Indigenous researchers are pressing the clock. Of the world’s 4,000 indigenous languages, one dies every two weeks along with its last speaker. “Within the next five to 10 years, most Native American languages will be lost in the United States,” says Michael Running Wolf, founder of Indigenous in AI, an international community of Native, Aboriginal, and indigenous engineers. says Mr.
Running Wolf has dedicated his career to preventing this loss. He leads First Languages AI Reality, an initiative of the Mira-Québec Institute for Artificial Intelligence. At the institute, researchers build speech recognition models for more than 200 endangered indigenous languages in North America.
But before that, he must overcome a major obstacle. There just aren’t enough Indigenous computer scientist graduates – people who understand the language and culture – to work on language preservation projects. Running Wolf emphasized that Indigenous scientists know to respect the data itself. “The core data we use is not just tweets or social media posts; it is deeply culturally specific information from speakers who may have passed away,” he said. Ta. “We need to ensure that the community maintains a constant relationship with their data.”
Running Wolf said that in his years of artificial intelligence research, he has only met about 12 indigenous North American AI scientists. “Every year, only one or two Indigenous PhDs in AI and computer science graduate,” he said.
Indigenous people make up less than 0.005% of the U.S. technology workforce, earn just 0.4% of bachelor’s degrees in computer science each year, and hold one seat on the board of directors of the top 200 technology companies. In 2022, native-founded companies received just 0.02% of all venture capital funding.
That’s where the few indigenous engineers that actually exist come into play. They train Native American, Alaska Native, and Native Hawaiian computer science students at leading organizations such as First Languages AI Reality, IndigiGenius, Tech Natives, and Wihamble S’a Center for Indigenous AI. Preserving indigenous cultures and languages.
“Traditionally, AI assumes that data is proprietary and can be harmful to Indigenous communities,” Running-Wolf says. “We want to demonstrate that ethical AI protocols can be used to succeed in the mission of ethically reclaiming Indigenous languages.”
Building a pipeline of indigenous technologies
Kayla Kaya is one of dozens of beneficiaries of a tech training program called Tech Natives, an organization of Indigenous women in technology that provides mentorship and recruitment opportunities on college campuses. Kaya, who is studying computer science at Yale University, was inspired to build an AI tool to honor her Native Hawaiian grandmother, with whom she spent summers on Maui. “I realized that many Native Hawaiians don’t have access to technology that many people take for granted,” the 20-year-old said.
She fed the tool Hawaiian Pidgin English, a much-maligned English-based creole language used by many Hawaii residents, and trained it to recognize spoken language. “I wanted to change the narrative of pidgins as ‘low-level languages,'” she said. “I typed in phrases that my grandmother, aunt, and mother used so I could identify them.”
Kaya hopes to turn her work into an app that other Hawaiians can use. “AI and the tech industry have the power to empower or silence marginalized groups like me,” she says. “That’s why Indigenous peoples can and should play a big role in that.”
Researchers say getting more Indigenous people into the tech industry starts with getting them interested at a young age. Held each summer in South Dakota, IndigiGenius’ Lakota AI Code Camp brings together Native teens for three weeks to design apps that document Lakota culture, including sacred plants and everyday Lakota words. . Since its inception in 2022, Code Camp has trained 33 students to contribute to the app, many of whom have returned as instructors or promoted other computer science projects.
To continue technology education throughout the school year, IndigiGenius also launched T3PD to train a group of 20 primarily Native high school teachers across the country to develop culturally relevant computer science courses in their schools. I did. Only 67% of native-born students have access to computer science courses, lower than any other student population, so the organization partners with teachers to provide laptops and computer science classes to students. Masu.
“It’s important to make AI education culturally meaningful to students,” said Andrea Delgado-Olson, executive director of IndigiGenius. “What sets us apart is that we leverage indigenous knowledge to bridge technology and tradition.”
Using AI to preserve other aspects of Indigenous culture
AI is also helping bridge cultural gaps for indigenous peoples beyond language. As a child, Madeline Gupta rarely visited Chippewa land. However, as she grew older, the desire to return to her ancestral land grew stronger. “I felt that I belonged to the land and that my ancestors wanted me to live there,” Mr. Gupta said.
Many were cut off from their roots after the government separated tribal people from their land and families from 1819 to 1969, she said. “There are about 50,000 people in my tribe, but only 2,000 of them live on reservations,” said the 21-year-old Yale University student. “That means thousands of students have never seen their land.”
While in college, Mr. Gupta participated in the Tech Native Program, which proposed an immersive virtual reality experience for Native youth to “visit” traditional lands in the Great Lakes region. After securing funding from the Aspen Institute and Yale School of Medicine, Ms. Gupta traveled to Mackinac Island this summer to film and record tribal elders’ talks, capture 3D spatial video on her cell phone, and record accompanying audio. I recorded the story. She wants to create a virtual reality map of the island that users can click to view videos related to the area.
“We want to reach people who don’t currently feel connected to the land and who don’t know our story,” she said.
In addition to cultural preservation, artists are also using artificial intelligence in their creative work. Suzanne Kite of Bard College’s Wihanble S’a Center for Indigenous AI describes herself as one of the first Native American artists to use machine learning in her art. “My question is simple: how can we apply indigenous ontologies to create ethical art using AI?” she said.
Kite delved into the Lakota dream language and processed the knowledge her family gained from spirits and animals through dreams. She spent three months last year recording all of her dreams and used machine learning to translate them into Lakota women’s geometric language, commonly used in beadwork and quilting. . She also turned her dreams into a graphic score for the American Composers Orchestra. “I try to resist the Western anthropomorphism of AI and instead dig into the very local, grounded, and practical framework of knowledge provided by Native American communities,” Kite said.
Many of these indigenous AI projects are in the early stages, but Running Wolf hopes that programs like his will be obsolete in 10 to 20 years. His dream is to revive dying languages and enable a new generation of native speakers to develop ethical skills. “I hope this technology will be remembered as a product of difficult times,” he said.
