Collective Intelligence

Background: Collective self-organising behaviour is ubiquitous in nature, whereby complex patterns emerge from the local interactions between individuals. Yet in humans, most group behaviour is attributed to explicit central control or pre-planning, rather than to the synergistic decentralised interplay between individuals.

Purpose: This paper introduces Synch.Live, a novel experimental paradigm for quantitatively studying collective motion in humans, framed as a game with an unspecified task and a group feedback mechanism that can be solved through cooperation by 10 participants walking together. We also present the results of a pilot study demonstrating its use.

Study sample: 10 partitipants took part in each trial, forming 16 independent groups.

Data collection and/or Analysis: We collected psychometrics, behavioural and trahectory data in order to explore the participants’ state of mind and the strategies they developed collectively to solve the group challenge.

Results: We found that more than half of the groups participating in the study self-organised to achieve collective flocking motion, and winning players showed higher connectedness to others compared to those who failed. Furthermore, individuals with an awareness of working strategies were more likely to be part of winning groups, suggesting the importance of individual contributions to the collective task.

Conclusion: This work demonstrates that solving an unspecified group challenge in response to group feedback is possible, and moreover, that flock-like collective movement has the potential to yield social benefits and well-being, suggesting new directions for exploring social aspects of consciousness and cognition.

Intelligence Everywhere is predicated on the seamless integration of INTERNET of Things (IoT) networks transporting a vast amount of data streams through many computing resources across an edge-to-cloud continuum, relying on the orchestration of distributed machine learning models. The result is an interconnected and collective intelligent ecosystem where devices, systems, services, and users work together to support IoT applications. This paper discusses the state-of-the-art research and the principles of the Intelligence Everywhere framework for enhancing IoT applications in vertical sectors such as Digital Health, Infrastructure, and Transportation/Mobility in the context of intelligent society (Society 5.0). It also introduces a novel perspective for the development of horizontal IoT applications, capable of running across various IoT networks while fostering collective intelligence across diverse sectors. Finally, this paper provides comprehensive insights into the challenges and opportunities for harnessing collective knowledge from real-time insights, leading to optimised processes and better overall collaboration across different IoT sectors.

In many domains, imitating others’ behaviour can help individuals to solve problems that would be too difficult or too complex for the individuals. In collective decision making tasks, people have been shown to use confidence as a means to communicate the uncertainty surrounding internal noisy estimates. Here, we show that confidence alignment, namely, shifting average confidence between dyad members towards each other, naturally emerges when interacting with others’ opinions. This alignment has a measurable impact on group performance as well as the accuracy of individual members following information exchange. It is suggested that confidence alignment arises among individuals from the necessity of minimising confidence variation arising from task-unrelated variables (trait confidence), while at the same time maximising variation arising from stimulus characteristics (state confidence).

In addition to radical reduction of greenhouse gas emissions, large-scale removal of anthropogenic CO₂ will be required to mitigate the impacts of global warming. However, many greenhouse gas removal (GGR) methods remain at a nascent stage of development. A case study was carried out on the application of the collective intelligence (CI) model to the UK GGR sector and the need to scale up and accelerate development in an economically, socially and environmentally sustainable way. Through systems mapping, thematic analysis, workshops and semi-structured interviews, a rich dataset was formed on the existing level of and potential for CI within the UK GGR ecosystem. It was found that implementing CI thinking could address the need for increasing the visibility of the system and its workings to participants and the public. This would contribute to the formation of coherent, shared vision for the role of GGR in the UK’s net zero strategy. It is proposed that these risks could be mitigated by creating a publicly accessible ‘commons’ to visualise the UK GGR ecosystem process, dynamics, components, and goals, allowing innovation policy to be more responsive to innovator and net zero policy needs. Increasing ecosystem awareness could lay the foundations for sharing of information, promotion of a more collective culture, and increased transparency and accountability, all of which are critical building blocks in establishing a robust GGR sector for the future.

Background: Models of collective decision-making typically assume that individuals sample information independently and decide instantaneously. In most natural and sociological settings, however, decisions occur over some timescale in which group members gather information—often from multiple sources. Information sources may persist for varying lengths of time or be viewed concurrently and identically by multiple group members. These tendencies introduce spatio-temporal correlations in gathered information with poorly understood consequences.

Research Design: Here, we develop a collective decision-making model in which individuals’ access and switch between two conflicting cues that differ in their spatio-temporal properties.

Results: Our model reveals that spatially and temporally correlated cues can profoundly affect collective decisions. Specifically, we observe that spatially correlated cues are dominant when individuals rarely switch between sources of information. Temporally correlated cues, on the other hand, have the strongest impact when individuals frequently switch between information sources. We also discuss how much the usage of independent information must be increased to counter the impact of correlation.

Conclusions: The present model represents a first step toward more accurately capturing the complex mechanisms underlying collective decision-making in natural systems and reveals multiple ways in which the properties of environmental cues can impact collective behavior.

