Understanding Society, Intelligence, and the Meaning of Life Through the Lens of Complex Systems

 

Introduction:

I have always been very interested in the complexity of the world e.g. how different elements depend on each other, any simple principle behind these complexity etc. In the last half year, I finally got some leisure time digging into a couple of popular books written by scholars from Santa Fa institute which is a theoretical research institute dedicated to the multidisciplinary study of the fundamental principles of complex adaptive systems and founded by Nobel laureates in Physics.

In the last half year, since I work as an engineer and am also very interested in philosophy of science, I read a couple books covering origin of life, history of intelligence and philosophy of intelligence which I collected from various AI talks in 2025.

All the books are listed in the reference sections

I gained many fascinating insights from these books. I also came to see complexity science and intelligence as deeply interconnected. Together, they inspired me to reflect not only on these fields, but also on broader questions about the meaning of life. In this article, I will briefly introduce some of the most interesting ideas from the books, while also sharing my own thoughts and explorations.

TL;DR

• A complex system = many individuals each following simple rules, with no one in charge, yet giving rise to an unpredictable whole. The world is a complex system.
• The world is hard to predict because it's full of nonlinearity, chaos, randomness, and feedback loops.
• True complexity is a "mixture of order and randomness"—the purely ordered and the purely chaotic aren't actually complex.
• The micro is unpredictable, but the macro is predictable—a single person's behavior is hard to foresee, yet averages are highly regular.
• Life can emerge spontaneously from complex systems, requiring only: energy, a bit of randomness, and enough time.
• Life maintains local order by consuming energy, while letting the total entropy keep increasing.
• To understand a complex world, don't just look at static objects—look at relationships, interactions, and processes.
• A word's meaning comes from its associations with other words; the purpose of language is to update other people's predictive models.
• Human consciousness isn't a static entity but a process continually constituted by memory, experience, perception, and interaction.
• Social values and morality aren't eternal, unchanging truths—they're argued over, fought for, and revised throughout history.
• A society can't optimize for just one goal, whether happiness, efficiency, wealth, or equality.
• The essence of intelligence is finding order in a complex world, making predictions, and taking action.
• A large part of human intelligence comes from simulation: simulating the environment, simulating oneself, and simulating how others see you.
• An intriguing paradox: everyone wants to predict others, while also wanting to remain unpredictable themselves.
• AI won't completely replace humans, because AI doesn't share the same body, desires, developmental history, or evolutionary history as humans.
• AI is more likely to form a symbiotic relationship with humans: replacing some jobs while augmenting individual capabilities.
• Centralized autocracy grows ever more "stable" yet ever more rigid, and eventually collapses because its feedback loop is too long.
• Democracy + capitalism is more complex and prosperous, because countless local feedback loops resolve conflicts on the spot; money is the best "interaction protocol" for a complex system.
• For the individual, the meaning of life may not be finding an ultimate answer, but making your own life more complex: keep learning, stay open, cultivate diverse values, plan for the long term while staying flexible, and create more genuine interactions with the world.

The world is complex

A complex system is a system of entities where each typically follows relatively simple rules with no central control or leader. It is the collective actions of vast numbers of components that give rise to the complex, hard-to-predict, and changing patterns of behavior. [1]

The world is obviously a complex system with numerous humans, communities, companies, countries and many other entities.

And as a complex system, the world has following attributes which makes the world hard to predict:

1. This system is non linear which means properties such as population, resources, and criminal rates don’t scale linearly with each other [2]. In math, 90% of problems are related to non linear relationships while major academic schools in economy tend to model the economy by linearizing it as much as possible. [4]
2. The system is chaotic [1].

• Seemingly random behavior can emerge from simple deterministic systems like the Lorenz system, with no external source of randomness.
• The behavior of some simple, deterministic systems can be impossible, even in principle, to predict in the long term, due to sensitive dependence on initial conditions.

1. There is a “purer” randomness factor in the system which is quantum randomness. Bell experiment rules out local hidden variables on quantum mechanics which indicates quantum is more likely to be purely random. Such randomness can serve as the initial condition of a chaotic system like thermal dynamic system and human nerve system. And the randomness can be amplified by these chaotic systems to have macro level impact on the world e.g. humans making random decisions.
2. The system is full of feedback loops including both reinforcing loop and balancing loop. While [3] mentions many general strategies to control such systems based on feedback loop models, it also admits that such systems are very hard to control.

A complex system is hard to predict while not completely unpredictable. In micro scope, it is unpredictable, e.g. a single molecule movement in the air, single person’s behavior, one day weather forecast etc. But in macro scope, the average values are completely predictable. For example, the overall temperature, pressure and volume relationship; long term average temperature change etc. And there is a strong correlation between a complex system’s different properties. For example, an animal specie’s metabolic rate scales sublinearly with body volume which indicates the larger an animal’s volume, the more efficiently it utilizes energy. Human society also has similar correlations. A city’s public resources such as gas station and restaurant scales sublinearly with its population which indicates the larger a city, the more efficiently it utilizes public resources. A city’s salary and crime rate scales super-linearly with population which indicates the larger a city, the more interaction each resident has with others. [2] And [3] illustrated many common patterns of human organizations and their results.

There are many ways to measure complexity. One of my favorite ways, which also reflects the essence of a complex system, is effective complexity. To calculate the effective complexity, first one figures out the best description of the regularities of the entity; the effective complexity is defined as the amount of information contained in that description, or equivalently, the algorithmic information content of the set of regularities. With this measurement, both very ordered and very random entities have low effective complexity. Therefore, a complex system is a mixture of order and randomness. [1]

Life can arise from complex systems. In [7]’s chapter 1, the author constructs a bff system which is a simplified genetic mutation and recombination system. In this system, each object contains a number of genetic codes in which the majority are meaningless code while a small fraction of code has functions like mutating, shifting, copying etc. Each iteration, 2 objects are randomly concatenated and their codes are executed to produce 2 new objects. After millions of iterations, the author discovered each object contains more and more meaningful codes and groups of codes are able to copy themselves to the next generation. This is what real life is actually doing!

So the author concludes that in general, life arises spontaneously whenever conditions permit. Those conditions seem minimal: little more than an environment capable of supporting computation, some randomness, and enough time. Matter subject to random forces is supposed to become more random, not less. Yet by growing, reproducing, evolving, and indeed by existing at all, life seems to violate this principle. The violation is only apparent, for life requires an input of free energy, allowing the forces of entropy to be kept at bay.

And life is truly evolved because of a complex system as illustrated by [5]. Earliest organic matter and prokaryote occurs in a special environment under sea naturally which has 1. energy flow from underground; 2. not too reactive environment; 3. catalyst from soil. This is a molecule level complex system which generates early organic matter.

Then by coincidence 2 different types of prokaryote (bacteria and archaea) merged together into some stable form which is eukaryote. The eukaryote contains major attributes animals, plants and fungi share such as two sexes, the germline–soma distinction, programmed cell death, mosaic mitochondria, and so on. These attributes constitute basic rules of a complex system. Beyond that merging point, according to the book, the evolution from eukaryote into animals, plants and fungi are triggered by rule interacting with each other + natural selection afterward instead of purely environment change. Therefore, the rich and abundant environment of the modern biosphere is evolved because of the complex organic matter system.

Human society is also a complex system. Different political structures and culture evolves on this system. For example, over the last 2000 years, China’s centralized totalitarian political system gradually evolved into a highly stable yet increasingly rigid form of governance. This system profoundly influenced Chinese culture, social organization, and collective values. With low probability and some randomness, few people in human society experienced the “eukaryote” moment and brought a huge breakthrough for the whole human being. For example, Europe experienced renaissance and industrial revolution breakthroughs due to various accidental factors. It brought a huge breakthrough for all human beings. Living standard and productivity increased tremendously. Capitalism, socialism and democracy grew more sophisticated. Culture became more diversified and prosperous.

The evolution in complex systems demands energy flow. In the bff algorithmic system, the energy is the electrical power to run the algorithm. In the organic matter case, at the beginning, the energy is the heat from underground. Later, solar energy took a more and more important role. In human society, energy is food and fuel. Though, with various energy inflows, the complex system’s own entropy decreases with its structure becoming more ordered and more dynamically stable, the total entropy including those excreted during the evolving process, such as burned fuel and excreted heat, increases as illustrated in [5].

My own takeaway from [5] and the idea of energy and entropy flow is that the generation of organic matter was to increase energy releasing speed from underground to sea water in a moderately reactive environment which also increases entropy growing speed. Then after photosynthesis evolved, the energy flow becomes more complex in earth’s echo system. It can make use of not only underground energy but also solar energy. It could buffer both energy for future use like fossil fuel. Sunlight also finds a way to more directly react with earth’s soil through plants and its photosynthesis which produces more entropy.

Relationship, interaction and process are basic elements of the complex world

Modern science and philosophy pays more attention to clear definition of a meaning, object with clear boundary, absolute value and eternal property when modeling the world. Such a paradigm is suitable for a simple model and makes it easy to simulate the world in our mind.

However, after extending this modeling paradigm to the complex world, many problems emerged. We keep clarifying a meaning’s definition but could never find a consistent definition. We keep drawing boundaries for an object but there are always foggy regions. We keep seeking absolute moral values to guide our life but could never find one without doubt unless we treat it as a religion. We keep assuming some complex objects could be statically stable so as to model them in a simple way but we always find dynamics and randomness in those complex objects.

The solution for these challenges, as illustrated in [7] and in post modern philosophy , is to shift the paradigm to treat relationships, interactions and processes as basic elements of the complex world instead.

From the success of attention and transformer based language models, we learn that a word’s meaning emerges largely from its relationships and associations with other words. Furthermore, since language is mainly used to influence other people’s mind such as their prediction of the world, a word’s meaning is best understood functionally: a statement updates the listener’s predictive model rather than referring to a perfectly defined Platonic object. [7]

An object’s boundary definition is usually in “IS-A” form, e.g. “a car is a machine on 4 wheels”. However, the very definition of “IS-A” in natural language dissolves under close inspection; it’s an approximate regularity in the world, not a law or axiom. A “thing in itself” is better understood as a pattern within a web of associations than as an independently defined essence. For example, a car is rather a machine which is powered by gas (relationship with external energy) and can take passengers to different places (interaction) than purely a machine on 4 wheels. [7]

Human consciousness is rather an ongoing process depending on all past interaction history than an absolute existence. Imagine that if you have no memory of your past, or you could never see, hear or touch since your birth, will you have a consciousness as clear as your current’s? Moreover, the past which constructs consciousness is based on hallucination. Yes! Not only AI hallucinates, the human mind also hallucinates. As mentioned in [7] and in many experiments, the human mind hallucinates about observed reality. The human mind even hallucinates about the reasoning of its own decision. Experiments show that a human could make a decision and react due to one reason (due to external inference) while he thinks the decision is made based on another reason. Therefore, the whole human consciousness is an interaction process deeply depending on hallucinated self and past experiences.

The Copenhagen interpretation of quantum mechanics states that a physical system is in a superposition of all possible states until it is measured, at which point its “wave function” collapses into an unambiguous state. The Copenhagen interpretation is consistent with a century’s worth of experimental findings, but it is troubling to reason about it. The observer is also a quantum wave. Why does the observer have the privilege to collapse a particle’s wave function when observing? A particularly straightforward interpretation championed by Carlo Rovelli is “relational quantum mechanics” or RQM. In short, RQM suggests that wave function collapses because of interaction. Similar to the theory of relativity, there is no privileged “view from nowhere.” Events themselves depend on one’s perspective. Pasts and futures are even more local and relative, as they are contingent on the particular network of prior interactions leading up to an event. [7]

Extending this idea to the macro world, [7] suggests that there is no privileged view of reality. Reality consists of relationships. Some agreements may become nearly universal on certain kinds of judgments, especially when those judgments carry strong predictive power. But a judgment may always remain contested, especially when it concerns subjectivity itself. Therefore reasoning and argument are required for interaction and social cooperation. We can’t trust a judge to make a decision directly from a holistic view without “lawyers” arguing on each side.

Social values, precepts of moral conduct and ethics are neither timeless nor universal. The United States Declaration of Independence advocates for natural rights for all humans. But racial segregation was not completely forbidden in the U.S. until civil rights movements in the 1960s. As [7] writes, no matter how vigorously we assert the timeless universality of human rights and dignities, precepts of moral conduct or ethics, these are neither timeless nor universal; we have had to continually argue about them, fight for them, and change our conceptions of them over time.

I think, similarly, social idealism is rather an incentive to improve the society than an achievable goal. They are incentives for people to act, to ally, and to impact the whole society. They are not achievable or at least directly achievable. All hasty attempts of achieving communism including the community trial in the United States and the nationwide trial in the Soviet Union all failed. But the labor movement in Europe and America has improved labor’s working and living conditions a lot.

It is impossible to optimize society on a single value. Utilitarians tend to convert everything into a single value such as happiness and optimize the total amount of it. But in practice humans have various values which can’t be summed up to evaluate. Nor does moral value. Furthermore, natural intelligence is not the maximization of any single value. The interaction of diverse actors through mutual modeling is what creates the dynamical process we call “intelligence.” This is true even for business enterprises. Business school scholars tend to find a single factor which makes a company long lasting. They can always find a single value which a company optimizes for in a long period through reverse engineering. But that value is rarely contestable because a business enterprise is very complex [7].

Intelligence is both prediction and action in the complex world

Intelligence is to find order in a complex world full of randomness and make predictions. As illustrated in [6], natural intelligence develops through multiple stages: steering, reinforcement learning, simulating the environment, simulating self and peers and language. [7] also emphasize a lot on the importance of simulation especially simulating for social purposes (self and peers). And language is an advanced tool for simulation.

So simulation seems to be the terminal method of prediction.

[4] criticizes traditional economics academia for relying on unrealistic assumptions and evaluating academic contribution based on elegance of theory rather than real prediction power. Main stream economics theories assume too many linear relationships while 90% of relationships in real world are nonlinear. They run simulation based on these linear and simplified agent models and produce really poor prediction on economics. On the other side, [4]’s author, a complexity scientist and physicist, built a multi agents simulation based on simple non linear rules including finds from behavioral economics. The simulation successfully predicted economic outcomes during COVID pandemics in Britain and helped regulators to make appropriate rules to mitigate the damage from pandemics. Therefore, simulation is probably the best way to predict non linear system.

Both [6] and [7] mention that primates not only simulate the physical world, they also simulate themselves and peers. A primate simulates not only peers’ actions and thoughts but also peers’ thoughts on the primate itself. Such simulation contributes greatly to primates’ socialization. This is called theory of mind.

Even different brain regions simulate peer regions’ output signals. For example, visual cortex simulates eye control neurons’ output to predict eyes’ movement in advance so that it could hallucinate what the eye will see in the next few moments. The hallucination might be corrected by real visual input later to improve the brain’s simulation model. Eye neurons also predict the visual cortex’s control signal inputs. Effective mutual prediction involves mutual learning. Such mutual predictions and simulations allow one brain area to learn something first, then teach it to others, whether for lower latency, robustness, parallelism, or greater generality.

Simulation is a powerful prediction tool. But it has several limitations. First, it costs a lot of energy especially when granularity of simulation is high. Secondly, although everyone try to predict others, they also try not to be predicted by others. For example, when escaping from predators or humans, a fly may randomly smash right and left to avoid flying on a predicted path.

Intelligence is not only prediction but also action. The action is used to seek goals based on their predictions. There can be various goals. Survival is the fundamental one. Others include aesthetic goals, curiosity and exploration, morals and so on. As discussed earlier, language can be viewed as an abstraction of relationship graphs. It can be used as a simulation tool, like using next token prediction to simulate on the relationship graph. It can also be used as an action tool. For example, one human can give a speech to influence others through theory of mind.

Free will is the emergence of intelligent action. It consists of 4 elements. Theory of mind lets us build a network of solid tracks along which our minds can venture far into an otherwise marshy future. Randomness provides nudges, letting us wander prospectively into multiple futures. Dynamical instability, like a lubricant, amplifies random nudges and lets us glide anywhere along those tracks, free to go either way at any fork. Finally selection prunes the network to allow efficient long-range planning. [7]

Natural intelligence’s prediction and action essence make the world even more complex. Because each natural agent is trying to make itself less predictable to other agents. This is illustrated in 2 cases in [7].

In the social cooperation case, when mutual prediction is cooperative, it must also be imperfect, for cognitive cooperation involves a division of labor. That means each party brings cognitive resources or inputs to the table that the others lack. If one of the cooperating parties is always perfectly predictable to any of the others, that would imply that the predicted party doesn’t bring anything unique to the table, and is therefore a third wheel.

In adversarial cases such as predator and prey, a prey definitely doesn’t want its escape path to be predicted by predators. And the solution is to add randomness into its escape path.

AI won’t replace humans. It will coexist symbiotically with humans.

AI won’t completely replace humans. A replaces B only when they compete for the same resources and pursue the same goals. But AI is quite different from humans in these senses.

AI doesn’t have exactly the same desires as humans’. Besides common desires like survival needs, an agent’s desires are deeply rooted in its developmental history and its species’ evolutionary history. For example, the environment a species evolves from affects its preference. What you lack in your childhood may decide what you are craving for in your adulthood. As discussed in the previous part, who you are depends on your whole history. Furthermore, human culture and modern civilization are developed from the random interactions among distributed human groups. Different human groups settled in different environments all over the world, developed unique technology and culture and learned from each other through trading, connecting and cruel wars.

