From bees to robots: The new power of swarm intelligence — DW Documentary

[00:00:00] Speaker 1: A flock of starlings in the sky is a system that offers advantages for each individual bird. [00:00:11] Speaker 2: A predator detected by part of the group causes a wave of turning, a sort of ripple through the group that allows individuals that don't even see the predator to move away from the threat. [00:00:22] Speaker 1: Seemingly solitary individuals benefit from the group. Gliders and paragliders alike depend on others. [00:00:34] Hannah Williams: If you fly with a group, you should be able to fly more efficiently and take higher risks. [00:00:40] Speaker 1: More safety, better solutions. Does swarm logic also apply to society? [00:00:50] Speaker 4: Swarm intelligence is basically a different way of thinking. The goal is to move away from control toward coordination. [00:00:56] Speaker 1: What happens when a factory is organized like an animal swarm? [00:01:04] Speaker 5: In experiments, we have seen performance increase significantly with swarm intelligence. Production can be up to three times faster. [00:01:20] Speaker 1: How much of today's world is based on swarm intelligence? [00:01:25] Myrta Galesic: Almost everything we have achieved, from finding shelter, building roads, economics, culture. This camera that is now recording us, all of that is a product of a series of really good, correct collective decisions. [00:01:40] Speaker 1: Can humans learn from collective intelligence? [00:01:49] Speaker 2: We can certainly learn from animals because many of the animals that we study, they have evolved to live and coordinate behavior at very large scales. Thousands, tens of thousands, even millions of individuals, now humans haven't. [00:02:07] Speaker 1: In swarms, information circulates. No one individual knows everything, not even bees. [00:02:14] Speaker 7: It is absolutely essential that unverified information is not passed on. Bees have known this for millions of years. Now think about social media and what it has done to humanity. There we are not seeing swarm intelligence. We are seeing swarm stupidity on a massive scale, because information is simply copied and shared without verification. [00:02:35] Speaker 8: Because information is simply copied and shared without verification. [00:02:52] Speaker 1: A swarm is a perfectly organized system, at least in the animal world. No single individual is responsible for everyone else. Information triggers a chain of reactions that makes the swarm smarter and more resilient as a whole. Scientists have long debated whether humans are a foolish herd or an intelligent collective. In 1906, British naturalist Francis Galton, a cousin of Charles Darwin and the founder of the pseudoscientific theory of eugenics, set out to prove what he called the "stupidity of the masses." At a cattle market, 800 visitors were asked to estimate the weight of a bull. Almost no one guessed correctly. Some thought the bull was far too heavy, others thought it was too light. But the average of all the guesses came remarkably close to the animal's actual weight. To this day, the experiment is considered evidence of the wisdom of collective intelligence. [00:04:09] Speaker 9: The people are judging independently. [00:04:11] Myrta Galesic: When they're somehow acquainted with the problem, they see the ox, ox is right there. So they're more, they're likely to be somewhere in the ballpark. This is not a completely unknown thing to them. Many problems are not like that. And this is why this whole idea of collective intelligence, well, it's an attractive term, it's a little bit deceiving. Everything is really about adaptation. It's about the group recognising in which situation they are. We all know people who score high on IQ tests, these conventional IQ tests, but then you ask them to make a sandwich or to organise a trip, you know, for their children and they are useless. So everything is about being able to adapt to problems people are facing. [00:04:53] Speaker 1: Psychologist Myrta Galesic studies how human groups function at the Complexity Science Hub in Vienna. Her work draws on large data sets, including information from social media, movement tracking and records of parliamentary decisions. [00:05:10] Myrta Galesic: Until recently, it has been very difficult to actually measure collective behaviour, the many beliefs of a collective, the ways collective make decisions. Of course, experimenting is almost impossible, but therefore you need a lot of data. But now this is possible. And also scientists are now more communicating across disciplines than ever before. So this institute is one example. There's so many people coming together to study this beast of human collective. [00:05:48] Speaker 1: Vultures spend most of their time gliding, allowing them to cover vast distances. Hannah Williams is investigating if and how vultures fly as a group. To find out, she works with paragliders. [00:06:02] Hannah Williams: When I went camping in Covid, you can see paragliders were out during Covid. They've got no interaction with anybody else, so they can fly. And