frontmatter 
title: Sunaŵi
aliases: [Home, Sunawi]
tags: [home]

Sunaŵi

"to think one's thoughts, to ponder, to think in existence" — Owens Valley Paiute

This site is a living document about what responsible AI looks like in endangered-language work. Rather than a single static report, it is a wiki: a growing, cross-referenced collection of examples, terminology, and people, meant to help communities, researchers, and practitioners have productive conversations about this topic.

It grew out of the Sunaŵi workshop (June 2026), but it is built to keep growing well beyond it.

Goals of This Wiki

A central takeaway from the workshop, and the reason this wiki exists, is that conversations about this topic are genuinely difficult, for many reasons. To name just a few:

  • Earned mistrust. Indigenous communities have a long history with extractive research, and researchers often don't understand that history or the concerns it created.
  • Jargon. Technical language gets in the way of real conversation between fields.
  • "AI" has become too broad a term to talk about meaningfully. People picture very different things when they hear it, and much lower-risk technologies (text-to-speech, OCR, and the like) get swept up in criticisms aimed at very different systems.
  • Heterogeneity. Indigenous communities' goals and concerns vary widely. There's no one-size-fits-all answer, and that makes it very difficult to rigorously define what good or bad work looks like.

So, in the spirit of producing guidance that is useful rather than prescriptive, this wiki is example-driven and includes good and not-so-good examples of tools, collaborations, and approaches, along with the upsides and downsides of each. We think this could be an important step toward helping communities, researchers, and practitioners have real conversations about this topic.

A few other ideas that came up and shaped how we approached this:

  • Indigenous language communities don't want technology; they want their languages to thrive. Technology is a means to an end, not the end itself.
  • Data is necessarily a flattening of reality. Computer scientists think a lot about how representative a dataset is, but there's a limit we often don't recognize in ourselves: you can't assess how representative a dataset is of a reality you don't know firsthand. What's left out is invisible unless you know what was there to begin with. For Indigenous language data, that's exactly the gap a strong community relationship can fill, which is why this work must be community-driven.
  • Nearly every major question society is asking about AI shows up here in concrete form: environmental impact, privacy, data sovereignty, consent, ownership, misrepresentation, and misinformation. It's also a space where AI has real potential to do real good. If you want to understand what responsible AI actually looks like in practice, not just in principle, this is one of the best places to look.

How this wiki is organized

  • Examples/: the heart of the wiki: concrete tools, collaborations, and approaches, each with its context, upsides, and downsides.
  • Terminology/: short, plain-language definitions of technical terms, concepts, etc. These are referenced throughout the examples and meant to be a resource for people from all backgrounds to get on the same page about what we're talking about.
  • People/: Sunaŵi workshop participants and contributors to this wiki.
  • About/: the workshop this grew out of, related efforts like the CORE Institute Fellowship, and the process for updating this document.
Cite this
@misc{coleman2026,
  title        = {Sunaŵi: A Living Document on AI & Indigenous Language Revitalization},
  author       = {Coleman, Jared and Coleman, Tainã and Iskarous, Khalil and Krishnamachari, Bhaskar and Nelson, Glenn and Anastasopoulos, Antonios and Andrews, Helema and Ankiel, Ben and Aplin, Chris and Arvizu, Citlali and Barragan, Luis and Bögre Udell, Daniel and Brixey, Jacqueline and Brunson, Lisa Mae and Cabrera, Mónica and Coto Solano, Rolando and Cox, Patrick and Dioniso, John David and Ebrahimi, Abteen and Floca, Melissa and Fountain, Amy and Frey, Ben and Gaeta, Alejandra and Hearrell, Chad and Heaton, Raina and Hildebrandt, Kristine and Huaute, Ray and Kellerstrass, Jordan and Kendall-Bar, Jessica and Krim, Jessica and Kroskrity, Paul and Ku, Tiffany Carmen and Lockwood, Hunter and Maillet, Kiana and Munro, Pamela and Murph, Nicole and Prud'hommeaux, Emily and Ramonetti, Pedro and Rosenthal, Nicolas and Scarcliff, Brent and Sullivan, Mairead and Thakur, Jay Prakash and Ticona, Belu and Toal, Kira and Toal, Ray and Torrence, Harold and Wilson, Katie and Wright, Tiffany},
  year         = {2026},
  howpublished = {Sunaŵi},
  url          = {https://sunawi.kubishi.com},
  note         = {Accessed 2026-06-18},
}
Created · Updated
Supported By the National Science Foundation Award 2542375.