Conference PaperPDF Available

Human Swarms, a real-time method for collective intelligence

  • Unanimous AI

Abstract and Figures

Much research has been done in the field of collective intelligence to aggregate input from human populations with the goal of amplifying the abilities of the groups. Nearly all prior research follows a similar model where input is collected from human participants in isolation and then aggregated statistically after the fact. The paper introduces a radically different approach in which the human participants is not aggregated statistically, but through a real-time dynamic control in which the participants act, react, and interact as a part of a system modeled after swarms in nature. Early testing of these "human swarms" suggest great potential for amplifying the intelligence of human groups, exceeding traditional aggregation methods. on the simulation of collaborative systems as it relates to the emergence of real-time collective intelligence. While theoretical studies are of great research value, there’s a growing need for real-world platforms that test the emergence of collective intelligence among human users. This short paper introduces such a platform. It enables networks of online collaborators to converge on questions, decisions, and dilemmas in real-time, functioning as a unified dynamic system. The dynamic system has been modeled after biological swarms, which is why refer to the process as “social swarming” or "human swarming". Early testing of human swarms suggests a great potential for harnessing collective intelligence.
Content may be subject to copyright.
Human Swarms, a real-time method for collective intelligence
Louis B. Rosenberg
Unanimous A.I., San Francisco, California, USA
Although substantial research has explored the emergence of
collective intelligence in real-time human-based collaborative
systems, much of this work has focused on rigid scenarios such
as the Prisoner’s Dilemma (PD). (Pinheiro et al., 2012; Santos
et al., 2012). While such work is of great research value, there’s
a growing need for a flexible real-world platform that fosters
collective intelligence in authentic decision-making situations.
This paper introduces a new platform called UNUM that allows
groups of online users to collectively answer questions, make
decisions, and resolve dilemmas by working together in unified
dynamic systems. Modeled after biological swarms, the UNUM
platform enables online groups to work in real-time synchrony,
collaboratively exploring a decision-space and converging on
preferred solutionsin a matter of seconds. We call the process
“social swarming” and early real-world testing suggests it has
great potential for harnessing collective intelligence.
Humanity is a tribal species, owing our evolutionary success
to our ability to collaborate in social groups (Axelrod, 1981;
Rand et al., 2011). This said, modern life has expanded the
scope of human interactions so widely, our tribal norms may
no longer be sufficient to maintain a cooperative stance
among dependent parties (Green, 2013). Even among small
social groups, collaborators rarely congregate in the same
place at the same time, decisions often being made via email
and text. For larger groups, discussion forums are commonly
used for distributed online decisions, with conclusions based
on asynchronous user inputs such as “likes” and “up-votes”.
Unfortunately, asynchronous polling doesn’t leverage our
natural capacity for compromise and consensus-building. In
fact, recent studies suggest that asynchronous polling, as used
by mainstream social media sites and forums, greatly distorts
group-wise decisions by introducing biasing effects known
commonly as herding or snowballing (Muchnik et al., 2013 ).
From Polls to Swarms
As introduced above, there is a growing need for new online
platforms that facilitate collective intelligence and support
collaborative decision-making without employing traditional
asynchronous polling. To address this, we developed UNUM,
a real-time collective intelligence engine that is modeled after
natural biological swarms. UNUM enables groups of users to
answer questions in synchrony, the participants working as a
unified dynamic system through real-time feedback loops.
When using the UNUM platform, swarms of online users
can answer questions and make decisions by collaboratively
moving a graphical puck to select among a set of possible
answers. The puck is generated by a central server and
modeled as a real-world physical system with a defined mass,
damping and friction. Each participant in the swarm connects
to the server and is provided a controllable graphical magnet
that allows the user to freely apply force vectors on the puck
in real time (Fig. 1). The puck moves in response to swarm’s
influence, not based on the input of any individual participant,
but based on a dynamic feedback loop that is closed around all
swarm members. In this way, real-time synchronous control is
enabled across a swarm of distributed networked users.
Figure 1: a human swarm of user-controlled magnets collaborate in
synchrony to move a graphical puck as a unified collective intelligence.
Through the collaborative control of the graphical puck, a
real-time physical negotiation emerges among the members of
the online swarm. This occurs because all of the participating
users are able to push and pull on the puck at the same time,
collectively exploring the decision-space and converging upon
the most agreeable answers. But do the answers have value?
