AI digital inclusion and the communities being left further
AI digital inclusion is already a measurable problem in Australian communities. Not For Humans documents the AI digital inclusion gap through the lens of community organisations closest to the people being excluded. AI digital inclusion failures are not primarily technical. They are the result of systems built without the communities they affect, deployed into service environments where AI digital inclusion has not been assessed. The Australian Digital Inclusion Index confirms that AI digital inclusion gaps track closely with existing socioeconomic disadvantage. Not-for-profits and NGOs serving digitally excluded communities face a dual challenge: their clients face AI digital inclusion barriers, and many organisations lack the governance capacity to assess whether their own tools are making exclusion worse.
How AI digital inclusion gaps compound existing disadvantage in Australia
Impact Area 3
Digital inclusion and access.
When services move online and become AI-mediated, not having reliable internet or digital confidence stops being an inconvenience. It becomes exclusion from the services people need most.
One in five Australians remain digitally excluded. As AI shapes access to services, that exclusion becomes structural. Not-for-profits and NGOs serving these communities are closest to the gap and best positioned to name it.
AI digital inclusion is a governance question for every not-for-profit and NGO using digital tools
What is happening
Digital exclusion existed before AI. AI is making it structural.
The digital divide is not new. Uneven access to technology has reproduced existing inequalities in education, employment, health, and civic participation for decades. What AI changes is the nature of what exclusion means. Previous digital exclusion meant missing out on convenience. When AI shapes triage decisions, assesses eligibility, and mediates communication between people and the services they depend on, exclusion becomes something different.
It is no longer primarily about access to opportunity. It is about whether the systems making decisions about you have any accurate knowledge of your circumstances. Rural and indigenous communities often remain invisible in the datasets that train AI systems, increasing the risk of algorithmic bias and exclusion from essential services. Exclusion from access and exclusion from the design of the systems that govern access compound each other.
Digital exclusion from AI-shaped services is not only a service access problem. It is a democratic participation problem. The communities most affected by AI-driven decisions are frequently the communities with the least voice in how those systems were built.
What the evidence shows
First Nations Australians are twice as likely as other Australians to be digitally excluded. The national digital inclusion gap is 10.5 points, more than doubling in remote communities. These are the communities most likely to be shaped by AI systems trained on data that does not represent them.
Australian Digital Inclusion Index, 202520.6 percent of Australians are excluded or highly excluded from digital participation. Exclusion rates are highest for people aged 75 and over (66.5%), those without secondary education (54.5%), public housing residents (45.2%), and people with disability.
Australian Digital Inclusion Index, 2025- 29% vs 40% Regional not-for-profits adopting AI versus metropolitan organisations. The communities with the highest rates of digital exclusion are served by the organisations with the least AI capability. National AI Plan, 2025
- Documented AI uptake tracks closely with existing patterns of socioeconomic disadvantage, geographic remoteness, age, disability, and cultural background. The AI divide reproduces the digital divide. ACS AI Divide Survey, November 2025
- Invisible Rural and indigenous communities often remain invisible in the datasets that train AI systems, increasing the risk of algorithmic bias and exclusion from essential services. UNDP, The Next Great Divergence, December 2025
- 14% Of not-for-profits have an internal AI policy or governance framework. The organisations serving digitally excluded communities are among the least equipped to govern AI tools in their own practice. Infoxchange, 2025
An honest note on the evidence
The evidence on digital exclusion in Australia is stronger than in some other impact areas. The Australian Digital Inclusion Index provides robust longitudinal data. The connection between digital exclusion and AI-specific harm is well documented in aggregate. What is less documented is the downstream impact at the community service level. Not For Humans is building that picture from what organisations are actually observing.
AI digital inclusion is already a measurable problem in Australian communities. Not For Humans documents the AI digital inclusion gap through the lens of community organisations closest to the people being excluded. AI digital inclusion failures are not primarily technical. They are the result of systems built without the communities they affect, deployed into service environments where AI digital inclusion has not been assessed. The Australian Digital Inclusion Index confirms that AI digital inclusion gaps track closely with existing socioeconomic disadvantage. Not-for-profits and NGOs serving digitally excluded communities face a dual challenge: their clients face AI digital inclusion barriers, and many organisations lack the governance capacity to assess whether their own tools are making exclusion worse.
The evidence on AI digital inclusion: what Australian data shows about who is being excluded
The participation dimension
The gap compounds. Here is how.
From exclusion to invisibility
Digital exclusion does not end at the service access point. When people cannot engage with digital services, they also cannot contribute the feedback, the data, and the community knowledge that would make those services more accurate and more relevant over time.
The other side of this argument
Digital tools and AI, used well, can also reverse this dynamic. Content production at scale, automated translation, chatbot support for after-hours queries, and AI-assisted outreach can all lower barriers and bring excluded communities into contact with services they could not otherwise reach. For under-resourced not-for-profits and NGOs, this is a genuine opportunity. The governance question is not whether to use these tools. It is whether they are deployed in ways that extend access or inadvertently restrict it.
Exclusion from participation produces under-representation in data. Under-representation in data produces AI systems that perform less accurately for excluded communities. Less accurate AI produces worse outcomes. Worse outcomes compound the original disadvantage.
This is the mechanism that makes AI-driven digital exclusion qualitatively different from previous waves. Previous exclusion was additive: each excluded community missed out on a service. AI-driven exclusion is recursive: each excluded community shapes the next iteration of the system in ways that entrench their exclusion further.
