The Case for Swarm A.I.™ technology

Humanity is a tribal species, owing our evolutionary success to our ability to collaborate in social groups.1,2 This said, society has grown so rapidly during modern times that our tribal norms are no longer sufficient to maintain a cooperative stance among interdependent parties.3 This is especially true when dealing with global problems like climate change and geopolitical conflict.  When confronting such issues, we often get mired in entrenchment and stagnation, devolving into inflexible factions that fail to find common ground, even when our collective future depends on it. This is highly concerning but hardly surprising, for our species evolved with the specialized skills, traits, and temperaments necessary to overcome division within small local tribes that congregate in the same place at the same time.4  Those days are long gone, tribal inter-dependency now spanning continents and time zones, leaving us evolutionarily unequipped to cooperate effectively in modern groups.

Even among small groups, our networked lifestyle is now such that we rarely congregate in the same place at the same time, decisions often being made via emails and texts.  Among larger groups, forums and blogs are commonly used to promote opposing views, with conclusions being based on drive-by inputs such as “likes” and “stars” and “up-votes”.  The problem is, these asynchronous modes don’t easily leverage our natural capacity for compromise and consensus-building.  businessmen shoutingAfter all, it’s quite difficult to explore sets of options and find common-ground when the participants pop in for brief moments, clicking “thumbs-up” or “five-stars” and then move on to something else.  The result is a growing societal trend towards division, entrenchment, and stagnation.

The fact is, polls are polarizing.  And yet poll-focused sites like Reddit, Yelp, and Digg are growing so popular, they’re influencing how we make decisions in a networked society.  This goes beyond explicit decisions, like whether to raise taxes or allow fracking, to implicit social views, like what new book, album, or movie we collectively believe is worthy of high praise.  Research shows that asynchronous polling leads to highly distorted outcomes. This is because the first few individuals to give a rating have a wildly disproportionate impact on the final output.  The process is called “herding” or “snowballing” and reflects the fact that voters in asynchronous rating systems are deeply influenced by the votes that precede their own. In a recent study performed by researchers at MIT and Hebrew University of Jerusalem, it was found that a single up-vote inserted as the first response on a forum like Reddit, boosts the likelihood of future positive ratings by an average of 32% and distorts the outcome by more than 25%.  The researchers warn that these herding effects are distorting our perceived views on everything from product recommendations on Amazon to electoral polling and market predictions.7

In other words, asynchronous polls fail to tap into the true intelligence of groups. Even worse, these methods can do damage, distorting outcomes and manipulating opinions.  In fact, I fear such tools are promoting the mindlessness of the mob rather than the wisdom of the crowd. Should we be surprised?  Probably not, for it’s not a new problem.  In 1895, renowned French sociologist Gustav Le Bon observed in his seminal book “Psychology of Crowds” that “the opinions and beliefs of crowds are specially propagated by contagion… however obvious the absurdity of the triumphant opinion.” 8 The key word here is ‘propagated,’ for the problem is a consequence of asynchronous behavior in which the opinions of a few are passed person-to-person over time, influencing the many without reasoned consensus.  Now 120 years later, social media has enabled the hyper-fast sharing of thoughts and ideas, allowing insignificant preferences to snowball into distorted group sentiments that impact global views.  Does this mean we should give up on the very idea that groups possess collective intelligence?  I think not…

Substantial research has shown that deep wisdom exists within groups and that collective intelligence is a very real phenomenon.3,6, 10  The problem is, current tools don’t tap into this wisdom in a meaningful and constructive way.  Instead, social media combined with asynchronous ratings have produced a perfect storm in which ideas, UNANIMOUS_SLIDE_01opinions, and beliefs spontaneously gain popularity without reflecting the true intelligence of the population.  Thus the critical question is, how do we escape the mindlessness of the mob and tap into the actual wisdom of the crowd?

I contend we must move away from asynchronous polling tools that separate us in time and bias our perceptions. Instead we should employ real-time dynamic systems that close-the-loop around social groups, empowering participants to converge on issues collaboratively and interactively. This paradigm shift is the basis for Swarm A.I. technology, enabling decisions to emerge synchronously from a whole group at once, nobody coming first and nobody coming last, everyone being a simultaneous participant in a unified intelligence.  This mitigates the problem of random biases snowballing into unjustified sentiments.  It also promotes exploration of the decision-space through a dynamic give-and-take that reduces polarization and fosters consensus-building.

We can model this as the difference between “a herd” and “a swarm”. In a herd, a single influential leader can gallop off a cliff, causing thousands of followers to make the same bad decision. And now, with social media driving society, that single influential leader schooling fishisn’t necessarily selected because of wisdom or skill, but  because he or she happened to cast the first up-vote or five-star rating. In a swarm, these biased dynamics don’t happen as easily, for participants contribute in parallel to the collaborative intellect, resulting in a real-time emergent sentiment that better reflects the wisdom of the crowd.  In essence, a swarm is stable closed-loop system that self regulates at every time-step, refining and converging to seek acceptable solutions, while a herd runs open-loop, driving itself to unstable extremes even if the biasing input is little more than insignificant noise.

Still, modern tools for group assessment are overwhelmingly poll-based and asynchronous.  This drives a herd mentality that polarizes populations, distorts decisions, and elevates flawed opinions.  Even more concerning, it encourages entrenchment of extreme views, for when herding takes hold, the majority tends to follow those who came before.  There’s no real means for negotiation.  No give and take.  No finding of common ground.  Although we believe the internet connects us, I fear this particular trend is pulling society apart.

Swarm A.I. technology was developed to help bring us back together, enabling us to leverage our natural instincts for collaboration and compromise when functioning in networked groups.  The basic architecture employs a “synchronous mediation engine” that builds dynamic feedback loops around groups of users, empowering participants to converge on insights by pushing and pulling on each other in real-time, exploring the decision-space and finding common ground.  Unlike asynchronous polls, this synchronous methodology enables users to form an intelligent and interactive swarm that tackles issues simultaneously, unleashing emergent thoughts, opinions, decisions, and ideas. We call the process Social Swarming, for it unlocks the wisdom of social groups by treating the participants as a singular closed-loop system, eliminating leaders and followers so that the true sentiment of the population can emerge naturally.

The world’s first Swarm A.I. system is currently embodied in a platform called UNU™ developed by Unanimous A.I. and currently being field-tested.  The system enables groups of friends, colleagues, or strangers to combine their thoughts and opinions in real-time, forming a Social Swarm can quickly answer questions on any topic.  The interface feels a little like an old-time “spirit board”, users posing questions to the UNU A.I. and then watching as a glowing puck moves under its own power, plucking out answers.  Of course the puck isn’t driven by ghostly intervention, but by something more intriguing – an emergent swarm intelligence that’s born from the group itself.  Every user controls a graphical magnet with which to contribute his or her own personal pull on the puck, but it’s the collective pull of the entire swarm that causes the puck to move.  In this way, everyone has an impact in a simultaneous real-time negotiation that drives the puck across the decision-space, swaying and swerving until it converges on an answer that reflects the current wisdom of the crowd.

Below is an image of an UNU session conducted in 2014. As shown, a group of users collaboratively answered a question posed to them by one member of the group.  The answer was reached in only 19 seconds, the users synchronously pulling on the puck in real-time, guiding it to the decision that was most agreeable to the participants.  This decision may not reflect the opinion of every member of the group.  It may not even reflect to opinion of any member.  But it does reflect the emergent sentiment that was most acceptable to the group as a whole.  The output does not represent an average or tally, as traditional polling yields, but an interactive negotiation wherein the members of the swarm push and pull in real time as the group explores the boundaries of the decision-space and converges on common ground.  The interface requires users to engage the system physically, and answer under time-pressure, both of which have been shown by research to encourage the expression of honest sentiments rather than biases and preconceptions.4,5

What UNU yields is a computer-moderated social swarm that taps the collective intelligence of diverse groups, facilitating a real-time negotiation that produces emergent views, opinions, and insights.  Of course, the temperament and perspectives of each “unu swarm” is dependent on the makeup of the particular group, impacted by its size and the views of its members. In this way, each computer-moderated swarm represents a unique entity – a Swarm A.I. system that can express unique sentiments which are distinct from any member of the group, as well as distinct from other groups.  In other words, each unique collaborative swarm possesses its own emergent personality.

In this way, Swarm A.I. technology facilitates the creation of unique group intellects – collaborative swarms that leverage the contributions of many minds at once to express creative ideas, insights, and conclusions by tapping the wisdom of social groups. Over the last six months, promising studies have been conducted using the UNU system to test the effectiveness of the swarming algorithms., exploring the accuracy of predictions, the quality of decisions, and the usefulness of insights.  Results of these studies will be reported in the coming months.  I believe we’ll see exciting evidence of something profoundly valuable – an emergent and collaborative form of A.I. that instills human sensibilities into a computer-moderated intelligence.


1:  Axelrod R, Hamilton WD (1981) The evolution of cooperation. Science 211:1390–1396.

2: Rand, D. G., Arbesman, S. & Christakis, N. A. Dynamic social networks promote cooperation in experiments with humans. Proc. Natl Acad. Sci. USA 108, 19193–19198 (2011).

3: Greene, Joshua (2013). Moral Tribes: Emotion, Reason, and the Gap Between Us and Them. Penguin Press.

4: Rand, Greene, Nowak, “Spontaneous Giving and Calculated Greed”, Nature, Sept 2012, VOL 489, pg 427 – 430.

5: Gauchou, H. L., Rensink, R. A., & Fels, S. (2012). Expression of nonconscious knowledge via ideomotor actions. Consciousness & Cognition, 21(2), 976-982.

6: Surowiecki, James. The Wisdom of Crowds: Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations. Doubleday, 2004. ISBN 978-0-385-50386-0.

7: Lev Muchnik, Sinan Aral, Sean J. Taylor. Social Influence Bias: A Randomized Experiment. Science, 9 August 2013: Vol. 341 no. 6146 pp. 647-651

8: Le Bon, Gustav. Psychology of Crowds. 1895.


10: Johnson, Steven Berlin. (2001). Emergence: The Connected Lives of Ants, Brains, Cities. Scribner. New York, NY

ABOUT THE AUTHOR: Louis Rosenberg, PhD did his doctoral work at Stanford University in robotics, virtual reality, and human-computer interaction. A pioneer in the field of Augmented Reality, he developed the Virtual Fixtures system for the U.S. Air Force in the early 1990s.  He then founded of the VR company Immersion Corporation (Nasdaq: IMMR).  Currently, Rosenberg is founder of Unanimous A.I. where he pursues his work on human swarming and other methods of Collective Intelligence.

This Post Has One Comment

  1. Pingback: UNU - a Collective Intelligence platform from Unanimous A.I.UNANIMOUS A.I.

Comments are closed.