Early Testing and Results
To test the value of human swarms, we enlisted groups of
novice users and asked them to make predictions on verifiable
events: the outcome of the NFL playoffs, the Golden Globes,
and the 2015 Academy Awards. In all cases, the predictions
made by swarms were substantially more accurate than the
predictions made by the individuals who comprised each
swarm. In fact, in all cases the predictions made by swarms
out-performed even the highest performing individual in each
group. The swarms also out-performed the average polling
results across the full population of participants. This suggests
that swarms offer a powerful alternative to the traditional poll-
based methods of harnessing the wisdom of groups.
Louis B. Rosenberg (2015) Human Swarms, a real-time method for collective intelligence. Proceedings of the European
Conference on Artificial Life 2015, pp. 658-659
For example, when predicting the 2015 Academy Awards,
we polled 48 individuals with a written survey, asking them to
predict the top 15 award categories. Using the most popular
predictions to represent the wisdom of the crowd”, the group
collectively achieved 6 correct predictions for the top 15
award categories (40% success). This was our baseline
dataset, the low success rate reflecting the fact that this group
of users had no special knowledge about movies.
To test swarming, we then selected a 7 person sub-group of
the full population and asked them make the same predictions,
but now as a unified dynamic system. The 7 individuals were
typical performers on the written poll, ensuring equity with
the full 48 person population. Each of the 7 individuals were
networked over standard internet connections to a central
server from different remote locations.
Working as a unified swarm, the group of 7 individuals
achieved 11 correct predictions for the top 15 award
categories (73% success). In other words, a sub-group that
was only 15% the size of the full population had a success rate
that was nearly double. We believe this is a highly promising
result and speaks to the potential for harnessing the wisdom of
social groups through real-time swarming.
It should also be noted that real-time swarming is a high-
speed process, all decisions made within 60 seconds or less.
Thus, in addition to improved accuracy of predictions, this
form of collective intelligence is uniquely efficient.
Figure 2: screen-shot of a real-time social swarm in the process of
predicting the Best Actor category of the 2015 Academy Awards.
As a point of reference, experts at the New York Times
made similar predictions for the 2015 Academy Awards.
These experts possessed far deeper knowledge than the novice
members of our study. We assume these experts invested far
more than 60 seconds on each prediction made. Still, the New
York Times only showed a 55% success rate.1 Thus, a group
of 7 novices, functioning as a social swarm, made predictions
that surpassed industry experts. Although not conclusive, this
result suggests that social swarming may provide a means of
achieving expert-level insights from groups of non-experts.
Discussion and Conclusions
Why are swarms better than polls? Our analysis suggest that
while polls are good at characterizing the average views of a
population, without real-time feedback control, polls offer no
means for groups to explore options and find consensus.
Swarms, on the other hand, allow users to continually update
their intent in real-time, assessing how their views combine
with the other participants to achieve an acceptable outcome.
In this way, each participant in a swarm is not expressing a
singular view, but is continually assessing his own personal
conviction across the range of possible options, weighing his
confidence and preference in real-time. With all participants
doing this in synchrony, the swarm quickly converges on
solutions that seem to maximize the collective confidence and
preference of the full group. We believe this is why swarms
are able so efficiently capture the group’s wisdom.
We are currently conducting additional studies to quantify
the effectiveness of social swarms, not just to make accurate
predictions but in facilitating group decisions. Of particular
interest is whether decisions made by real-time swarms are
more or less satisfactory to the participants than decisions
made by traditional polling. Initial results suggest that social
swarms yield more satisfactory decisions than votes or polls.
Finally, to help drive exploration of social swarming, we
have made the UNUM platform accessible to any academics
who wish to run their own user tests. Academic researcher can
request a free account at
Acknowledgements: This study was funded by Unanimous A.I. with
support from California State University (CalPoly, San Luis Obispo).
Axelrod R, Hamilton WD (1981) The evolution of cooperation. Science
Greene, Joshua (2013). Moral Tribes: Emotion, Reason, and the Gap
Between Us and Them. Penguin Press.
Lev Muchnik, Sinan Aral, Sean J. Taylor. Social Influence Bias: A
Randomized Experiment. Science, 9 August 2013: Vol. 341 no.
6146 pp. 647-651
Rand, D. G., Arbesman, S. & Christakis, N. A. (2011) Dynamic social
networks promote cooperation in experiments with humans. Proc.
Natl Acad. Sci. USA 108, 1919319198.
Pinheiro, F. L., Santos, F. C., and Pacheco, J. M. (2012). How selection
pressure changes the nature of social dilemmas in structured
populations. New J. Phys., 14(7):073035.
