What's in a name?

This whole project started with a gap we couldn't ignore. While building our product, we went looking for illustrations and icons that showed people with disabilities in successful roles at work, alongside people without disabilities, and there was almost nothing out there. What did exist mostly seemed to show someone with a disability sitting alone in a dark room, staring sadly into the distance. If there's one thing our participants have told us again and again, it's that they don't want your pity. Empathy over sympathy, every time.

Keep looking and you'll find a handful of disability focused illustration libraries. Very few of them extend that same style to people without disabilities, or to people of different cultures. And there's even less choice again once you're looking for something professional rather than cutesy clip art.

So that's what we did. We're a tiny, early stage startup, and we've invested in a suite of custom illustrations that we're now making freely available to anyone who wants them, open source and at no cost. No pity party in any of them, and every asset is co-designed with the people it represents. It's the small details that matter here: not just someone at work in a wheelchair, but someone at work in a power wheelchair too.

We're genuinely comfortable talking about inclusive language. We spend our days encouraging organisations to focus on respectful language, and to ask the person if they're ever unsure, rather than defaulting to overly positive framing, or avoiding the words altogether. But building this library put us in front of a question we hadn't had to answer before: how do we represent different cultures respectfully, at scale, across hundreds of illustrations? That's what the rest of this page is about, sharing our research and the approach we landed on.

We're learning as we go, and we mean well. If we get something wrong here, please tell us.

Nobody sets out to get this wrong. Most people who hesitate over how to describe someone's background, particularly when it comes to how they look, aren't being careless, they're being careful, and that carefulness is exactly the instinct that makes vague language feel safer than specific language.

For example, take See Me Please's open source Inclusive Images Library. Search tags that describe a photo of a Samoan woman as simply "diverse" mean nobody searching for Pacific representation will ever find her, and we'll never know if we're missing that representation in the first place. The exact gap we observed, a lack of inclusive, person first representation, isn't solved, because no one can find an image that reflects what they're looking for.

Respect and specificity aren't in tension. They're the same instinct, pointed at the same goal: seeing people accurately, and letting them be found. Many people avoid specific terminology for fear of offending someone, and in this case, avoidance is the enemy.

Every country that runs a census has to answer the same question we're answering here, and they've each landed somewhere slightly different. It's worth knowing the landscape before picking a standard, because the differences are the whole story, and the links below are there so you can look at each one yourself and decide what fits your own project, not just take our word for it.

The 9 broad groups and 28 narrow groups, transcribed directly from the ABS's own published structure, exactly as the ABS groups them:

Australia is home base for See Me Please, and the ABS runs the most actively maintained, most rigorously reviewed classification of the lot, revised on a public cycle with a formal consultation process behind every change, most recently in 2025.

We use the full ASCCEG, all 276 detailed cultural and ethnic groups underneath its 9 broad and 28 narrow groups, as our default. That's a deliberate choice for specificity over convenience: the whole point of the Samoan example above is that "Pacific" isn't a specific enough tag to be found by. A broader working set (even ABS's own 9 broad groups) would still only get a searcher as far as "Oceanian," the broad group Samoan sits under. The full 276-group classification is what actually gets a searcher to "Samoan." When a project genuinely only needs the broad picture, the coarser levels are still there inside the same standard, we're not asking anyone to memorise 276 terms, only to have them available when precision is the point.

The classification standards above answer "what are the categories." A separate, equally useful set of resources answers "what's the respectful way to talk about them," which is a different question with a different kind of answer. We keep an eye on a few of these as a secondary check on wording, not as a source of categories:

Respect is a method, not a mood. Adopting an external standard doesn't mean we treat it as gospel. Every category we use still passes a community-review gate before it's used to tag a published asset, consistent with how every other representation value in the library is handled.

If a term ever reads as outdated by the time you read this, tell us. Language earns its keep by staying current, and a page like this one is only doing its job if it's willing to be rewritten again.

For the full taxonomy this facet sits within (subjects, categories, art styles, and the naming grammar used across the library), see the Cohorts reference for how SMP's six testing cohorts relate to the accessibility categories used throughout the platform.

Here's the part that's easy to miss: precise, standardised terminology isn't just the respectful choice, it's also the discoverable one. Someone searching "Aboriginal and Torres Strait Islander illustration" or asking an AI assistant to find "accessible design imagery representing South Asian users" is searching in the exact vocabulary a real classification standard uses. A generic tag like "diverse person" matches nothing specific, ranks for nothing specific, and gets recommended by nothing specific, not a search engine, and increasingly, not an AI answer engine trying to match a query to the right source either. The words a census bureau chose after years of public consultation tend to be the same words real people type into a search bar, which makes "use the standard, precisely" a rare case where the accessible choice and the SEO choice are the same choice.

The classification standards cited throughout this page, direct from source: