When it comes to social or environmental issues, far too many people are content to “like” a Facebook page devoted to it and feel content with that being the extent of their effort. “[People] talk about it, but nothing ever happens, nothing ever gets solved,” says Raymond Ferrero, an attorney who focuses on crisis intervention in mental health and substance abuse in his paid and pro bono work. “They move to the next trend,” adds Rob Hust, a software developer who works in artificial intelligence (AI).
To offer a more effective alternative, Ferrero and Hust have developed a new online network for social causes called the Knowledge Ecology Engine (KEE). Prior to the launch, Fast Company got an exclusive first look at the project, which uses AI to match up not only people, but also information and even objects like physical tools for tackling challenges such as drug addiction, environmental protection, and economic development.
As Facebook or LinkedIn recommend friends based on someone’s social circles, KEE will recommend people based on complementary expertise or prior work on similar projects, such as education programs for prison inmates (something Ferrero is working on). It can also introduce individuals who work on the same issues in the same geographic area but may not know each other. And as Netflix suggests TV shows and Amazon suggests books, KEE will suggest things like research papers or databases that can help professionals and activists get their job done.
KEE isn’t simply built on the hope that people will gravitate to a neat idea. The network already has sections for more than 100 cause areas, from well-defined ones such as cancer, clean energy, and GMOs to broader concepts like freedom, inequality, and peace. KEE is also recruiting veterans in each field to lead many of those areas and bring in others from their professional networks. For example, Florida Judge Ginger Lerner-Wren coordinates the criminal justice and mental health sections. In 1977, Wren created the first mental health court in the U.S., which seeks to direct mentally ill offenders into treatment programs rather than prisons. (It served as the model for a federal program started in 2000.)
KEE will accept members until it reaches 10,000 people, then go invitation-only, with an ultimate cap of 1 million participants. “This is a community that we want to populate with passionate people who are problem solvers, people who get things done,” says Ferrero. If a network grows too large, it loses focus, says Hust: “Instead of a tool, it becomes a pastime.”
KEE grew out of work in 2012 to combat the drug abuse epidemic in South Florida, where Ferrero was a board member of the Broward County United Way Commission Against Substance Abuse. “[We] tended to experience these drug epidemics that would sweep through the community as these new drugs would enter,” he says.
Fellow board member and epidemiologist James Hall developed models to successfully track and even predict drug trends for targeting at-risk communities, says Ferrero. Hall has been frequently quoted by publications like the New York Times and the Wall Street Journal in articles about the drug epidemic, which helped catch the attention of the Organization of American States (OAS), the association of 35 North, Central, and South American nations for international collaboration. The OAS wanted to apply the United Way’s tools for tracking how drug epidemics break out in their own communities. “South Florida shares the problems of Latin America and the Caribbean in regard to trafficking issues,” Ferrero wrote in an email.
“That really was the impetus for the KEE community,” says Ferrero. “What we realized was that we needed a better and more sophisticated way for these levels of professionals to engage each other and interact.”
Language was one of the stumbling blocks, and KEE incorporates a basic “machine translation” engine similar to Google Translate to give a rough, instant rendering of text. “If it’s something of value, the contributors to the community can quickly change it,” says Hust, meaning that people proficient in the languages can look over and refine the translations.
Another challenge was developing an online system that’s accessible even to people with very limited technical resources. “When you are talking about dealing with another culture, you can’t automatically assume that the infrastructure they have over there is anything like the infrastructure we have here,” says Ferrero. “That’s why the cloud-computing aspect of it became so important.”
All the heavy work in KEE, like the translation and artificial intelligence, is handled on the back end, while the users need only minimal resources—just a web browser to open pages that are optimized for low bandwidth. “If you are in sub-Saharan Africa, and you happen to get some Wi-Fi for half a second, you can actually get some work done,” says Hust.
KEE offers more sophisticated tools such as video for people who have better bandwidth, but the initial barrier to entry is set very low. Networks like Google’s new balloon-based Project Loon may eventually bring high-speed Internet access to more of the world, but that will take years to go widespread, even under the rosiest predictions.
Hust and Ferrero claim to be applying a novel form of AI, one that Hust developed for his business-intelligence startup, Prefrent. KEE was coming together at the same time, and Hust says that he applied some lessons from KEE in developing Prefrent.
The difference, according to Hust, is that traditional AI relies on text processing and ontologies—a structured tree of terms. “When you get into a community where targets are moving faster, they break down,” says Hust, “especially when you’re talking in terms of a multilingual community.”
Hust claims to use a different, “signal-processing,” approach. Any potential resource, be it a PhD expert, a report, a tree-planting project, or even shovels that can be used in planting those trees, is recognized as an object with certain properties—knowledge, types of tasks it can perform, etc. KEE looks at those properties and suggests any objects that might interest a member of the network. “They’ll pop up on someone else’s screen and say, this is pertinent to what you’re doing, it has a high value for the effort that you’re in, so you might want to either join that conversation or contribute, or simply download the document.”
Most of these matches are easy for the AI to make, according to Hust. Some are less obvious to the algorithms, but they can learn from the matching of resources that the human experts initiate. “As we’re successful, as is the nature of AI, we’re storing those successes,” says Hust, “and that’s contributing to likelihoods of how that information might be treated in the future.”
KEE is a for-profit company, but it will not feature advertising. Ferrero instead hopes to fund it through membership subscriptions and corporate sponsorship. “Corporations can show their logo and talk about their support, but there is no direct solicitation,” Ferrero says in an email. Hust describes his involvement as partly pro bono. “Essentially we’re working for way below our cost,” he says.
The two have been friends and social activist collaborators for years, and Hust says that he was drawn in by the promise of using artificial intelligence and social networks for more than commercial gimmicks.
“There’s a million AI projects out there right now, whether it’s to make a prettier website, whether its to dig into big-data stores, but it’s all kind of Wizard of Oz stuff,” says Hust. “We haven’t solved these problems we’ve been working on for hundreds of years. And the Internet can be used to actually solve these. Instead, we keep creating flash mobs about them.”