Santos, F. C., Pinheiro, F. L., Lenaerts, T., and Pacheco, J. M. (2012). The
role of diversity in the evolution of cooperation. J. Theor. Biol.,
Louis B. Rosenberg (2015) Human Swarms, a real-time method for collective intelligence. Proceedings of the European
Conference on Artificial Life 2015, pp. 658-659
... The technology of ASI was first proposed Presented at the Works-in-Progress & Demonstration track, AAAI HCOMP 2023, Nov 6-9, 2023. Copyright by the authors. in 2015 and uses algorithms modeled on the emergent behaviors of schooling fish, flocking birds, and swarming bees (Rosenberg 2015). Biologists refer to these emergent behaviors as Swarm Intelligence because it turns biological groups into "super-organisms" that can make more effective decisions than the individuals can make on their own (Seeley 2010, Seeley 2012. ...
Full-text available
Conversational Swarm Intelligence (CSI) is a new method for enabling large human groups to hold real-time networked conversations using a technique modeled on the dynamics of biological swarms. Through the novel use of conversational agents powered by Large Language Models (LLMs), the CSI structure simultaneously enables local dialog among small deliberative groups and global propagation of conversational content across a larger population. In this way, CSI combines the benefits of small-group deliberative reasoning and large-scale collective intelligence. In this pilot study, participants deliberating in conversational swarms (via text chat) (a) produced 30% more contributions (p<0.05) than participants deliberating in a standard centralized chat room and (b) demonstrated 7.2% less variance in contribution quantity. These results indicate that users contributed more content and participated more evenly when using the CSI structure.
... A variety of methods have been explored over the last hundred years for harnessing groups to drive more precise forecasts, predictions, and decisions (Boland, 1989;De Condorcet, 1785;Malone 2019;Larrick and Soll, 2006). Artificial Swarm Intelligence (ASI) is a relatively new technique that was first proposed by Rosenberg in 2015 as a real-time alternative to traditional collective intelligence methods and is modeled on biological swarms (Rosenberg 2015;Rosenberg 2016). A variety of subsequent studies have shown that ASI can significantly amplify the decision-making accuracy of networked human groups Metcalf et al., 2019;. ...
Conference Paper
The biological phenomenon of Swarm Intelligence (SI) enables social species to converge on group decisions by interacting in real-time systems. Studied in schools of fish, bee swarms, and bird flocks, biologists have shown for decades that SI can greatly amplify group intelligence in natural systems. Artificial Swarm Intelligence (ASI) is a computer-mediated technique developed in 2015 to enable networked human groups to form real-time systems that can deliberate and converge on decisions, predictions, estimations, and prioritizations. A unique combination of real-time HCI methods and AI algorithms, ASI technology (also called “Human Swarming” or “Swarm AI”) has been shown in many studies to amplify group intelligence in forecasting tasks, often enabling small groups of non-professionals to exceed expert level performance. In the current study, small groups of approximately 24 amateur sports fans used an online platform called Swarm to collaboratively make weekly predictions (against the spread) of every football game in four consecutive NFL seasons (2019 - 2022) for a total of 1027 forecasted games. Approximately 5 games per week (as forecast by the human swarm) were identified as “predictable” using statistical heuristics. Performance was compared against the Vegas betting markets and measured against accepted performance benchmarks for professional handicappers. It is well known that professional bettors rarely achieve more than 55% accuracy against the Vegas spread and that top experts in the world rarely exceed 58% accuracy. In this study the amateur sports fans achieved 62.5% accuracy against the spread when connected as real-time “swarms.” A statistical analysis of this result (across 4 NFL seasons) found that swarms outperformed the 55% accuracy benchmark for human experts with significance (p=0.002). These results confirmed for the first time that groups of amateurs, when connected in real-time using ASI, can consistently generate forecasts that exceeded expert level performance with a high degree of statistical certainty.Keywords: Swarm Intelligence, Artificial Swarm Intelligence, Collective Intelligence, Wisdom of Crowds, Hyperswarms
... At the organizational multi-team level, such as when policymakers carry out new policy reforms but tap into different groups' judgments using CI methods, such as swarm intelligence when each participant connects to the server and is allocated a controllable graphical magnet that allows the user to freely apply force vectors on the puck in real-time (Rosenberg, 2015). It could also be citizens' deliberation on participatory platforms, supported by natural language processing (NLP) to sort and categorize the open-text contributions of interested citizens. ...