The Closing the Gap framework recognises this dynamic. Target 17 sets a goal of equal digital inclusion for First Nations Australians by 2026. Progress is being made. But the deployment of AI into service delivery is outrunning that progress in some contexts, adding new layers of exclusion before the foundational ones have been resolved.
The design dimension
The International Association for Public Participation establishes that communities have a right to participate in decisions that affect them. As services and the systems governing them become AI-mediated, that participation right is at risk. Not through hostility, but through design.
Most AI systems deployed in service delivery contexts were designed without meaningful involvement from the communities they serve. The data used to train them reflects historical service patterns that encoded existing exclusions. Communities least able to participate in design processes are the communities whose knowledge is most absent from the systems that will shape their access to services.
For not-for-profits and NGOs working in communities with high rates of digital exclusion, the governance question is not only about the AI tools they adopt. It is about whether they have a role in advocating for the design of systems that will shape their clients’ lives.
Epistemic Justice is the relevant concept in the Not For Humans Behavioural Foundations Lexicon: whether the knowledge of affected communities is treated as credible and valid in the systems that govern their access to services. When community knowledge is absent from AI design, the resulting systems do not just fail to serve those communities. They actively misrepresent them. Read the Epistemic Justice entry in the Not For Humans Lexicon.
AI digital inclusion and what it means for for-purpose organisations serving excluded communities
AI digital inclusion is already a measurable problem in Australian communities. Not For Humans documents the AI digital inclusion gap through the lens of community organisations closest to the people being excluded. AI digital inclusion failures are not primarily technical. They are the result of systems built without the communities they affect, deployed into service environments where AI digital inclusion has not been assessed. The Australian Digital Inclusion Index confirms that AI digital inclusion gaps track closely with existing socioeconomic disadvantage. Not-for-profits and NGOs serving digitally excluded communities face a dual challenge: their clients face AI digital inclusion barriers, and many organisations lack the governance capacity to assess whether their own tools are making exclusion worse.
Who bears the heaviest cost
Populations facing compounded disadvantage.
Digital exclusion and AI exclusion reinforce each other. For the communities already furthest behind, the gap is not narrowing at the pace that AI is advancing.
The intersection of the highest rates of digital exclusion and the highest risk from AI systems not trained on their communities’ data. A 10.5 point national digital inclusion gap that more than doubles in remote communities. First Nations communities are both the most excluded from AI-mediated services and the least represented in the data those systems are trained on. The Closing the Gap Target 17 framework sets a goal of equal digital inclusion by 2026. AI deployment is in some contexts outrunning that progress.
Australian Digital Inclusion Index, 2025Two thirds (66.5%) of Australians aged 75 and over are digitally excluded. This is not primarily a literacy problem. It is a design problem. Systems built without older Australians’ input do not accommodate their access patterns, communication preferences, or life circumstances. As health, aged care, and government services become AI-mediated, the proportion of inaccessible services for this group grows.
Australian Digital Inclusion Index, 2025People with disability are disproportionately affected by digital exclusion through inaccessible websites, lack of assistive technology, and limited digital literacy support. AI adoption frequently outpaces the development of accessible interfaces. The access gap between Australians with and without disability is 6.3 points nationally. When AI tools are adopted without accessibility assessment, that gap widens in practice.
Centre for Accessibility Australia, 2025Both infrastructure barriers (connectivity, affordability) and under-representation in AI adoption conversations. Outside capital cities, the gaps in digital ability and affordability are largest. One in ten Australians (9.7%) rely solely on mobile connectivity. In remote communities, access limitations restrict the kind of sustained digital engagement that AI-mediated services increasingly require.
Australian Digital Inclusion Index, 2025Language barriers, unrecognised qualifications, and AI systems that may not reflect diverse cultural contexts compound each other. The communities most likely to encounter culturally inappropriate AI outputs are frequently the communities with the least recourse to challenge them. AI systems trained predominantly on English-language data perform less accurately for communities communicating in other languages.
63 percent of low-income Australians are digitally excluded, often relying on prepaid mobile as their only access point. Public housing residents face a 45.2 percent exclusion rate. Affordability stress is a documented structural driver. As more services shift online and become AI-mediated, the cost of being disconnected is not stable. It rises with every service that moves beyond the reach of a prepaid mobile connection.
Australian Digital Inclusion Index, 2025
Australian context
The Australian Government’s National AI Plan identifies digital inclusion as a prerequisite for equitable AI benefit. The Senate Select Committee on Adopting AI found that the communities least equipped to navigate the AI transition will bear its heaviest costs. The gap between policy intent and community reality remains significant, and the for-purpose sector is the closest observer of where it falls.
Read the committee reportAI digital inclusion and the communities being left further behind
What it means for organisations
What this means for your organisation.
Digital inclusion is not a background condition. For not-for-profits and NGOs serving communities with high rates of exclusion, every service design decision is an inclusion decision. Tap each point to read more.
If you are seeing this in your community, we want to hear about it.
Join the networkHas your organisation assessed whether its digital or AI tools are accessible to the clients you serve?
Not For Humans is building a picture of how the for-purpose sector is navigating digital inclusion in practice. Your response is anonymous and takes less than a minute. The open field is optional.
The communities most likely to be excluded from AI-mediated services are the communities most dependent on the services they mediate. That is the gap not-for-profits and NGOs are closest to. It is also the gap they are best positioned to name.
Not For Humans: Because AI’s greatest risk isn’t technical, it’s Human.