Swarm Intelligence: Lessons from Birds, Bees, and Human Systems

A flock of starlings demonstrates a system where individual birds benefit from the group. When a predator is detected, a wave of turning ripples through the flock, allowing individuals who haven't seen the threat to move away. This concept of "swarm logic" or "swarm intelligence" suggests that seemingly solitary individuals can benefit significantly from collective behavior. Vultures and paragliders, for instance, depend on others to fly more efficiently and take higher risks, leading to greater safety and better solutions.

Swarm Intelligence in Society and Industry

Swarm intelligence is fundamentally a different way of thinking, shifting from control to coordination. When applied to a factory setting, experiments have shown significant performance increases, with production becoming up to three times faster. This suggests that organizing systems like animal swarms can yield substantial benefits.

The concept of swarm intelligence is deeply embedded in human achievements. From finding shelter and building roads to economics, culture, and even the development of advanced technology like cameras, collective decisions have played a crucial role.

Learning from Animal Collectives

Humans can learn from animals, particularly those that have evolved to live and coordinate behavior at very large scales, involving thousands, tens of thousands, or even millions of individuals. Unlike humans, these animal swarms have developed sophisticated mechanisms for information circulation.

In swarms, no single individual possesses all information. A critical aspect of animal swarm intelligence, exemplified by bees, is the rigorous verification of information before it is passed on. Bees will not share unverified information; they will personally confirm a food source before communicating it to others. This contrasts sharply with the "swarm stupidity" often observed on social media, where information is copied and shared without verification, leading to widespread misinformation.

An animal swarm is a perfectly organized system where no single individual is solely responsible for the entire group. Information triggers a chain of reactions that enhances the swarm's overall intelligence and resilience.

The Wisdom of Crowds: Francis Galton's Experiment

For a long time, scientists have debated whether human collectives are foolish herds or intelligent groups. In 1906, British naturalist Francis Galton, a cousin of Charles Darwin, conducted an experiment to test the "stupidity of the masses." At a cattle market, 800 visitors were asked to estimate the weight of a bull. While individual guesses varied widely, the average of all guesses was remarkably close to the bull's actual weight. This experiment is now considered evidence of the "wisdom of collective intelligence."

However, this concept has limitations. The success of collective intelligence, as seen in Galton's experiment, relies on individuals making independent judgments and having some familiarity with the problem. Many real-world problems are far more complex and less straightforward. The true essence of collective intelligence lies in a group's ability to adapt to specific situations. As psychologist Ma Galacles notes, individuals who score high on conventional IQ tests might struggle with practical tasks, highlighting that intelligence is about adaptability.

Studying Human Groups with Big Data

Ma Galacles, a psychologist at the Complexity Science Hub in Vienna, studies how human groups function using large datasets, including social media information, movement tracking, and parliamentary decisions. Historically, measuring collective behavior and decision-making was challenging, but with the advent of big data and interdisciplinary collaboration, it has become possible.

Vultures, Paragliders, and Social Information

Hannah Williams investigates how vultures fly as a group, leveraging the efficiency of gliding to cover vast distances. She works with paragliders, who, like vultures, rely on air currents for flight without engines. Williams observed that paragliders exhibit similar behaviors to birds, allowing her to predict their movements. She learned to paraglide herself to gain a deeper understanding of the air and the dynamics of flight.

While individual vultures may appear solitary, they move through the landscape as a group, benefiting from each other's presence. Williams describes this as a "swarm," even if it's not always visible from the ground. From a higher perspective, data reveals many linked "dots" in the sky, indicating a collective movement. Vultures and paragliders have a much larger context for their spatial use than humans on the ground can perceive.

Williams has also studied the flight behavior of Andean condors in Argentina, the world's largest flying birds. Condors are incredibly efficient, gliding for hours and covering hundreds of kilometers daily without flapping their wings, utilizing rising air currents (thermals) along mountain ranges. Paragliders, lacking engines, are similarly reliant on these updrafts. Williams uses paragliders as a proxy to understand the cognitive processes of large birds, as it's difficult to observe their eye movements and decision-making in the wild.

For her research, Williams collaborates with the British National Paragliding Team. Pilots wear tags that track head and body movements, allowing researchers to determine who they are looking at and what others are doing. This data helps understand how pilots use social information during cross-country competitions, where they can fly up to 60 kilometers without stopping.