This short piece shares thoughts on some recent research and books related to collective intelligence - on topics ranging from democracy and institutions to LLMs and animals.

Designing interaction techniques for end-users often involves exploring vast design spaces while balancing many objectives. Bayesian optimization offers a principled human-in-the-loop method for selecting designs for evaluation to efficiently explore such design spaces. To date, the application of Bayesian optimization in a human-in-the-loop setting has largely been restricted to optimization, or customization, of interaction techniques for individual user needs. In practice, interaction techniques are typically designed for a target population or group of users, with the goal is to produce a design that works well for most users. To accommodate this common use case in interaction technique design, we introduce two practical approaches that facilitate multi-objective Bayesian optimization at the group level. Specifically, our approaches streamline the process of (1) deriving designs suitable for a group of users from data collected in individual user evaluations; and (2) deriving an initialization from group data to improve the efficiency of design optimization for new users. We demonstrate the advantages of these practical approaches in two multi-phase user studies involving the design of non-trivial interaction techniques.

Digitally enabled means for judgment aggregation have renewed interest in “wisdom of the crowd” effects and kick-started collective intelligence design as an emerging field in the cognitive and computational sciences. A keenly debated question here is whether social influence helps or hinders collective accuracy on estimation tasks, with recent results on the role of network structure hinting at a reconciliation of seemingly contradictory past results. Yet, despite a growing body of literature linking social network structure and collective accuracy, strategies for exploiting network structure to harness crowd wisdom are underexplored. We introduce one such strategy: rewiring algorithms that dynamically manipulate the structure of communicating social networks. Through agent-based simulations and an online multiplayer experiment, we provide a proof of concept showing how rewiring algorithms can increase the accuracy of collective estimations—even in the absence of knowledge of the ground truth. However, we also find that the algorithms’ effects are contingent on the distribution of estimates initially held by individuals before communication occurs.

Scientists must choose which among many experiments to perform. We study the epistemic success of experimental choice strategies proposed by philosophers of science or executed by scientists themselves. We develop a multi-agent model of the scientific process that jointly formalizes its core aspects: active experimentation, theorizing, and social learning. We find that agents who choose new experiments at random develop the most informative and predictive theories of the world. The agents aiming to confirm, falsify theories, or resolve theoretical disagreements end up with an illusion of epistemic success: they develop promising accounts for the data they collected, while misrepresenting the ground truth that they intended to learn about. Agents experimenting in these theory-motivated ways acquire less diverse or less representative samples from the ground truth that also turn out to be easier to account for. Random data collection, on the other hand, combines virtues of diverse and representative sampling from a target scientific domain which enables cumulative development of the successful theoretical accounts of it. We suggest that randomization, already a gold standard within experiments, is also beneficial at the level of experiments themselves.

Humans and other intelligent agents often rely on collective decision making based on an intuition that groups outperform individuals. However, at present, we lack a complete theoretical understanding of when groups perform better. Here, we examine performance in collective decision making in the context of a real-world citizen science task environment in which individuals with manipulated differences in task-relevant training collaborated. We find 1) dyads gradually improve in performance but do not experience a collective benefit compared to individuals in most situations; 2) the cost of coordination to efficiency and speed that results when switching to a dyadic context after training individually is consistently larger than the leverage of having a partner, even if they are expertly trained in that task; and 3) on the most complex tasks having an additional expert in the dyad who is adequately trained improves accuracy. These findings highlight that the extent of training received by an individual, the complexity of the task at hand, and the desired performance indicator are all critical factors that need to be accounted for when weighing up the benefits of collective decision making.

Decision-making in cultural consumption often hinges on social signals, yet the relative power of anonymous mass behaviour versus local opinion leaders remains unclear. The article presents a controlled lab. experiment that attempted to measure, in real time and in a network setting, the degree to which different mechanisms of influence – expert preferences, local networked opinion leaders’ preferences, and collective social dynamics – shape decision-making in the selection of music, as a case of cultural consumption. The results of the experiment provide evidence that decentralised, impersonal social aggregation can outweigh the influence of local influencers in shaping cultural preferences. The article contributes to research on collective intelligence by demonstrating that mass-level signals can more reliably catalyse coordination around cultural goods than top-down endorsements, with implications for digital platform design and early-warning trend detection.

Evolution of beliefs of a society are a product of interactions between people (horizontal transmission) in the society over generations (vertical transmission). Researchers have studied both horizontal and vertical transmission separately. Extending prior work, we propose a new theoretical framework which allows application of tools from Markov chain theory to the analysis of belief evolution via horizontal and vertical transmission. We analyze three cases: static network, randomly changing network, and homophily-based dynamic network. Whereas the former two assume network structure is independent of beliefs, the latter assumes that people tend to communicate with those who have similar beliefs. We prove under general conditions that both static and randomly changing networks converge to a single set of beliefs among all individuals along with the rate of convergence. We prove that homophily-based network structures do not in general converge to a single set of beliefs shared by all and prove lower bounds on the number of different limiting beliefs as a function of initial beliefs. We conclude by discussing implications for prior theories and directions for future work.