Full-text available
This handbook aims to synthesize key ideas from the emerging field of collective intelligence and place them more clearly in the context of public problem-solving. Groups, under certain conditions, can solve problems and accomplish complex tasks in a way that surpasses the abilities of any individual within them. How can we understand this phenomenon, and what are the conditions necessary to bring it about? The notion of collective intelligence is polysemic, informed by disciplines as diverse as conflict mediation and computer science. This poses challenges as well as opportunities: Though our notions of cognition, collaboration, and citizen engagement are enriched by studies from a diversity of fields, there is also a risk that these fields talk past one another or that crucial links go unexplored. Our intent in this handbook of collective intelligence for democracy and governance is to add clarity and coherence to the field of collective intelligence as it matures into a scientific discipline in its own right. The increased salience of the field in recent decades may be correlated with some of the key trends of our time, notably the revolution in digital technology that has enabled new modes of largescale collaboration (McNutt et al., 2016), the growing complexity and interdependence of public problems (Noveck, 2015), and a crisis of trust in public institutions that has inspired a “wave” of deliberative experiments worldwide (OECD, 2020).
... The knowledge graph used in the Google search engine contains a large amount of information about the knowledge graph constructed by manual means in the Freebase project [25][26]. Several research works have also shown that the inclusion of human feedback in the automated construction of knowledge graphs can significantly improve the quality of the construction of knowledge graphs [27][28][29][30]. ...
Full-text available
Network culture is a culture based on computer network information technology. With the popularity of “Internet+”, the influence of network culture on college students’ ideological views, behavioral patterns, value orientations and psychological development is also increasing. While Internet culture brings opportunities to higher education ideological and political education, it also makes education work face serious challenges. This paper proposes a collaborative knowledge graph construction method for Civic Education based on Internet + group intelligence. The core of the method is a continuously running loop that contains 3 activities: free exploration, automatic integration, and active feedback. The experimental results show that the knowledge graph fusion algorithm can effectively utilize the structural information of the knowledge graph as well as the semantic information of the nodes to form a high-quality knowledge graph fusion scheme. The collaborative approach based on the “explore-fusion-feedback” loop can improve the scale of group construction of knowledge graphs and the efficiency of individual construction of knowledge graphs, and exhibit better scalability of group scale.
... Artificial Swarm Intelligence (ASI) is a recent real-time technique that's been shown to significantly amplify the decision-making accuracy of networked human groups using intelligence algorithms modeled on biological swarms. Unlike votes, polls, surveys, or prediction markets, which treat each participant as a separable datapoint for statistical processing, the ASI methodology treats each individual as an active member of a real-time dynamic system, enabling the full group to collaboratively converge on solutions as a unified intelligence in real-time [4,5,[9][10][11]15]. ...
Conference Paper
Full-text available
Artificial Swarm Intelligence (ASI) is a powerful method for harnessing the collective intelligence of decentralized human teams and amplifying their decision-making accuracy. Previous studies have shown that ASI collaboration environments, such as the Swarm® software platform, can significantly increase the collaborative accuracy of decentralized groups across a wide range of tasks from forecasting and prioritization to estimation and evaluation. In this paper, we introduce a new ASI method for amplifying group intelligence called the “slider-swarm” and show that networked human groups using this method were 11% more accurate in generating collaborative forecasts as compared to traditional poll-based Wisdom of Crowds (WoC) aggregation methods (p<0.001). These results were generated by groups tasked with forecasting three different real-world events. These results suggest slider-swarms can amplify group forecasting accuracy across a range of authentic forecasting applications.
... Pilot tests of artificial human swarms have demonstrated accurate predictions and estimations, outperforming votes and polls and other traditional methods of harnessing the collective intelligence of groups (Rosenberg 2015). To support the ongoing study of human swarms, the UNU platform is available to researchers who wish to run their own experiments. ...
Full-text available
Most research into Swarm Intelligence explores swarms of autonomous robots or simulated agents. Little work, however, has been done on swarms of networked humans. This paper introduces UNU, an online platform that enables networked users to assemble in real-time swarms and tackle problems as an Artificial Swarm Intelligence (ASI). Modeled after biological swarms, UNU enables large groups of networked users to work together in real-time synchrony, forging a unified dynamic system that can quickly answer questions and make decisions. Early testing suggests that human swarming has significant potential for harnessing the Collective Intelligence (CI) of online groups, often exceeding the natural abilities of individual participants.
Full-text available
Artificial Swarm Intelligence (ASI) is a powerful method for amplifying the collective intelligence of decentralized human teams and quickly improving their decision-making accuracy. Previous studies have shown that ASI tools, such as the Swarm® software platform, can significantly amplify the collaborative accuracy of decentralized groups across a wide range of tasks from forecasting and prioritization to estimation and evaluation. In this paper, we introduce a new ASI method for amplifying group intelligence called the “slider-swarm” and show that networked human groups using this method were 11% more accurate in generating collaborative forecasts as compared to traditional polling-based Wisdom of Crowds (WoC) aggregation methods (p < 0.001). Finally, we show that groups using slider-swarm on three real-world forecasting tasks, including forecasting the winners of the 2022 Academy Awards, produce collective forecasts that are 11% more accurate than a WoC aggregation. These results suggest slider-swarms amplify group forecasting accuracy across a range of real-world forecasting applications.
