1.7: Social Intelligence
Pioneering Socially Intelligent Systems Other Information:
Emerging types of collaboration, communication, and cooperation – from open source software development, crowdsourcing, and
clickworking – illustrate the potential for a new form of “social intelligence” that melds human and computer abilities. Socially
intelligent systems can range in scale from one person and one machine to many people and many machines distributed over the
globe. WeCompute envisions enhanced environments and tools for large-scale collaborative problem-solving in which the intelligence
of large numbers of people, operating in real-time computational environments, can accelerate solutions to the most complex
problems by simultaneously letting humans do what they do best (e.g., deriving meanings from sensory inputs, synthesizing
disparate experiences, drawing inferences) and machines do what they do best (e.g., fast, complex computation). Such environments
would include the far more powerful analytical tools described above to enable people to make effective decisions. The “intelligence”
exhibited in these systems will mimic human capability to reason, perceive the environment, and collaborate with humans and
other machines. An intelligent system will learn from all past experience and adapt over time as well as understand people’s
cognitive and social abilities. The “social” aspects of the system will result from optimizing human actions, interactions,
knowledge, and skills in relation to overall goals. Socially intelligent systems may be designed to act autonomously so that
humans can remain “out of the loop,” as in dynamic allocation of bandwidth or recovery of transportation systems during emergencies.
But humancomputer partnerships allowing for new forms of complementary engagement may be the most effective of all. Where
we are now: Online commerce and communication have been revolutionized through innovations such as micro-blogging, security,
video, recommender and reputation systems, and science has been advanced through cyber-enabled discovery and virtual organizations.
Even so, today’s systems are merely suggestive of the powerful versions of social intelligence that may be developed in the
future. The goal is to advance knowledge at the frontiers of computationally mediated human-machine interactions that would
reframe the meaning of intelligence. Research needs: New findings about how the mind perceives, evaluates, categorizes, synthesizes,
analyzes, retains, and makes decisions about inputs – both internal and from the outside world – will provide researchers
with models for engineering intelligence into the capabilities of networking and computing technologies. Research needs to
focus on computers as intelligent participants whose perception, computational capabilities, and learning may be unique and
able to scale at rates difficult for humans to grasp. At the same time, we also need a better understanding of how people
best coordinate, collaborate, and participate in collective action at such large scales and in real time. One goal is to better
understand the types of problems that are best suited for these types of human-computer partnerships. Key areas of research
include: machine learning and artificial intelligence; immersive environments and 4-D touch displays; understanding, modeling,
and managing complex systems; computational photography; graphics and visualization; social and humanoid robotics; speech
recognition, natural language understanding and dialogue systems; and mechanization of economic theories (e.g., n-way kidney
exchanges). Some research questions include: What is the nature of this collective intelligence? Can we build computational
models of political discourse to better understand each other’s point of view and resolve disputes? Can a distributed collection
of machines and people learn to collect data, analyze them, and ask good research questions, ultimately changing the way science
is conducted? How do we best understand the human capabilities that outperform computers and then harness those assets in
new humancomputer partnerships? How are value systems (i.e., cultural, ethical, legal, etc.) embedded in algorithms and collective
enterprises and how should they be evaluated? What new design techniques and methods would result in a broad array of behaviors
and achievements that could effectively address current social issues (e.g., emergency response)? How can human needs and
values be strengthened through socially intelligent systems? What new theories could explain the behaviors of these complex,
dynamic systems?
Indicator(s):
|
