2023-05-29
Indigenous Peoples: AI is inherently ridden with algorithm bias and poisoned data
In this article, the Asia Indigenous Peoples Pact (AIPP) summarises the main points that came out of a collective session organised at DRAPAC23 by Forus, ADA - Asia Development Alliance, ECNL, the Asia Pacific Forum on Women, Law and Development (APWLD), the Asia Indigenous Peoples Pact (AIPP) and the Asia Center on the risks and opportunities when it comes to AI and its applications.
Asia Indigenous Peoples Pact (AIPP)
The Asia Indigenous Peoples Pact (AIPP) is a regional organization founded in 1992 by indigenous peoples’ movements. At present, AIPP has 46 members from 14 countries in Asia with 18 indigenous peoples’ national alliances/networks (national formations), 30 local and sub-national organizations.
AI is inherently ridden with algorithm bias and poisoned data that will have massively negative repercussions on Indigenous Peoples including both connected and unconnected or poorly connected communities. Below 5 points that showcase the negative impact AI has on Indigenous Peoples.
1. Economic impact: AI seeks to automate low skill jobs and since many indigenous bear the brunt of poor access to education and high skills, many are employed in the unskilled and disorganized sectors. AI will effectively decimate their access to employment including disguised employment.
2. Exclusionary: Most AI tools such as ChatGPT are grouped into free and premium (paid) services. That leaves out even the ones among the Indigenous Peoples who can access all spectrums of the internet who do have the technical skills or general awareness of AI to access the premium services. Therefore, it is not inclusive of vulnerable groups. Furthermore, the online world is largely ableist and tends to be aimed at the English-speaking world with other mainstream global languages taking over. Indigenous languages are yet to be integrated fully into this eco-system, exacerbating the widespread erasure of languages, cultural diversity, richness, good practices of natural resource management, spiritual world view of Indigenous Peoples since documentation, journalism, research and online resources are written by the non-indigenous, catering to the mainstream gaze. Automated translation services do not include indigenous languages.
3. Negative narratives: Indigenous Peoples are overwhelmingly targeted by structured and strategic disinformation/misinformation as well as State sponsored campaigns against them. Indigenous Peoples activists which include human rights defenders, journalists, climate defenders and all whom the State declares its enemy. AI is designed to further existing harmful narratives on Indigenous Peoples. For example, the search word “terrorist” will show up Indigenous Peoples activists’ names since that is what exists as data and stories online. This has been the case in the Philippines with the red tagging of indigenous activists who are often termed terrorists. Similarly, in Thailand IOs (Information Operations) are targetting Indigenous Peoples to sustain and prolong bullying tactics, land grabbing and imaginative forms of violence on the ground. In India, under Modi’s repressive regime, anyone of critical of the State is branded an anti-national with Adivasi activists called Maoists, the terms are almost interchangeable. Indigenous women are particularly vulnerable to doxing that will certainly increase due to the risks posed by AI.
4. Post pandemic post truth world: Covid 19 made the Digital Divide - i.e lack of devices, data and limited control over discourse - painfully obvious. Access to the internet is a fundamental right to many globally and yet is viewed as a privilege for indigenous communities who are often unconnected or poorly connected to the internet and its eco systems powered by AI and the Metaverse.
5. Invisibility: Non recognition of Indigenous Peoples by the State is one of if not THE reason for the barrier to the self-determination of Indigenous Peoples. According to the United Nations Declaration for the Rights of Indigenous Peoples, “self-identification as indigenous is considered a fundamental criterion.” The Declaration refers to their right to determine their own identity or membership in accordance with their customs and traditions. Lack or little existing data on Indigenous Peoples including challenges posed to indigenous data sovereignty along with biopiracy continue the exploitation of indigenous knowledge and traditions, patented for profit by the non- indigenous. Further, Indigenous Peoples are not at decision making tables in the tech world engaged in policymaking.
The fundamental aspect that was the lodestar of the focus group discussion facilitated by AIPP during the Digital Rights Asia-Pacific 2023 was:
"Do digital rights belong only to the connected or are they inclusive of those who are unconnected or poorly connected as are many of our indigenous brothers and sisters who are overwhelmingly negatively impacted by the virtual world?"
AI alarmingly is designed to widen existing structural inequalities through integrated algorithm bias and poisoned data aimed at sustained invizibalization of Indigenous Peoples. It is therefore crucial to include indigenous voices in order to influence AI policy and practice if we are to retard the rampant abuse of our rights both online and offline.