1. Rangatiratanga | Authority: ...Māori have an inherent right to exercise control over Māori data and Māori data ecosystems. This right includes, but is not limited to, the creation, collection, access, analysis, interpretation, management, security, dissemination, use and reuse of Māori data...Whenever possible, Māori data shall be stored in Aotearoa New Zealand...
2. Whakapapa | Relationships: All data has a whakapapa (genealogy). Accurate metadata should, at minimum, provide information about the provenance of the data, the purpose(s) for its collection, the context of its collection, and the parties involved...Māori data shall be collected and coded using categories that prioritise Māori needs and aspirations...
3. Whanaungatanga | Obligations: ...In some contexts, collective Māori rights will prevail over those of individuals...
4. Kotahitanga | Collective benefit: ...Māori Data Sovereignty requires the development of a Māori workforce to enable the creation, collection, management, security, governance and application of data...
5. Manaakitanga | Reciprocity: ...The collection, use and interpretation of data shall uphold the dignity of Māori communities, groups and individuals. Data analysis that stigmatises or blames Māori can result in collective and individual harm and should be actively avoided...
6. Kaitiakitanga | Guardianship: ...Māori shall decide which Māori data shall be controlled (tapu) or open (noa) access."

❓ Consider:

• How protective do you feel about your personal data? Have you ever been discriminated against for some aspect of your identity? If not, do you think you would take a similar stance to the above principles if efforts to destroy your language and culture were still in recent memory for your community?
• Do you feel any sense of collective privacy? Can you think of an example of when you protected the privacy of your friends, family, or colleagues?

We got excited about the opportunity to fine-tune Whisper to train a bilingual Māori and English model. But the more concerning question, raised by Whisper's ability to transcribe te reo Māori, was where did Whisper get its data? The Whisper model was trained with 1381 hours of te reo Māori and 338 hours of ʻōlelo Hawaiʻi. The paper doesn't explicitly state where this data was taken from, but one can infer the data was scraped from the web. Scraping data from the web is particularly alarming when they don't have the right to use that data or to create derived works from it...

...While individuals within Big Tech might genuinely believe they're supporting minority languages, the reality is that their actions create more opportunities for non-indigenous peoples and corporations than they do for the minority communities themselves. Authoritarian governments, for example, have the power to use models like Whisper to improve their existing technologies, some of which are used for surveilling indigenous groups. They have the means to take a model like Whisper, fine tune it, and build tools to enhance their surveilling...

...If anyone is to profit from te reo Māori it should be Māori and Māori alone, especially considering the fact that non-Māori once sought to make the Māori language extinct. Our position is only Māori should be selling Māori language as a service...Ultimately, it is up to Māori to decide whether Siri should speak Māori. It is up to Hawaiians to decide whether ʻōlelo Hawaiʻi should be on Duolingo. The communities from where the data was collected should decide whether their data should be used and for what. It's called self determination. It is not up to foreign governments or corporations to make key decisions that will affect our communities."

❓ Consider:

• Are you surprised at this reaction to languages being included in Whisper and Duolingo? If you were developing either of these, would you have expected native speakers' reactions to be positive?
• OpenAI's Whisper model was published in 2022. The Māori Data Sovereignty principles in the first reading were published in 2018. Do you think anyone at OpenAI was aware of them? What actions could OpenAI take to be more aware of and more respectful of other cultures and legal rights in other countries?
• What actions could OpenAI have taken in this specific case to respect the Māori Data Sovereignty principles before they trained the model on te reo Māori speech samples? Do you think the principle of Rangatiratanga would mean they would need to localize their training infrastructure to New Zealand? What about the fully trained model?
• The article provides examples of Whisper transcribing te reo Māori speech inaccurately. Do you think this violates the GDPR principle of accuracy? If not, why? Do you think it is unreasonable to expect a machine learning model to be 100% accurate? Should all models be exempt, or is accuracy more important in some cases than others?

<- Back

Next ->