that's when I started looking at them and thinking they're doing what the birds are doing, I can predict where they're moving next. And so if they're doing what the birds are doing, let's study them. And then I happened to know some people and they got me into the sport. And now I've learned to paraglide as well. It's terrifying, but really awesome for data and understanding the air. [00:06:40] Speaker 1: Although these birds may appear to be loners, they move through the landscape as a group and benefit from one another. [00:06:46] Hannah Williams: That comes from a place where I see data on a big screen where you can zoom out of context and you see many, many dots in the sky that look linked. But if you stand on the ground and look up, you probably won't see a swarm. You'd see a few individuals, but they can see for kilometers, these individuals. So they have a much bigger context to their space use and swarm than we can see from the ground. So, yeah, they're very comparable, but from a different context. [00:07:29] Speaker 1: In Argentina, Williams has also studied the flight behavior of the Andean condor. The world's largest flying bird often glides for hours at a time. [00:07:41] Hannah Williams: They're super efficient. It's amazing. They can flap very, very little and then use the airwaves to fly a great distance. And they can fly 100 kilometers a day without flapping their wings. And then you scale that up and it's hundreds of kilometers over a certain time period. We're watching birds go up and down the Andes several times within a year. It's amazing. [00:08:10] Speaker 1: Along mountain ranges, rising air currents create thermals, columns of air that lift both birds and paragliders back into the sky. [00:08:21] Hannah Williams: They have a similar movement efficiency or a movement capacity. So a paraglider has no engine. It means that they're completely reliant on the air flows to be able to travel. And they're using updrafts to be able to travel. A condor is limited really to be able to use the same updrafts. It can flap its wings to fly and you'll see many pigeons flapping. But that's very, very costly for a big animal. So the bigger the animal, the more like a paraglider or something without a motor. So I use the paraglider because I want to get inside the head of a bird. And by wanting to look at the eyes and where they're examining, it's very, very hard to do that in the wild with a large bird. So we do that with paragliders given their similarity. [00:09:10] Speaker 1: For this research, Williams is working with the British National Paragliding Team. [00:09:19] Hannah Williams: Show me what you're seeing, but it needs to be relative to your movements. And so this tag is giving me the movements of your head. And with the tag that's in your equipment, the two compare to each other. Tell me what your body and your head are doing. So that's going to give me this movement as well as your direction. Piece it together in the computer with everybody else's helmets. And I know who you're looking at and what the others were doing when you looked at them. And that's going to record the whole time you're flying. [00:09:49] Speaker 1: During cross-country competitions, pilots can fly distances of up to 60 miles without stopping. Altitude and speed vary depending on weather conditions and the strategic decisions of the competitors. For pilots, close observation is critical. As soon as one pilot finds the perfect thermal, it becomes visible and valuable information for everyone else. [00:10:13] Hannah Williams: This is the way we think that they are using social information, because we can see it in their risk-taking in their glides. If you take a bigger risk, a greater risk, then you can increase your flight speeds, but you will hit the ground quickly if you don't get things right in your decision. So if you fly at a high speed, you have a really steep glide angle and that will mean you're going to hit the ground unless you've got some information that tells you otherwise. Because we see that there is an increase in speed in the glides when they're flying towards where other individuals are, it indicates to us that they're using that information for a benefit. Overall, that will have a big consequence for movement performance. If you fly with a group, you should be able to fly more efficiently and take higher risks. [00:11:06] Speaker 1: Humans are social beings. We live together in many different forms: families, among friends, in the workplace or as citizens of a nation. To navigate these communities, humans have developed an awareness of their surroundings, a kind of social sense. [00:11:24] Myrta Galesic: One is that people are really attuned to their immediate social environment. They approximately know who is around them. They will know approximately, you know, who earns what, who has what education, but also including who has what kind of broad values. Are they politically leaning this way or that way or are they religious or not? People know these things and that's important because we are constantly always evaluating our social environments in search for friends, to be worried of people who could do us harms. And so we do have this very fine social sense. Is that a mess? [00:12:03] Speaker 1: That social sense can also be used in election forecasting. The 2016 US presidential elections demonstrated that traditional models can fail. Almost no one believed Donald Trump had a realistic chance of winning. Myrta Galesic analysed elections in several countries, including multiple US elections. To make more accurate predictions, she believes researchers need to tap into people's social sense. In practical terms, that means don't ask people whom they plan to vote for, ask about their friends instead. [00:12:43] Myrta Galesic: But there is an additional reason and that is that our friends define us. So in situations where a person themselves are not, are not yet decided about who to vote for, but they say, oh, most of my friends are going to vote for this person. It is likely that in a month or two, when the elections come, this person will also become similar to their friends. So that's kind of, you're kind of looking at the crystal ball of yourself by reporting your friends today. [00:13:09] Speaker 1: So if people tend to behave much like their friends, what distinguishes us from the swarms we see in the animal kingdom? [00:13:23] Speaker 2: When we look at the greater complexity of humans, such as our ability to exchange information using language, this is something that we don't find in the animal world. Humans have this unique ability. And of course, we have invented technologies that allow us to communicate near instantaneously over the entire planet, which is absolutely remarkable. But remember, we haven't evolved to have this type of information capability. [00:13:51] Speaker 1: Biologist Ian Cousin is considered one of the pioneers of collective animal behaviour. At the Centre for the Advanced Study of Collective Behaviour at the University of Constance, researchers use cutting-edge technology to study the behaviour of swarming animals and the way they perceive their environment. [00:14:19] Speaker 2: The more we learn about animal sensing, the more amazing it is. You know, they can sense magnetic fields, they can sense odours in ways that we can't sense. They have extra cells in their eyes that allow them to see in a different way to humans as well. So these are really the frontiers of our understanding. So we're moving away from these sort of old particle models, almost like physical particles interacting with each other, really to consider the sensing and the complexity of cognition that's required to make sense of the world. [00:14:56] Speaker 10: We're now inside the imaging hangar. [00:14:59] Speaker 4: This is the largest experimental space we have here in Constance. Mounted on the ceiling, you can see more than 40 cameras specifically designed to track animals. This is the arena, where the ants move around and search for food. The panoramic image surrounding the arena was taken just 100 meters from here, near St. Catharina. We tried to create a natural visual environment for the ants. [00:15:34] Speaker 1: Ants are among the most complex social insects. Their antennae play a key role, allowing them to detect odours and chemical signals and communicate with one another. In this experiment, researchers are trying to determine how important vision is for this particular ant species. [00:15:54] Speaker 2: We can now embed animals like fish or locusts into immersive volumetric virtual realities. So we're projecting the world, but from their perspective, it looks 3D. So, you know, we can't put glasses on these animals like, you know, humans do, but we can immerse them nonetheless in completely three-dimensional volumetric environments. And this has really been a revolution. [00:16:23] Speaker 4: This is our virtual reality room. Here we conduct experiments using virtual reality setups, and we also develop our own VR setups for insects. The ants you saw earlier in the hangar are brought in here afterward. [00:16:41] Speaker 1: Using a microscope, researchers attach a tiny rod to the ants' back. To keep the insects still during the procedure, it's cooled beforehand, placing it into a completely harmless, sleep-like state. [00:17:01] Speaker 4: These are small styrofoam balls that function like treadmills for the ants. They rest on a cushion of air, and once I turn on the airflow, the balls float without friction, allowing the ants to walk on them. [00:17:21] Speaker 10: When we start the experiment, the ants see a reconstruction of the hangar we were just in. [00:17:29] Speaker 4: They are now placed at the center of this virtual world and perceive it exactly as if they were actually there. So you can test whether they behave the same way here? In the real world, the ants have already learned how to find their way home. The question is whether this visual information alone is enough for them to relocate their nest. If it is, then we can manipulate specific elements of the virtual environment, and identify exactly which visual cues the ants use to navigate back home. [00:18:07] Speaker 1: After the experiment, once the glue has