AI has a major challenge when it comes to bias. But what if the technology could be flipped to actually help the preservation of culture and indigenous knowledge?
While digital transformation is essentially reshaping how a business operates using tech, artificial intelligence (AI) is recognised as one of the main technologies that is enabling this transformation to take place.
But while businesses across a wide range of industries have to figure out how to bring digital transformation into their own areas, it’s important to remember the importance of ethics as technology advances.
AI has been often spoken about alongside algorithmic bias and a limited understanding of different cultural contexts.
“We now need to make sure AI-suitable datasets accurately represent the human experience, especially including under-represented groups and traditional societies that are not technologically savvy,” said Davar Ardalan, founder of IVOW AI.
‘Making AI wise is only going to further make humanity thrive’
– DAVAR ARDALAN
Ardalan is a journalist and National Geographic’s executive producer of audio, where she oversees the podcast series Overheard at National Geographic, as well as other narrative storytelling podcasts in production.
As the daughter of an architect and a scholar, she grew up valuing culture and history and tried to ensure her journalism work was always anchored in historical context.
“Looking at the future of automation, I see a gaping hole in AI algorithms that will define our future stories. I created IVOW to champion culturally conscious data strategies across multiple industries, from academia to development and enterprise. IVOW stands for intelligent voices of wisdom.”
She said that while technologists have gathered data and AI algorithms to advance everything from self-driving cars to healthcare, a critical eye is needed to ensure deep learning is inclusive of diverse cultural contexts.
“As children, we learn our history through the stories our family and friends tell us about our community and where we come from. We can take the same approach to teach machines about our heritage, our communities, our myths and our legends. It’s not OK for AI to dictate to us how we think if it doesn’t even know what we think,” she said.
“Remember, even in sci-fi movies, the most sought-after characters are the wise sages! Making AI wise is only going to further make humanity thrive.”
Addressing AI bias
One notable example of bias being introduced in AI came from an MIT image library. Last year, Lero and UCD researcher Abeba Birhane helped uncover racist and misogynistic terms in the database of 80m images. This dataset was used to train AI systems, and could have contaminated them.
“It’s no wonder that there is so much bias,” said Ardalan.
“The datasets that power many AI products and solutions have been informed by mostly male developers from primarily western backgrounds. This means that AI algorithms and datasets are limited in understanding different cultural contexts, which inhibits the effectiveness of businesses and government from expanding into new markets and better serving citizens.”
However, it’s not all doom and gloom. It was these challenges and knowledge gaps that led to Ardalan founding IVOW, in the hope that technology could be the solution and actually help preserve cultural context and indigenous knowledge.
“We are building the foundations of what we call an Indigenous Knowledge Graph (IKG) – a central repository of culturally relevant narratives with tagged metadata related to indigenous traditions. This will be an open-source project,” she said.
“We’re urging the AI community to help fix the bias in its systems by pausing to nurture accurate and diverse training data. In order to build and support intelligent AIs that are more culturally aware, we need to provide a source of data to develop, train and test with.”
Last month, Ardalan spoke at the American Indian Science and Engineering Society (AISES) summit to showcase IVOW’s latest tools, including IKG. While there, she asked the audience to form small teams and gave them cultural images to review and write a caption for, along with seven tags or labels for each image.
“The Native American students and educators were engaged and passionate. When it came to the particular image below, they captioned it ‘ceremonial preparation’ and chose these labels: plants, sage, cedar and sweetgrass, ceremony, prayers, medicine, sacred, cedar, blessing, protection, spiritual.”
“If you put that same photo into a popular image recognition app, it will tell you the image is more than 95pc ice cream and a dessert.”
Ardalan is also working with the Institute of Traditional Psychoethics and Guidance to build a conversational AI to bring the late Islamic and Sufi scholar Dr Laleh Bakhtiar’s body of wisdom to the public.
“Laleh was my beloved mom, a prolific scholar and maverick who taught me so much about cultural expression. She passed away last October,” she said.
“We will create structured data from her letters, speeches, books, notes, videos, audio, podcasts, lectures, interviews and more and preserve Laleh’s work for the ages.”
Women in tech
As well as the work she has done around cultural knowledge, Ardalan is also a passionate advocate for women in tech.
“My journey to tech would not have been possible without pioneering women such as Anousheh Ansari of XPrize, Sepideh Mousakhani of Cooley, Moojan Asghari and Eve Logunova-Parker of Women in AI, and veteran journalist Kee Malesky who believed in me. They have opened doors in more ways than they know.”
For her own part, Ardalan is now a senior adviser to the Women in AI global network of more than 8,000 data scientists and engineers.
In her role with National Geographic, she has also been able to hire two women from diverse backgrounds. “It’s this intentionality to nurture talented women in leadership roles that gives me the greatest purpose,” she said.
“I try to foster an innovative environment while sharing my personal experiences as a mother and stepmother to eight, and grandmother to four. But self-reflection is also key. I admit that I am not always successful and I have at times let my exhaustion get in the way of letting down my children or other women by overextending myself.
“That said, I think it’s important to be transparent about one’s own shortcomings, growth and learnings. I pride myself in balancing professional and personal life and thriving as a mother, grandmother, entrepreneur and executive who’s knocking on 60.”
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.