5 Ways We Nurtured Cultural Context in AI in 2020!
The challenges in 2020 made our work at IVOW even more critical.
Happy Holidays from all of us at IVOW! Despite its roadblocks, 2020 was a year of accomplishments as we helped our partners cultivate new data ecologies with a focus on culture. Our aim is to fill the gap of online content for underrepresented groups and cultures –both in availability and bias.
This year has allowed us to grow our relationships in the industry and we are thankful for the contributions of our collaborators and supporters — especially Women in AI, KiwiTech, WeWork Labs, IBM Netherlands, Microsoft, Topcoder, XPrize, ITU, and Cooley LLP. Each shares our passion to make AI and data more culturally aware and inclusive.
One of the largest gaps in designing for new frontiers in AI is the lack of knowledge of social sciences and deep learning from our heritage and traditions. At IVOW, we are focusing on creating better data ecologies that account for culture and history.
Here are our top 5 accomplishments in 2020!
1. Indigenous Knowledge Graph: Data + Culture
We believe the digital transformation of cultural knowledge into AI-suitable datasets is an important step in helping future machines like our AI Sina Storyteller (a prototype is available on Google Home) become deeply aware of global cultures. With this in mind, we created an Indigenous Knowledge Graph (IKG) to demonstrate how creating structure around stories can foster reasoning and cultural intelligence in machines and conversational AIs.
The focus for our demo IKG was food knowledge of indigenous peoples. Even a fairly simple traditional recipe that has been passed down through generations represents a collection of components (ingredients, instructions, techniques, tools, occasions), each of which has its own set of aspects relating to origin and tradition that differ from culture to culture.
AI researcher Victor Yarlott is a member of the Crow Tribe of Montana. Together with IVOW and our collaborators Tracy Monteith of the Eastern Band Cherokee and Chamisa Edmo of the Navajo Nation, Victor designed our Indigenous Knowledge Graph in Neo4j. “While it may seem obvious to us, as humans, that all the stories about food we’ve collected are, in fact, about food, machines start from zero knowledge about the world: and so, all the stories and recipes being tied to Sustainable Development Goal 2 (Zero Hunger) give the machine additional information that may impact how it parses that information and where it might look to answer questions.”
The Indigenous Knowledge Graph is an important foundational step towards building more culturally-aware intelligent agents, not only because it gives such agents a broader view of the world, but also because it is a step towards structuring cultural data. Our demo graph was realized through the voice of Sina and presented in June 2020 at the virtual AI for Good Summit. XPrize included our presentation in a subsequent podcast. We also presented our IKG as part of the 2020 Global Online News Association Summit as part of the Emerging Technology Track.
2. Diverse Datasets with Topcoder: Data + Women
We partnered with Topcoder on a project that can have a significant impact on the problem of biased datasets: the Women in History Data Ideation Challenge.
Our goal was to collect, organize, and make easily usable public data sources of profiles of women throughout history, and to suggest how that data could be used to gain new insights for AI products and solutions with a focus on women.
The challenge proved that we need to create more free and open source datasets featuring women in science, engineering, technology, arts and culture; we need to clean-up current datasets and tag women based on their contributions in history and not with generic labels; and we must create machine-ready datasets focused on gender and culture so that AI products and services can be more inclusive. Please read our full report, Shaping AI Systems with Cultural Data.
The Women in History Challenge was nominated for a Topcoder 2020 Cutting Edge Award, which highlights a company using the Topcoder platform for the “project of the year, which is downright cool, edgy, important, and disruptive.”
3. Cultural IQ with NGA: Data + Analytics
NGA is a leading provider of AI and data insights that enable businesses to intelligently automate their back and front-office processes. Their focus has primarily been the finance and banking industry. Through a partnership with IVOW AI, we are bringing a cultural intelligence (CI) lens to NGA’s existing data.
“As a big data and AI fanatic, I see how Cultural Intelligence is what makes us human, and insights gained in the future from AI and Big Data by using a cultural lens will aid Government and Companies to better converse and communicate. This will in turn impact every aspect of people’s lives by showing more authentic empathy and understanding,” according to Mark Germishuys, AI and Big Data Pioneer and CEO at NGA.
IVOW and NGA’s CultureGraph will be a decision engine that is fully automated and helps to configure cultural analytics and contextual information for determining the next course of action at a given moment to make stronger business decisions and more authentic and feasible customer interactions.
In mid-2021, this new analytics offering and lens on culture will support a range of decisioning methods including decision trees, decision tables, reports, dashboards, scorecards, and can also perform quasi-real-time monitoring. Non-technical decision-makers who understand both the data and business goals will be able to create their own custom reports based on various dashboards.
