By Leila Toplic, NLG Tech Task Force, and Michael Tjalve, Microsoft
Artificial Intelligence (AI) has become an integral part of our lives whether we’re aware of it or not. From smart filters keeping our email inboxes organized, to Google search, or Skype speech translation helping us converse in different languages, AI is becoming seamlessly integrated into our lives. And this is just the beginning.
At its core, AI is a set of computational capabilities that can perform tasks we typically associate with human intelligence, such as speaking, seeing, hearing, and understanding how to interpret incoming signals and reason over them. AI is also exceedingly good at processing and learning from data, and it opens up a wide range of opportunities for new experiences and interfaces that even just a few years ago seemed inconceivable.
Today, conversation about AI is overwhelmingly defined by fear, shaped by Hollywood and media. There are concerns about AI taking our jobs, worsening existing inequalities, creating autonomous killing machines, and impacting financial markets. These are valid concerns, and it’s important to have an open discussion around these topics to proactively address potential risks. But there is more to AI than these stories.
In this post, we want to focus on the opportunities for AI to power the development of solutions to some of the toughest problems in the world and help us build a better future for all. This is something we explored at SXSW EDU earlier this month.
We are excited about the potential of AI to help humanity solve its biggest problems. We are also mindful of how the steps we take today — in terms of where we apply AI, who participates in creating it, who gets access to it, and how informed we all are about its impact on our daily lives — will play an important part in shaping the future of humanity.
We shouldn’t assume that this will happen organically, and as a society we need to put the best conditions in place for making sure that we’re ready for a future where AI has an increasingly important impact from which we can all benefit.
Here is what we propose:
AI can help us do good better
There are big problems out there — from refugee crises and poverty to climate change and disease. We are in the midst of the worst refugee crisis since World War II. Every minute, 20 people somewhere in the world become refugees, and over half of these refugees are under the age of 18. They are children with needs that range from basic human rights (shelter, food) to being able to access education, livelihoods, and meaningful opportunities for participation in their communities.
Natural disasters caused by climate change impact all of us. When disasters strike, the ability to act quickly and deliver assistance where it’s needed the most depends on access to accurate and timely information.
At the same time, new technologies are emerging and gaining traction in the commercial sector. Key aspects of AI have matured and are ready for an entirely new generation of applications. We have an opportunity to focus human creativity and available technologies, such as AI, on tackling real issues we face as a humanity, and we can do that today. There’s no reason for the private sector to wait to figure out all the possible use cases for AI for all commercial contexts before starting to engage on humanitarian and social impact use cases. The problems are too urgent, and the innovation cycles are getting shorter.
We argue that we should be able to look at humanitarian use cases for emerging technologies sooner — as soon as they’re out of the lab and validated for the commercial sector. Together, we’re doing exactly that.
Our organizations — NetHope and Microsoft — are working jointly, along with the humanitarian sector and conflict-affected youth, to leverage technology to address the challenge faced by conflict-affected youth, including education, livelihoods, and participation. Refugee youth are some of the most vulnerable people in the world. They have been uprooted from their homes, they are living in challenging conditions, yet they are eager to learn and create a better future for themselves and their communities. AI can power the solutions to some of their needs and aspirations. We’re integrating conversational AI into a chatbot experience aimed at addressing the problems of discovery and access to relevant learning resources. When embedded within humanitarian support programs, a chatbot can help conflict-affected youth learn whenever and wherever it’s convenient for them, based on their individual preferences and needs.
Another opportunity for AI to support humans in doing good better is in the context of mentoring — how might we enable employee volunteers from companies like Salesforce to remotely mentor conflict-affected youth living in host communities like Jordan and Lebanon? AI can help match mentors with mentees based on interests and availability while allowing humanitarian workers to focus on other urgent needs. It can facilitate communication across language boundaries by leveraging real-time speech translation. It can support mentors with relevant assistance resources and mentees with learning resources.
One of the fields in AI — Reinforcement Learning — could be used to prevent famines by simulating crop growth. Tech startups like bons.ai in the U.S. could use crop simulators developed by institutions like the University of Queensland in Australia to train an AI model that could optimize crop yield. AI could also help us diagnose diseases, treat cancer more accurately, or solve global environmental challenges. These are just some of the examples of what’s possible when AI is used in humanitarian, development, and conservancy contexts.
AI should be shaped by diverse voices
The direction of AI cannot and should not be driven by tech companies alone. Active participation from nonprofits, policy makers, academia, as well as end users will help address the value-alignment problem, ensuring that the goals we implement in AI systems are aligned with human values. At the same time, close collaboration between diverse stakeholders, e.g., nonprofits and tech companies, can help stress test technology components in the field and bring back crucial findings on how to improve the technology to meet the needs of all users.
Building advanced AI capabilities and services is complicated and expensive, and today there are only a relatively small number of tech companies with the full range of AI capabilities in house. It’s important that AI capabilities are made available as externally facing services, such as Microsoft’s Cognitive Services, which allow individuals to build their own experiences, applications, and businesses on top of these. Making AI capabilities available for anyone to use is one of the key driving forces in the field of AI for social good.
With diverse users comes a greater variety of preferences and needs, and those who are building AI-based experiences should consider how to integrate that diversity for more inclusive and personalized experiences. For example, Joi Ito (Head of MIT Media Lab) argues that those who don’t follow the traditional model of learning by reading textbooks and instead learn through observing and doing ought to be integrated to enable us to build AI-powered solutions with and for the full diversity of human experience.
AI is the new digital literacy
With AI being integrated steadily into all aspects of our lives, it’s already affecting us in different ways whether we know it or not — from what emails we respond to or what we watch, to how we order household items and about whose perspectives we read.
While not everybody will choose to be an active participant in the development of AI-powered solutions, everybody — including refugees and other vulnerable populations — should have the opportunity to access the benefits of AI and understand how AI-powered products and services might impact them in their daily lives. And while we work together to integrate human rights considerations into AI-based systems, we all need to be aware of the potential negative impact of the products and services we’ve created so far that might be amplifying marginalization and discrimination in the areas of employment, health, credit, and public services.
We believe that in the future, everyone will need to be AI-literate, and we encourage everyone to start learning about it now. Here are a few resources we’re aware of: AI4Al Education Programs, Microsoft Professional Program, OpenAI Research, AI.Google.
Filed Under: No Lost Generation Tech Task Force