dried, the rod eventually detaches on its own, and the ant returns unharmed to its nest. In this particular setup, researchers can observe large swarms and analyze the behavior of every individual animal. When scientists studied a massive swarm of locusts, they made a surprising discovery. [00:18:33] Speaker 2: Initially we thought, well, they must be exchanging information very much like fish schools and bird flocks. But we found that's not the case at all. In actual fact, these little individuals are really trying to survive in a harsh environment. They run out of food. And sort of counterintuitively, when they run out of food, they gather together and they all start marching together. But one of the driving factors is cannibalism. Each individual is trying to eat those ahead and trying to avoid being eaten by those approaching from behind. The outcome looks like it's cooperative. It's coordinated, but it's far from cooperative. It's a sort of selfish, sort of cannibalistic horde. Everyone trying to eat those ahead and avoid being eaten by those behind. And so we have to be very careful when we think of what drives coordinated behavior. And it's not always cooperation. [00:19:24] Speaker 1: In the shallow waters of the Maldives, black-tip reef sharks are hunting through a school of fish. [00:19:35] Speaker 2: I mean, a few years ago, I realized that I always show these beautiful videos of schooling fish and predators attacking these schools in my talks. And I thought to myself, wow, no one's actually studied this. Like, no one ever has studied this. And now we have the technology that allows us to fly drones with really high-resolution cameras above these fish schools and to use AI to track the movement, both of the predators, in our case sharks, hunting the schooling fish. And so initially I was thinking, well, the schooling is going to be the really interesting collective part. But of course, you know, what a surprise it was to discover that the sharks themselves are also coordinating their behavior. [00:20:22] Speaker 4: Here you can see white dots on the shark. These are tracking points automatically detected by the computer. Thanks to them, we can analyze exactly how the sharks swim, how fast they move their tails, and how their behavior changes when they are inside the school compared to when they are outside of it. [00:20:45] Speaker 1: The sharks are juveniles most likely practicing how to hunt. For the fish, the school provides ideal protection. If even a single fish senses danger, the entire group reacts immediately. [00:20:58] Speaker 2: This allows the group itself to collectively sense way beyond the capabilities of the individual. And so we've shown schooling fish. For example, each individual is not aware of these long-range gradients, but they've evolved to have social interactions and to change their local behaviors to allow the group to sense this. So there's a genuine emergent intelligence at the level of the collective that doesn't exist at the level of the individual. And that's what's so surprising. We all think of the brain as being where information is processed. But here, we almost have a meta-brain, this invisible connection among all of these individuals, allowing them to achieve greater intelligence. And this holds from schooling fish to flocking birds, and potentially even humans have this type of collective intelligence as well. [00:21:48] Speaker 1: Collectives are capable of remarkable things, but they are also vulnerable. In an age of constant communication and global connectivity, it is crucial to know whose voices are heard and whose are not. [00:22:01] Myrta Galesic: Groups across our human history, across different timescales, groups of different sizes, where all somehow, given their current constraints and problems they were facing, they managed to make reasonably good solutions that brought us where we are. Then the problem is that, unfortunately, there are always, in collectives, there is this additional factor of manipulation. So there are people who have, or organizations that have special interests, that make it more difficult for experts to be recognized and appreciated. [00:22:33] Speaker 1: In the United States, the president is drastically scaling back climate research. Funding is being cut, researchers are being dismissed, and institutes are being shut down. [00:22:45] Speaker 11: It's the greatest con job ever perpetrated on the world, in my opinion. Climate change, no matter what happens, you're involved in that. No more global warming, no more global cooling. All of these predictions made by the United Nations and many others, often for bad reasons, were wrong. They were made by stupid people that have cost their country's fortunes, and given those same countries, no chance for success. If you don't get away from this green scam, your country is going to fail. [00:23:20] Myrta Galesic: We are exposing as traitors, people who are speaking, who are saying something else. And this way we are losing, basically, access to valuable expertise about things from technology, to climate change, to medicine. Oftentimes, experts such as