4. Storytelling with Soul Machines: Data + Stories
Conversational chatbots like Siri, Alexa, and Google Home are taking the internet by storm, but many lack cultural context. Our AI Sina Storyteller (SEE-na) combines the power of artificial intelligence with ancient storytelling wisdom to create an interactive user experience that is culturally-specific, narrative-rich, and customizable for global appeal.
Sina provides curated stories on food, recipes, and healthy living. She’s a unique chatbot who brings the perspective of cultural heritage to the culinary world. Interested users can opt in for push notifications to access more detail. Sina’s simple yet innovative design makes the information easy to digest and is ideal for busy Millennials who are looking for an interactive way to explore global cultures through food.
Sina is also now available as a “digital human” thanks to our collaboration with Soul Machines, where we are now members of their Champion group. Working with Soul Machines, we are able to bring a cultural focus to the world of exponential AI. As a messenger of human stories, Sina can be seen with an expressive human face, combining AI technology together with storytelling.
In this way, we believe that automation does not need to be synonymous with the loss of ancient wisdom and cultural knowledge. With Cultural Intelligence, AI can help, rather than hinder, humans in our quest to share our unique stories.
5. NY Academy of Sciences: Data + Collaboration
IVOW was selected as one of a group of communications and AI experts to join a series of roundtables to identify and fill gaps in education, research, and policy with the goal of generating consistent and pragmatic solutions to some of the most important problems now facing society. Our contribution helped the New York Academy of Sciences shape parts of their AI report, especially around communications, data equity, and diversity in algorithms. They reported the results in Collaborative Intelligence; a Blueprint for a New Artificial Intelligence Institute.
We emphasized that content and story decisions cannot be made based on current data ecosystems. In addition to reflecting the highest standards for journalism and ethics, the future of media must also be expansive in its coverage of communities, issues, and ideas.
Many media leaders marching into automation fail to understand that our current data ecology is flawed, racist, and derogatory. That lack of diverse data causes problems. Let’s take the case of MSN.com. In early June, just ten days after Microsoft announced it was laying off human editors and journalists, their newsroom AI made a racial blunder.
As reported by Mashable, “Microsoft’s AI confused two members of the pop band Little Mix, who both happen to be women of color, in a republished story originally reported by The Independent. Then, after being called out by band member Jade Thirlwall for the screwup, the AI then published stories about its own failing.” This came just days after human journalists whose job was to test these AIs had been fired.
More significant is the fact that some of the foundational datasets that form the backbone of computer vision are also flawed. One example: In early July MIT announced that its 80 million tiny images dataset would be taken down immediately.
According to MIT and the authors of the dataset, “Why it is important to withdraw the dataset: biases, offensive and prejudicial images, and derogatory terminology alienates an important part of our community — precisely those that we are making efforts to include. It also contributes to harmful biases in AI systems trained on such data. Additionally, the presence of such prejudicial images hurts efforts to foster a culture of inclusivity in the computer vision community. This is extremely unfortunate and runs counter to the values that we strive to uphold.”
How media organizations can organize to join the AI revolution
Education and Training Units: The next generation of diverse editors and journalists need to be trained on the promise of artificial intelligence and machine learning (AI/ML). Educational institutions can create cross campus learning hubs where teams of data scientists, cognitive scientists, computer scientists, cultural anthropologists, social scientists, artists, and musicians come together to educate journalists and communications students on how to apply AI/ML tools in the context of media.
Data Architect Hubs: Similar to daily editorial meetings, diverse data scientists come together daily or weekly to accelerate AI innovation by making sure their data ecosystem is growing, their taxonomy is relevant and being trained ethically. These hubs are also solving challenging NLP extraction problems and sharing knowledge on how these findings can be applied to other use cases.
Open Source Public Editors: Aggregating a wide source of content that your public editors can not only experience but influence and train to make better. This can be done on an honor system to ensure that people are contributing relevant and accurate information.
Data Storytelling: Data Storytelling is an impactful visual reporting tool. “A machine will win a Pulitzer one day… We can tell the stories hidden in data,” Kris Hammond of the Narrative Science, an AI company specializing in natural language generation, said in The Guardian, April 3, 2016.
Quality Assurance & Testing Units: Journalists and editors can work together with QA teams to make sure that what they are serving the public is not only user friendly from a tech standpoint, but also culturally relevant, respectful, and sensitive.
Looking to the future
Heading into 2021, we plan to expand our work on cultural intelligence in AI with a focus on Sina and our CultureGraph, and to develop Sina beyond a prototype.
We invite you to access the prototype on Google Home by asking for “Sina Storyteller.” Next time you are in your kitchen, ask Sina to tell you a story about the traditional foods of the ancient festival of Diwali; she also knows recipes from the Navajo and Cherokee nations. Or ask her how machine learning can help us keep ancient cultural knowledge alive.
IVOW AI is an early stage startup focusing on cultural intelligence in AI.