Close observation is crucial for pilots. When one pilot finds a good thermal, it becomes valuable information for everyone else. Williams' research suggests that pilots take greater risks and increase their flight speeds when flying towards other individuals, indicating they are using social information to their advantage. This collective behavior leads to more efficient flight and allows for higher risks.

Human Social Sense and Election Forecasting

Humans are social beings, constantly evaluating their social environments to identify friends and potential threats. This "social sense" extends to understanding who holds what values, education, or political leanings.

This social sense can be applied to election forecasting. Traditional models failed to predict the 2016 US presidential election. Merta Galac suggests that to make more accurate predictions, researchers should ask people about their friends' voting intentions rather than their own. This is because friends often define us, and if a person's friends lean a certain way, that person is likely to follow suit, even if undecided initially.

The Complexity of Human Communication vs. Animal Swarms

While humans exhibit complex social behaviors, our ability to exchange information through language and advanced technology sets us apart from animal swarms. However, biologist Iain Couzin, a pioneer in collective animal behavior, notes that humans haven't evolved to handle this type of information capability.

At the Center for the Advanced Study of Collective Behavior at the University of Konstanz, researchers use cutting-edge technology to study swarming animals and their perception of the environment. Animals possess extraordinary sensory capabilities, such as sensing magnetic fields, odors, and seeing in ways humans cannot. This research moves beyond simplistic particle models to consider the complex cognition required for animals to make sense of their world.

Ant Navigation and Virtual Reality

In the imaging hanger at the University of Konstanz, over 40 cameras track ants in an arena designed to mimic their natural visual environment. Ants, complex social insects, use their antennae to detect odors and chemical signals for communication. Researchers are investigating the importance of vision for specific ant species.

The lab also utilizes immersive volumetric virtual realities to study animal behavior. They can project a 3D world from an animal's perspective, allowing them to immerse fish or locusts in virtual environments. For ants, a tiny rod is attached to their back after being cooled to induce a harmless, sleep-like state. The ants are then placed on styrofoam balls that function as treadmills, allowing them to walk without friction. In this virtual world, they see a reconstruction of their real-world environment.

The goal is to determine if visual information alone is sufficient for ants to navigate back to their nest. By manipulating elements of the virtual environment, researchers can identify the specific visual cues ants use. After the experiment, the rod detaches, and the ant returns unharmed to its nest.

Locust Swarms: Selfishness vs. Cooperation

When studying massive locust swarms, scientists made a surprising discovery. Unlike fish schools or bird flocks, locusts don't exchange information cooperatively. Instead, their collective movement is driven by individual survival in harsh environments, particularly when food is scarce. The swarming behavior is a result of cannibalism: each locust tries to eat those ahead and avoid being eaten by those behind. What appears to be coordinated behavior is, in fact, a "selfish, cannibalistic horde." This highlights that coordinated behavior isn't always a result of cooperation.

Shark Hunting and Emergent Intelligence

In the Maldives, blacktip reef sharks hunt in schools of fish. Researchers use drones with high-resolution cameras and AI to track the movements of both sharks and fish. While the schooling of fish is a fascinating collective behavior, the sharks themselves also coordinate their hunting strategies.

Tracking points on the sharks allow analysis of their swimming patterns and how their behavior changes inside and outside the fish school. These juvenile sharks are likely practicing hunting. For the fish, the school provides protection. If one fish senses danger, the entire group reacts, allowing the collective to sense beyond the capabilities of any individual. This "emergent intelligence" at the collective level, a "metabrain," allows for greater intelligence that doesn't exist at the individual level, a principle applicable from schooling fish to flocking birds and potentially humans.

Vulnerabilities of Collectives and the Role of Experts

Collectives are capable of remarkable feats but are also vulnerable. In an age of constant communication, it's crucial to understand whose voices are heard. Throughout history, human groups have found reasonably good solutions to problems, but manipulation remains a constant threat. Powerful interests can undermine experts, labeling them as "fakes" or "haters" to deceive the collective and advance their own agendas. This is evident in situations like the scaling back of climate research or the dismissal of experts in various fields.

The Bee Colony: A Model for Verified Information

The bee colony offers a powerful counter-example to misinformation. Bees have known for millions of years that unverified information should not be passed on. When a bee finds a food source, it informs only a few other bees. These bees then verify the information themselves before passing it on to others. This ensures that only successfully confirmed information is shared, preventing the spread of "swarm stupidity" seen in human social media.

Robotics and AI in Bee Research

Biologist Tomas Schmikler leads the artificial life lab at the University of Graz, studying swarm logic using cutting-edge technology. Robo Royale, an autonomous robot equipped with a camera, continuously films the queen bee inside a hive. The red light used is invisible to bees.

Observation hives have a long history, from ancient Roman bone panels to 18th-century glass hives. While humans find it fascinating to observe bees, maintaining focus for extended periods is challenging. Robots, however, can record continuously, and AI analyzes the data to identify patterns, such as the queen's interactions, feeding events, and the formation of her retinue (worker bees that feed, groom, and clean her, distributing her pheromones throughout the colony).

Historically, the queen was believed to control the entire hive. Later, the view shifted to a bottom-up, decentralized system where the queen was merely an "egg-laying machine." However, research now shows that a bee colony has both bottom-up and top-down regulation. The queen plays a central role by producing chemical signals that influence worker behavior, creating a complex regulatory system that is still not fully understood.

Collective Intelligence for Societal Challenges

The question arises: which system—grassroots democracy or authoritarian control—produces better outcomes for humans? For simple problems requiring fast action, a tightly connected, hierarchical structure with a strong leader can be effective. However, for complex problems where no single optimal answer exists, a more participatory approach is needed.

Dirk Helbing, a complexity researcher at ETH Zurich, studies how collective intelligence can strengthen democracy. He advocates for a more participatory model where ideas are collected, collaboratively developed, and then voted on. Budget control is key, similar to participatory budgeting, where voters allocate a single budget across multiple projects. This encourages complementary projects rather than one large project consuming all resources, leading to greater satisfaction.

Helbing emphasizes the importance of giving children and young people a voice in shaping their cities. The "Children and Youth Million" initiative in Vienna is an example, where young people submitted ideas for projects, and all children and young people aged 5-20 could vote. This process allows for the consideration of ideas that might otherwise be overlooked by adult planners, leading to better solutions for the community.

The Future of Collective Intelligence

There is no single "holy grail" for structuring groups. The optimal structure depends on the complexity of the problem. For complex issues, inviting diverse opinions and fostering discussion is crucial. Researchers are developing toolkits to help collectives recognize their situation and organize effectively. Diversity and homogeneity, large and small groups, all have their contexts.

A bee colony, despite its homogeneity (mostly female bees sharing the same queen mother), is incredibly precise. Researchers at the University of Graz study how this precision works. They use smoke to calm bees, as it signals fire and prompts them to prepare to leave the hive. They also observe that bees react defensively to dark, fuzzy objects, which they perceive as threats like bears.

Uta Fulman studies young bees, which are ideal for experiments because they cannot fly yet and cannot regulate their own body temperature. This allows researchers to expose them to signals like airflow or heat and observe their responses. Young bees are responsible for preparing cells for the queen to lay eggs.

Experiments with young bees in arenas revealed a clever mechanism for finding the warmest spot: they move randomly and adjust their behavior based on local bee density and warmth. To prove this, researchers used a swarm of robots programmed with this specific behavior. The robot swarm successfully located the brightest or warmest spot, demonstrating the underlying principle. The presence of outliers, individuals who don't follow the crowd, is also beneficial, as they may discover new options.

Applications of Swarm Logic

Findings from collective intelligence research have practical applications:

• Traffic flow: Traffic lights could react independently to current conditions rather than being centrally controlled.
• Drone swarms: Drones could search for missing people in hard-to-reach areas.
• Factory production: A microchip factory aims to switch from central control to swarm intelligence. In this system, individual agents (products) make their own decisions based on their environment and information exchange, similar to natural swarms. This could lead to significantly faster production.

While directly copying nature is not feasible, the underlying principles of swarm algorithms can inspire engineering processes to model workflows and decision-making. Full implementation of swarm intelligence in factories may take years, but experimental settings show potential for production increases of up to three times.

Ultimately, collective intelligence is not just about individual intelligence but about how individuals contribute to a greater good. Researchers are still at the beginning of understanding how collective bodies function, but the goal is to develop a "collective health checklist" to guide societal organization. The logic of the many creates groups that are more resilient and intelligent.