climate scientists, or epidemiologists, or social scientists warning about misinformation, or the dangers of AI, they stand in the way of very powerful interests. And so, these interests will, are often calling these experts fakes, or they hate us all, they are haters of our group, don't listen to these experts. And so, this deceives the collective. [00:24:03] Speaker 1: One collective that's immune to this kind of misinformation is the bee colony. [00:24:11] Speaker 7: It is absolutely essential that unverified information is not passed on. Bees have known this for millions of years. One bee flies out, finds a food source, returns, and tells one or two other bees, there is good food over there. But those bees would never simply pass that information along. [00:24:29] Speaker 8: They do not immediately start dancing and saying, I heard there is good food over there. [00:24:33] Speaker 7: The first thing they do is fly there themselves and verify it. Is there really good food? If true, they return and inform one or two more bees. And those bees also verify it themselves before passing it on. In other words, information is only shared that has been successfully confirmed. Now, think about social media and what it has done to humanity. There we are not seeing swarm intelligence. We are seeing swarm stupidity on a massive scale. Because information is simply copied and shared without verification. [00:25:07] Speaker 1: Biologist Thomas Schmichle founded and leads the Artificial Life Lab at the University of Graz. There, researchers use cutting-edge technology to study the logic of swarms. This is Robo-Royal, an autonomous robot equipped with a camera. [00:25:37] Speaker ?: And this is a matter of fact. [00:25:39] Speaker 1: Around the clock, it follows and films the queen bee inside a hive. The red light is invisible to bees. [00:25:57] Speaker 7: Observation hives, based on the same principle as this one, have existed for a very long time. Even in ancient Rome, people likely built beehives out of extremely thin bone panels that allowed them to glimpse what was happening inside. Later in the 18th century, around 1730 in France, honeybee research really began to emerge as a scientific discipline, using glass observation hives. Many people spent countless hours sitting in front of those hives, trying to understand what was happening inside. But in some respects, humans are not particularly well suited for this kind of observation. It is fascinating to look inside for a while, but remaining focused for hours and trying to take in everything at once, quickly pushes us to our limits. [00:26:44] Speaker 1: The robot's recordings are stored and later analyzed using artificial intelligence. [00:26:55] Speaker 7: Here you can see a live view from the system. At the moment, the queen is interacting with another worker bee. We may soon see a feeding event. You can also clearly see the queen's retinue forming around her. These are worker bees that position themselves around the queen, feed her, groom her and clean her. They also absorb her pheromones and distribute them throughout the colony so that all worker bees know the queen is present. All of this can also be analyzed automatically. Here, for example, you can see a video of the system identifying how many bees are currently part of the queen's retinue, counting them and evaluating those patterns over the course of an entire season. [00:27:43] Speaker 1: The queen bee has a small marker on her back, allowing the robot camera to recognize and follow her. [00:27:52] Speaker 7: Historically, people always believed the queen dominated everything. And that is why she is called the queen. For centuries, even millennia, people assumed she controlled the entire hive. Then came a period when people believed everything was grassroots and bottom-up, decentralized. At that point, they said, actually, the queen bee is just an egg-laying machine controlled by the collective. Suddenly, there was nothing regal left about her. The idea became that she was simply an egg-laying machine. Protein goes in the front, eggs come out the back. You can see quite clearly how she extends her proboscis. All these interactions involving the antennae are incredibly fascinating. We also know that with bees, either their left or right antennae is dominant. The antennae are used differently. It is not symmetrical. There is a lateral specialization. There's a left and a right? Exactly. A bee colony is organized in a complex way. Part of its regulation is indeed bottom-up and decentralized. Everything said about that was correct. But there is also top-down regulation coming from the queen. For example, she suppresses other bees from laying fertilized eggs or developing into queens themselves. The queen plays a central role because she produces large amounts of chemical signals. These are spread throughout the hive by worker bees that groom and lick her. The signals then influence the behavior of the workers and shape these processes. That closes the loop between top-down regulation and bottom-up regulation. The result is an extraordinarily complex regulatory system that even today we do not fully understand. [00:29:52] Speaker 1: Grassroots democracy or authoritarian control – which system produces better outcomes for humans? [00:30:03] Myrta Galesic: Some problems are very simple or require very fast action. Then it is good to be connected. Then even the brightest experts – if something needs to be solved now, this moment, we need to do anything. It doesn't have to be the optimal solution, just something. Then it is good to be connected. Often it is good to be hierarchically arranged, to have a strong leader. No matter if the leader is the smartest. If another group is in front of the door and we need to defend ourselves. If pandemic strikes, we need to do something tomorrow. Then we need to do – then a different, more tightly connected network structure is better. [00:30:42] Speaker 1: So when can society benefit from collective intelligence? How should a group be structured to arrive at innovative solutions? [00:30:56] Myrta Galesic: And so really, there is not one holy grail. You need to recognize when the problem is complex. Then, oh, let's invite people from all over the place of many different opinions. Let's discuss and deliberate. And this is the knowledge that we kind of sometimes intuit, but not really have yet. We didn't internalize this as a collective. And this is something that we are studying now. But I'm hoping that – I know that we will – we are developing basically a toolkit, so we will know better as collectives to recognize where we are and how we need to – how we need to organize. So yeah, I mean, in sum, diversity is not always good. Homophilia is not always good. Large groups are not always good. Small groups are not always good. But there are clear ways to recognize when we are, where we are and what we need to do. [00:31:45] Speaker 1: A bee colony is highly homogenous. Nearly all bees are female and share the same bee as their mother, the queen. This system is incredibly precise. Researchers at the University of Graz are figuring out exactly how this works. [00:32:00] Speaker 5: "This is smoke. It calms the bees. There's an evolutionary reason for that. To bees, smoke signals fire. They begin preparing to leave the hive in case something nearby is burning." [00:32:20] Speaker 1: Carniolan honeybees are considered especially gentle. Even so, they attack the black boom mic. [00:32:25] Speaker 5: "That is black and fuzzy. To the bees, that looks like an enemy, like a bear. You can see I'm wearing light-colored clothing. Anything dark, large and fuzzy is more likely to trigger defensive behavior in bees, [00:32:49] Speaker 12: while lighter colors trigger less of a response. Bees do not perceive the environment the way humans do. [00:33:01] Speaker 5: Their vision is more like a grid or mosaic. Anything large, dark and fuzzy can trigger an attack response. "They can break the attack response." [00:33:27] Speaker 1: Jutta Vollmann is searching for honeycombs containing young bees, which she will then bring into the laboratory for research. [00:33:34] Speaker 5: "A bee is hatching here right now. I would select this comb." [00:33:48] Speaker 7: "There are several things that make young bees. What many people call baby bees, so interesting. First, they are ideal for experiments because they can't fly yet. That means we can place them in walking arenas and observe how they move. Second, they cannot regulate their own body temperature yet, so they depend on the surrounding temperature. That allows us to send them signals and observe how they respond. Young bees are the ones that prepare and clean the cells for the queen to lay eggs. The queen lays fresh eggs only in prepared cells, never in unprepared cells. [00:34:33] Speaker 1: In this arena, researchers can expose the young bees to different signals, such as airflow or heat, and observe how they react. After the experiments, the bees are returned to a hive. [00:34:50] Speaker 7: "We discovered that baby bees use a very clever mechanism to identify the warmest spot among several options. They move randomly and adjust their behavior according to the local density of bees, how often they encounter one another, and how warm a location is. Through these simple mechanisms, they consistently manage to find the warmest spot available. We wanted to prove that this was really how it worked. That is difficult with bees, because they do what they want. So instead, we used a swarm of robots and programmed only this specific behavior into them. That way, we knew nothing else was influencing the outcome. We recreated the experiments and showed that the robot swarm can also locate the brightest or warmest spot in the arena. [00:35:42] Speaker 1: The EPUC robots move randomly and stop when they encounter one another. The brighter the area, the longer they remain there. Over time, nearly all gather in the brightest spot. [00:35:57] Speaker 7: "All except one. That one is the outlier. It refuses to join the group. And that is actually a good thing. There should always be a few individuals that do not follow the crowd, because they may discover new options or new light sources." [00:36:19] Speaker 1: Findings from collective intelligence research can also be applied to technology. Traffic flow, for example, could be improved if traffic lights reacted independently to current conditions rather than being centrally controlled. The principles of swarm behavior could also improve drone swarms searching for missing people in hard-to-reach areas. This factory produces microchips. Production is currently controlled centrally. In the future, however, the company hopes to switch to swarm intelligence. [00:37:00] Speaker 5: Swarm intelligence systems contain agents that behave like natural swarms do, [00:37:08] Speaker 13: whether ants, bees, fish or birds. [00:37:12] Speaker 5: They abide by shared environmental factors. That means each individual agent or ant makes its own decisions, depending on its environment and, in some cases, exchanges information with others. As a result, the collective as a whole achieves its objective. [00:37:34] Speaker 1: This semiconductor factory operates under ultra-clean conditions. Employees must follow strict hygiene protocols. [00:37:43] Speaker 4: "This area is cleaner than on Austria's Grossglockner mountain, and 20,000 times cleaner than in an operating room. Employees must wear special clean room clothing. They're not even allowed to wear jewelry. If you slide your finger across a ring, a few gold atoms can become loose. And that alone can destroy a chip." [00:38:04] Speaker 1: The containers transport semiconductor wafers between machines. The sequence varies depending on the product. Some stations operate at full capacity while others wait for incoming material. Around 1,600 different products are being processed at once. [00:38:20] Speaker 4: "This is where the chips are manufactured. Producing a chip is a major challenge. We manufacture thousands of different products, all with very different production times. Some take one or two months, others up to four months to be completed." [00:38:39] Speaker 5: "In a traditional technical system, there is usually some central point that collects all the data, [00:38:45] Speaker 13: makes all the decisions and instructs the machines what to do. In collective intelligence, that does not exist. Each individual agent makes its own decisions." [00:39:03] Speaker 1: The goal is to move away from centralized control towards swarm-based principles. Melanie Schrantz of Lakeside Labs in Klagenfurt is researching how that could work in close cooperation [00:39:14] Speaker 5: with the company Infineon. "We have an entire library of different algorithms inspired by many kinds of systems. That includes bees, ants, birds and bats, but also slime molds and hormone-based algorithms. It's a very broad field. We study all kinds of self-organizing algorithms, especially those based on swarm intelligence. [00:39:44] Speaker 1: To explore these possibilities, researchers have created a digital twin of the production facility. The difference is that processes are no longer centrally controlled. Instead, the products behave more like individuals within a swarm. [00:40:04] Speaker 5: "It's not as simple as saying, let's take a few ants and bees and create an algorithm to optimize production. We study the biological model, create an abstract environment and simulation and use these algorithms only as inspiration. We cannot simply copy nature one-to-one. An ant has completely different characteristics from a manufactured product. So this is really an engineering process in which we examine how the underlying principles of these algorithms can be adapted to model workflows and decision-making processes." [00:40:50] Speaker 1: For now, the factory is still centrally controlled. It will likely take years before swarm intelligence can be fully implemented. [00:41:06] Speaker 5: "In experimental settings, we can achieve a very significant performance increase with swarm intelligence, meaning production can be up to three times faster." [00:41:21] Speaker 1: If swarm logic can optimize production processes, could it also improve democracy, especially in a world that is constantly changing? [00:41:33] Speaker 4: "In reality, we often face complex problems today where nobody truly knows the optimal answer. We know parts of the solution or possible solutions, but by tomorrow there may already be an innovation that completely changes everything." [00:41:56] Speaker 1: "Dirk Helbing is a complexity researcher at ETH Zurich, a university. He and his team are studying how collective intelligence could be used to strengthen society and make democracy more resilient and fair. [00:42:22] Speaker 4: "The problem is that we now live in a situation where a majority, typically a coalition, decides how resources are distributed, and this often leads to clientelism. In other words, many people end up receiving very little. Over time, they feel excluded and become dissatisfied. We therefore believe that democracy can and must be upgraded in a way that is more participatory." [00:43:07] Speaker 1: This kind of participation model works in stages. First, ideas are collected, developed in a collaborative fashion, then finally voted