An analogy is drawn between the study of human behavior and the study of plutonium to demonstrate that soft and hard sciences are more similar than different, making the distinction moot and unproductive. The studies of human behavior and plutonium follow a common scientific research cycle that aligns with Thomas Kuhn’s views of scientific change. This common research cycle provides evidence that the thought processes and methodologies required for success are congruent in the soft and hard sciences. The primary implication from this analogy is that scientists in all disciplines should eradicate the distinction between soft and hard sciences. Focusing on similarities rather than differences among researchers from different disciplines is necessary to enhance collective intelligence and the type of transdisciplinary collaboration required to tackle difficult sociotechnical problems.

CCS Concepts: • Social and professional topics • User characteristics • Cultural characteristics.

Society is characterized by the presence of a variety of social norms: collective patterns of sanctioning that can prevent miscoordination and free-riding. Inspired by this, we aim to construct learning dynamics where potentially beneficial social norms can emerge. Since social norms are underpinned by sanctioning, we introduce a training regime where agents can access all sanctioning events but learning is otherwise decentralized. This setting is technologically interesting because sanctioning events may be the only available public signal in decentralized multi-agent systems where reward or policy-sharing is infeasible or undesirable. To achieve collective action in this setting, we construct an agent architecture containing a classifier module that categorizes observed behaviors as approved or disapproved, and a motivation to punish in accord with the group. We show that social norms emerge in multi-agent systems containing this agent and investigate the conditions under which this helps them achieve socially beneficial outcomes.

Previous efforts to support creative problem-solving have included (a) techniques such as brainstorming and design thinking to stimulate creative ideas, and (b) software tools to record and share these ideas. Now, generative AI technologies can suggest new ideas that might never have occurred to the users, and users can then select from these ideas or use them to stimulate even more ideas. To explore these possibilities, we developed a system called Supermind Ideator that uses a large language model (LLM) and adds prompts, fine tuning, and a specialized user interface in order to help users reformulate their problem statements and generate possible solutions. This provides scaffolding to guide users through a set of creative problem-solving techniques, including some techniques specifically intended to help generate innovative ideas about designing groups of people and/or computers (“superminds”). In an experimental study, we found that people using Supermind Ideator generated significantly more innovative ideas than those generated by people using ChatGPT or people working alone. Thus our results suggest that the benefits of using LLMs for creative problem-solving can be substantially enhanced by scaffolding designed specifically for this purpose.

The global research community responded with speed and at scale to the emergence of COVID-19, with around 4.6% of all research outputs in 2020 related to the pandemic. That share almost doubled through 2021, to reach 8.6% of research outputs. This reflects a dramatic mobilisation of global collective intelligence in the face of a crisis. It also raises fundamental questions about the funding, organisation and operation of research. In this Perspective article, we present data that suggests that COVID-19 research reflects the characteristics of the underlying networks from which it emerged, and on which it built. The infrastructures on which COVID-19 research has relied – including highly skilled, flexible research capacity and collaborative networks – predated the pandemic, and are the product of sustained, long-term investment. As such, we argue that COVID-19 research should not be viewed as a distinct field, or one-off response to a specific crisis, but as a ‘pandemic veneer’ layered on top of longstanding interdisciplinary networks, capabilities and structures. These infrastructures of collective intelligence need to be better understood, valued and sustained as crucial elements of future pandemic or crisis response.

This paper explores the notion of complementarity in the modeling of social systems. Complementary variables in physics exhibit a duality, and similar dualities are found in the analysis of social, political, cultural, and neurophysiological structures. In this paper, I show that the management of uncertainty in a hierarchical model exhibits certain dualities that closely parallel those observed in social systems. I argue that, due to the necessity of coordination within collectives, agents will mirror each other to a large extent, and will develop heuristics and tactics that are aligned with how each agent is managing uncertainty dualities. I connect sociological literature on social structures with neuroscientific, economic, and psychological developments, and show a simplified derivation of these connections from free energy principles. I then explore how these connections can be used to gain insights into observable socio-cultural processes.

Why do bad methods persist in some academic disciplines, even when they have been widely rejected in others? What factors allow good methodological advances to spread across disciplines? In this paper, we investigate some key features determining the success and failure of methodological spread between the sciences. We introduce a formal model that considers factors like methodological competence and reviewer bias toward one’s own methods. We show how these self-preferential biases can protect poor methodology within scientific communities, and lack of reviewer competence can contribute to failures to adopt better methods. We then use a second model to argue that input from outside disciplines can help break down barriers to methodological improvement. In doing so, we illustrate an underappreciated benefit of interdisciplinarity.