Timing and cultural circumstances are makers and breakers of successful commercial endeavors. For Dr. Jeff Orkin, co-founder of the new company “Giant Otter Tech”, it seems both elements are in his favor. Gaming and virtual environments have made waves in the last few years, and Orkins’ vision for his company lands right in the cross-section of the exploding artificial intelligence industry and education in a media-driven world.
Giant Otter Tech (how they chose their unique name and mascot can be found on their website) was chosen as one of 128 global finalists to participate in MassChallenge, a renowned startup accelerator supporting early-stage entrepreneurs, in the 2013 accelerator program. Driving the company’s initiatives is a combination of artificial intelligence, natural language processing, crowd sourcing, and data science; put all of these parts together and come out with an array of possibilities for social simulations as a tool to be used across industries.
Dr. Orkin took a bit of a backwards path, a “Benjamin Button” approach as he coins it, to arrive at his current undertaking. As an undergraduate at Tufts University, he studied computer science and art. He then got into game development and animation, and found that he became more interested in what drives the action behind animation, the artificial intelligence bit. Dr. Orkin spent the next 10 years working in the video-gaming industry, leading the creation of systems recognized by the computer gaming industry for artificial intelligence.
One of the things that Dr. Orkin began noticing during his burgeoning career in AI was that it had become very fragmented, with many specialized fields. Often the result was a lack of coming together and producing coherent products that were able to leverage progressing technology in various fields. For example, immersed in the gaming industry, he noticed that not nearly as much progress had been made in language and social interaction as compared to spatial manipulation, the action part of the games. His interest in making progress in language and social interaction was piqued.
Making the decision to return to academia, Dr. Orkin is now a recent graduate of MIT’s Media Lab, where his research leveraged the phenomena of crowdsourcing to amass data on social interaction and natural language, i.e. dialogue, in virtual settings. At MIT, Dr. Orkin developed an end-to-end process for an automated AI character. He recorded 16,000 people playing different roles – that of customer or waiter/waitress – and interacting in restaurant-type settings. This crowd-sourced data was then used to automate an AI waitress, who can interact with humans and say and do 18,000 different things. This accomplishment led to Dr. Orkin’s interest in how these types of simulated interactions could apply commercially to Education and other fields in which human beings learn and evolve through social interaction.
Being in the right place at the right time, Dr. Orkin met and partnered with researcher Geoff Marietta from the Harvard Grad School of Education – they have quite a meeting-of-the-minds story. While Dr. Orkin was actively learning about the media business and searching out avenues for taking his research in a commercial direction, Marietta’s research interests focused in the social processes that drive role-taking and how these processes inhibit or facilitate learning and communication between people across backgrounds. His interest in the virtual environment included exploring and building on research that shows how interacting with gaming characters in a virtual environment has the potential to increase empathy and perspective-taking.
With Dr. Orkin’s success in automating socially-intelligent characters and Marietta’s research background in changing human behavior through virtual social interaction, their uniting created a fertile ground for brainstorming and building real-world applications. Giant Otter Tech has opened its minds and doors to creating virtual learning environments that cross many different fields and leverage the complex milieu of social interactions. Currently working on an anti-bullying program for middle school and high schools students using simulated characters, the team is also looking at how these simulations could apply to language learning; therapy for autism; transfer of knowledge at big corporations; and “tough conversation” training, from medical students speaking with cancer patients and their families to management leadership situations; basically, “anything where you’d like to practice in a safe environment before facing the real thing.”
So asides from the potential for bettering social skills, what other potentials might gaming hold for the future of humanity? Dr. Orkin notes that games easily allow you to collect data, with the ability to record people’s moment-to-moment interactions and decisions; this is invaluable information in creating better simulation of human cognition, abstract concepts, and the use of language in AI. Furthermore, games are a valuable vehicle for crowd-sourcing solutions. This idea is basically the inversion of the traditional “humans-managing-computers”. As Dr. Orkin notes, there’s been a “gradual shift towards the idea that you can just push a button and something will happen, who cares what’s on the other side of that, it might be human or machine, and more and more it’s becoming a mix of the two.”
Crowdsourcing melding humans and AI is a fascinating idea, and one that could revolutionize the way we interact with each other and solve problems. Dr. Orkin references the research-based game Foldit, created and delivered to the public in 2008 by researchers at the University of Washington. More than 100,000 proteins exist in the human body, and while researchers have mostly mastered genetic sequence, the complex shapes into which these proteins fold is still not clear, with computers often failing at bigger and harder protein structures. The idea behind the delivery to the public is to see if human intuition and diverse human brainpower can help construct these complex structures. Playsourcing, or using virtual games that can reach the masses and potentially further scientific research, is a provocative and lucrative idea. Certainly the idea of crowdsourcing for both research and commercial ventures has its valid critiques, such as risk of a lower quality product or contribution with less face-to-face time, but the inherent idea, if engineered in a right-fit context, is powerful.
In response to key considerations for progressing AI research and businesses toward a continual betterment of society, Dr. Orkin brings up three key ideas: transparency, communication, and collaboration. “Whatever can be done to build technology in a way that other people can build on it – I think that’s the key…to be able to stand on the shoulders of others.” This is especially important in academia, where a sense of isolationism is often inadvertently fostered. Finish the research and the paper, but then find a way to get those findings out into the public’s hands. Finding ways to better disperse and integrate information will make a profound impact, furthering the AI industry at a more accelerated pace and in more holistic fashion, opening doors for more quality products and services delivered by visionaries – like those behind the face of the Great Otter.