on. Helbing believes that budget control is key to this kind of collective planning. A concept similar to participatory budgeting, which originated in the late 1980s. Thanks to digitalisation, this idea has spread globally. Now ideas are assigned budgets and voters can allocate a single budget across multiple projects. [00:43:45] Speaker 4: The best solution is usually when there are many projects that complement each other, rather than one large project that consumes the entire budget. Under today's majority voting system, a single large project often absorbs most of the funding, [00:44:07] Speaker 14: leaving little for everything else. And then nobody is really satisfied. [00:44:19] Speaker 15: I think we need much more of this, especially when it comes to children and young people, giving them a voice and enabling them to implement what they actually want to see in their city. We have no idea what the world would look like if children had a say and could co-decide just like adults. [00:44:50] Speaker 5: I don't really say I'm a city kid or a country kid. I say I'm a Robbie kid, because it's part of who I am and how I've been shaped, very connected to nature. It's been a really important place. The game is a bit tricky if you haven't heard of it before. You slapped the Nazis, not the police. That's important. [00:45:27] Speaker 15: And as we've seen, more and more of it is falling apart and basically being used as a storage room. We could paint the outside to make it look nicer, or create a small area inside where people can hang out or meet. Look how torn up everything already was. It was really, we realized there was more rust than expected. [00:46:09] Speaker 1: This renovation was made possible through the Participatory Children and Youth Million Initiative, in which the city of Vienna allocated 1 million euros for projects. All children and young people could submit their ideas. [00:46:30] Speaker 5: The application itself wasn't actually that difficult. It was just uploading a photo. But it became clear early on that it wouldn't be just for us as young people, but more for the generation after us. [00:46:55] Speaker 1: After submission, the projects were reviewed. The next step was planning how to carry them out with the help of experts. [00:47:08] Speaker 15: I found it a bit exhausting. I think it was just a lot for both of us because we were still in school full time, and had to manage everything while also planning and thinking along with this project. But we both really liked having the opportunity to have a say and help decide. We were very happy that it could become our own space. As Robbie youth, who don't quite fit into any one group, creating a place of our own. [00:47:46] Speaker 1: All children and young people in Vienna aged 5 to 20 could participate in the voting process, regardless of citizenship. [00:47:58] Speaker 15: It was very easy, and you could vote for multiple projects. I think you could even distribute 1 million across different ideas. It was also really nice to see what others had submitted, different ideas from different age groups. There were even primary school kids sharing what they wanted, and it was interesting to see their interests and what they wanted to change. So I didn't just vote for our project, but also supported other ideas I thought were good. I think it's good when people have more of a say in our everyday lives and how our city looks. In the end, we all live here and participate in daily life in Vienna. Being able to decide what we want is very important for everyone. Give a voice to people who actually know from experience what they need and where support is needed. [00:49:00] Speaker 5: Through things like this, it becomes easier to consider things that would otherwise be overlooked. The people who plan things might simply not think of them, and then something gets missed that later causes problems or inconvenience. [00:49:21] Speaker 4: Collective intelligence can now be integrated into all processes. Good ideas can ultimately help shape the future of our cities, and our world. [00:49:40] Speaker 2: There really is no such thing really as individual intelligence. It's all kind of collective. And I think we, in certain Western society, we don't really appreciate the fact that we're both special and also contributing to this greater good potentially. [00:50:00] Speaker 7: We're still focused on understanding the swarm intelligence of bees and building models from that. That is what complexity research is doing. And it's why we need autonomous robots to observe it, and artificial intelligence to recognize patterns and process the data. Because up to now, this has been largely misunderstood. [00:50:27] Myrta Galesic: The same way we can learn about how our collective body functions, we're still at the beginning. If we just manage to not start the Third World War very soon, I'm thinking that, I'm convinced that we can come to that. We will know it. And then, it will be like a collective health checklist. [00:50:48] Speaker 1: One thing is clear: the logic of the many creates groups that are more resilient and more intelligent.