In the last few years, breakthroughs in computational and experimental techniques have produced several key discoveries in the science of networks and human collective intelligence. This review presents the latest scientific findings from two key fields of research: collective problem-solving and the wisdom of the crowd. I demonstrate the core theoretical tensions separating these research traditions and show how recent findings offer a new synthesis for understanding how network dynamics alter collective intelligence, both positively and negatively. I conclude by highlighting current theoretical problems at the forefront of research on networked collective intelligence, as well as vital public policy challenges that require new research efforts.
Full-text available
Follow the Leader? The Internet has increased the likelihood that our decisions will be influenced by those being made around us. On the one hand, group decision-making can lead to better decisions, but it can also lead to “herding effects” that have resulted in financial disasters. Muchnik et al. (p. 647 ) examined the effect of collective information via a randomized experiment, which involved collaboration with a social news aggregation Web site on which readers could vote and comment on posted comments. Data were collected and analyzed after the Web site administrators arbitrarily voted positively or negatively (or not at all) as the first comment on more than 100,000 posts. False positive entries led to inflated subsequent scores, whereas false negative initial votes had small long-term effects. Both the topic being commented upon and the relationship between the poster and commenter were important. Future efforts will be needed to sort out how to correct for such effects in polls or other collective intelligence systems in order to counter social biases.
Full-text available
When members of a population engage in dyadic interactions reflecting a prisoner’s dilemma game, the evolutionary dynamics depends crucially on the population structure, described by means of graphs and networks. Here, we investigate how selection pressure contributes to change the fate of the population. We find that homogeneous networks, in which individuals share a similar number of neighbors, are very sensitive to selection pressure, whereas strongly heterogeneous networks are more resilient to natural selection, dictating an overall robust evolutionary dynamics of coordination. Between these extremes, a whole plethora of behaviors is predicted, showing how selection pressure can change the nature of dilemmas populations effectively face. We further show how the present results for homogeneous networks bridge the existing gap between analytic predictions obtained in the framework of the pairapproximation from very weak selection and simulation results obtained from strong selection.
Human populations are both highly cooperative and highly organized. Human interactions are not random but rather are structured in social networks. Importantly, ties in these networks often are dynamic, changing in response to the behavior of one's social partners. This dynamic structure permits an important form of conditional action that has been explored theoretically but has received little empirical attention: People can respond to the cooperation and defection of those around them by making or breaking network links. Here, we present experimental evidence of the power of using strategic link formation and dissolution, and the network modification it entails, to stabilize cooperation in sizable groups. Our experiments explore large-scale cooperation, where subjects' cooperative actions are equally beneficial to all those with whom they interact. Consistent with previous research, we find that cooperation decays over time when social networks are shuffled randomly every round or are fixed across all rounds. We also find that, when networks are dynamic but are updated only infrequently, cooperation again fails. However, when subjects can update their network connections frequently, we see a qualitatively different outcome: Cooperation is maintained at a high level through network rewiring. Subjects preferentially break links with defectors and form new links with cooperators, creating an incentive to cooperate and leading to substantial changes in network structure. Our experiments confirm the predictions of a set of evolutionary game theoretic models and demonstrate the important role that dynamic social networks can play in supporting large-scale human cooperation.
Understanding the evolutionary mechanisms that promote and maintain cooperative behavior is recognized as a major theoretical problem where the intricacy increases with the complexity of the participating individuals. This is epitomized by the diverse nature of Human interactions, contexts, preferences and social structures. Here we discuss how social diversity, in several of its flavors, catalyzes cooperative behavior. From the diversity in the number of interactions an individual is involved to differences in the choice of role models and contributions, diversity is shown to significantly increase the chances of cooperation. Individual diversity leads to an overall population dynamics in which the underlying dilemma of cooperation is changed, benefiting the society as whole. In addition, we show how diversity in social contexts can arise from the individual capacity for organizing their social ties. As such, Human diversity, on a grand scale, may be instrumental in shaping us as the most sophisticated cooperative entities on this planet.
Moral Tribes: Emotion, Reason, and the Gap Between Us and Them
  • Joshua Greene
Greene, Joshua (2013). Moral Tribes: Emotion, Reason, and the Gap Between Us and Them. Penguin Press.