AI at the Crossroads: Advancing Inclusion or Deepening Divides?

6 May 2025: Artificial intelligence (AI) has become the defining zeitgeist of our era, with some hailing it as a “new electricity”, a dynamic technology that has the potential to transform education, health, work, and daily life. Conversation surrounding AI often focuses on the arms races and policymaking on risks. However, the focus on AI should move beyond risks and rivalries to also consider the possibilities it offers, possibilities ultimately shaped by people’s choices. This year’s Human Development Report by the UN Development Programme (UNDP) examines what distinguishes AI from previous digital transformations and its implications on human development (chapter 1), including how AI can enhance or subvert human agency (chapter 2).

The scale of AI’s reach is already staggering: nearly two-thirds of people in low, medium, and high Human Development Index (HDI) countries expect to use AI in education, health, or work within a year. While this surge presents enormous opportunities for social development, it also highlights profound risks of social exclusion if biases are left unchecked or benefits are unevenly distributed.

For persons with disabilities, AI holds tremendous potential to expand human development, yet persistent inequalities, design biases, and techno-deterministic approaches risk excluding those it seeks to empower.

Opportunities for Persons with Disabilities:

• Assistive technologies made more accessible: Smartphones can now function as cost-effective assistive tools such as live captioning, image recognition, text-to-speech and speech-to-text solutions.
• Generative AI breakthroughs: Generative AI can now create descriptions of images for blind and visually impaired people, or convert text into easy-to-read formats for people with development or intellectual disabilities.
• Human-machine interaction improvements: Large language models can support communication for users of alternative and augmentative communication or to translate sign language into voice or text.

Potential Risks and Challenges:

• Inequalities in access: Around 2.5 billion people need access to assistive technologies, but access remains highly unequal. Internet use averages only 28% among people with disabilities compared to 39% for people without.
• Digital exclusion: Around 96% of the top million websites do not comply with the International Web Content Accessibility Guidelines, limiting AI’s accessibility benefits.
• Patent concentration: Most patents for assistive technologies are filed in a handful of high-HDI countries (e.g., U.S., China, Japan), leaving Global South countries underserved.
• Bias in datasets and models: Underrepresentation of persons with disabilities in training data leads to inaccuracies, misinformation, and reinforcement of stereotypes.

The broader message extends to all aspects of social inclusion: AI has the potential to bring great benefits, especially to those who face social exclusion such as persons with disabilities, older persons, and those facing lack of access to education. The true test is whether we can ensure AI serves as a driver of human development that advances inclusion and agency for all, rather than deepening divides.

As preparations for the Second World Summit for Social Development gain momentum, these findings highlight a central challenge. The emergence of AI is not simply a matter of technological advancement but a question of governance and equity. The benefits for persons with disabilities demonstrate the immense promise of AI for advancing inclusion — but the persistence of gaps underscores the risks of leaving these issues unaddressed. Overcoming those gaps, and embedding accessibility as a principle of technological development, will be essential if AI is to contribute meaningfully to the Summit’s vision of inclusive social progress.

Read the full report by UNDP here.