging staff to do the same. Numerous free AI courses are available online that can help build practical skills and improve overall competence.

Before adopting AI tools, it is useful for OPDs to reflect on their organizational needs and capacity. Many OPDs operate with small teams where staff handle multiple responsibilities, from communications and proposal writing to member outreach.

AI can potentially support areas that consume significant time, such as content creation, grant writing, translation for local language engagement, organizing data, drafting routine documents, or producing meeting summaries.

To make an informed decision, organizations can reflect on questions such as:

• Where do we currently experience the greatest capacity constraints as a team?
• Which tasks take a lot of time but follow predictable or repetitive processes?
• Are there areas where faster communication or reporting would strengthen our advocacy and member engagement?
• Do we struggle with translating information into accessible formats or local languages for our members?
• Which routine administrative tasks consume time that could be redirected toward advocacy or program work?

Reflecting on these questions can help organisations determine whether and where AI might add practical value to their work.

Rather than introducing AI across many activities at once, it can be helpful to begin with a single tool applied to one specific task. This allows teams to explore how the tool works, understand its strengths and limitations, and build confidence before expanding its use to other areas.

Starting small also makes it easier to evaluate whether the tool genuinely improves efficiency or supports the organization’s work.

The table below step 6 shows some recommended starting tools and their use cases.

Using AI effectively requires learning how to communicate clearly with it. Unlike a search engine that simply retrieves information, AI responds to instructions and context. The quality of the output often depends on the clarity of the request.

·       Ask Better Questions: Train your team to give the AI context. If you ask a bad question, you will get a bad answer.

·       Use Natural Language: Speak to the AI as if you are giving instructions to a human assistant.

·       Refine and Repeat: If the first answer isn't right, don't give up. Tell the AI what

A Simple Formula for OPDs

Role + Context + Task + Format + Tone

Example of a WEAK prompt:

"Write about a public participation in Nakuru to inform members."

Example of a STRONG prompt:

"You are a communications officer for a Kenyan OPD. Write a 200-word WhatsApp message in English informing our members that the County Assembly will hold a public participation session on the disability services budget on Friday 15th November at County Hall, Nakuru. The tone should be warm and encouraging and include a call to action to attend or send written submissions."

Even a simple one-page policy can help guide how AI is used within an OPD and provide basic safeguards for the organization. It does not need to be complex. Clear, practical guidelines are often enough to support responsible and consistent use.

An AI policy for an OPD can outline a few clear principles to guide responsible use. These may include:

• Permitted uses: Clarify the types of tasks where AI may be helpful, such as drafting, summarising, or translating content, while noting that decisions affecting members should always remain human-led.
• Data protection: Emphasise that personal information about members such as names, identification numbers, or medical details should not be entered into public AI tools.
• Human review: Ensure that any AI-generated content is reviewed and approved by a staff member before it is shared or published.
• Transparency: Consider whether and how the organization will acknowledge the use of AI in its communications.
• Accessibility checks: Confirm that AI-generated materials are reviewed to ensure they meet accessibility standards before being distributed.

The following is a link of an AI Usage Policy template

Engagement with AI should go beyond internal organizational use. There is an increasing need to ensure that AI technologies are designed and deployed in ways that are inclusive of persons with disabilities.

OPDs have an important role to play in raising awareness, contributing perspectives from lived experience, and advocating for AI systems that are accessible, equitable, and responsive to the realities of persons with disabilities in Kenya and across the region.

AI systems learn from existing data. If that data lacks diversity or representation, the outputs may reflect those same gaps.

• Stereotypes and exclusion: AI tools may unintentionally reproduce stereotypes or overlook accessibility needs if such perspectives were not well represented in their training data.
• Global data imbalance: Much of the data used to train AI systems does not always reflect the realities of many parts of the world. As a result, AI tools may sometimes struggle to fully understand local contexts in Kenya, including cultural dynamics, local languages, and the various barriers, physical, communication, attitudinal, and institutional that persons with disabilities encounter.

AI systems generate responses based on probability rather than verified truth.

• Potential inaccuracies: AI may produce confident-sounding responses that are incomplete or incorrect. This can be particularly problematic when discussing sensitive topics such as legal rights, policies, or disability-related services
• Importance of verification: AI-generated outputs should always be reviewed and validated by a knowledgeable person before being shared.

OPDs often manage sensitive personal information about their members, including identification details, contact information, and disability-related data.

• Risks with public tools: Uploading documents or personal information into public AI systems may expose that information beyond the organization’s control.
• Protecting confidentiality: Personal details such as names, identification numbers, member records, or confidential organizational documents should not be entered into public AI platforms without proper safeguards.

Maintaining strong data protection practices helps ensure that the rights, dignity, and privacy of persons with disabilities are respected when using AI tools.

AI is designed to support human work, not replace human judgment.

• Maintaining human connection: Maintaining human connection: Relying heavily on automated tools for communication or support may weaken the personal relationships that are central to community-based advocacy.

The use of AI should therefore complement not replace authentic storytelling and the lived experiences of persons with disabilities, which remain at the heart of effective advocacy and social change.

• Building internal capacity: Organizations benefit most from AI when staff understand how to guide and manage these tools responsibly.

The most practical approach for many OPDs is to start small by experimenting with one tool and one task. By learning, experimenting, and engaging with AI early, OPDs can help ensure that technology supports greater accessibility, participation, and disability inclusion.

Article by: Maryanne Emomeri