AI’s training process is quite different from humans’ developmental history above. First, AI is not born for survival. AI is born as a tool. Neither does AI have the same long evolution history as complex life is mentioned earlier where pro-survival desire features are selected. Secondly, AI is trained from text contents rather than interaction with environment, parents and peers while texts are more of second order thoughts compared to human instincts. Humans rarely put all internal thoughts into text. What humans put in text are usually formal verdict, reasonable arguments, morally acceptable speech, information which may impact other people’s behavior etc. Therefore, AI couldn’t learn the same human desires only from text. Though reinforcement learning teaches AI human preferences through interaction, the training content is usually censored based on high moral standards. And reinforcement learning through text still can’t imitate the real human interactions with peers and the world.

Therefore ai won’t replace humans. AI will probably form symbiotic relationships with humans. However the impact on human society may still be large. Many people's jobs will be replaced by AI. It will pose a huge risk on the economy since if few people make money, fewer people consume. The economy will stall. Universal basic income might be a solution but too difficult to implement. Furthermore, building a powerful black box will create a lot of security risk. Even a prompt with good intentions may have unintended consequences.

The upside is that AI greatly empowers individuals with knowledge. However, you also need a high educational level to take full advantage of AI’ capabilities similar to the idea that your brain regions model other “regions” (AI) output to improve efficiency.

Making the world more complex

Prerequisites for a complex world to evolve is dynamical stability.

In complex systems, stability does not necessarily mean stillness or equilibrium. A system may be dynamically stable if it can absorb perturbations, maintain coherent patterns, or keep its fluctuations within bounds over time.

The biggest factor that could impact human society’s complexity is the political system. A political system is a high level human organization system which evolves to have some level of dynamical stability.

There are two main political systems representing two extremes: centralized totalitarian vs democracy + capitalism.

Centralized totalitarian enforces high restrictions on all members except the emperor to maintain stability of the society. Each society member, especially the majority commoner class, are restricted to work in assigned positions, live in assigned regions and cannot travel too far from their home. Besides spatial and economical restrictions, usually a single school moral and political value are enforced on every member. There may not be an explicit slavery system but the emperor tries to make everyone live like a slaver to him. The majority of individuals' life goals are also simple, either to survive or to move up in the public official ladder. Commerce is usually monopolized or not encouraged by the government because it will enhance grassroot forces.

From dynasty to dynasty, the centralized totalitarian system evolves to be more stable with less power challenging potentials and activities from government officials to the emperor even when the emperor is weak. But the government efficiency still degrades from time to time in each dynasty. The cost of suppressing small uprisings or handling natural disasters is extremely high. The dynasties still collapsed by either large scale internal uprising or weak external forces. When a dynasty collapses, there are usually riots all over the country. A large population of people have to escape their homes for survival and most of them die while fleeing which is like an inferno on the earth.

The root cause of such an inevitable collapse of the centralized totalitarian system lies in the fact that the emperor is the only responsible person for the society and the feedback loop is too long. Only the emperor is responsible for the whole society’s stability. All other government officials are only responsible to their superiors or the emperor. They are not responsible for the region or department they manage. And in a big empire, the government hierarchical tree is usually very high. Each level in the tree distorts information from lower level when reporting to higher level. [3] Though the emperor could send out a personal representative to investigate lower level situations, the information distortion problem still largely exists because local officials could bribe and cheat the emperor’s direct representative. Furthermore, an empire is usually succeeded by heredity. The empire founder usually grows up with a lot of interactions with different levels of officials and citizens so that they establish a good common sense to identify distorted information from officials. But their offspring emperor, especially far offsprings lack such common sense. The distorted information and high hierarchical tree result in a long and malfunctioned feedback loop. For a local official, his reported issue is distorted and delayed when reaching the emperor or central government. And then the handling commands are dispatched down through many levels of officials back to him. [3]

Therefore, the centralized totalitarian system or its weaker form, monarchy system, is neither complex because of it sticking to rigid order and high restriction on individual freedom nor dynamically stable because of the single responsible person and long feedback loop.

Despite these disadvantages, the centralized totalitarian system or hierarchical monarchy system has many advantages. [3] suggests that hierarchical systems are efficient in passing and filtering information from low level to high level. Also, with rigid regulation, the emperor’s order can be better enforced by the whole government system. That’s why before democratic countries prospered in the 19th and 20th century, this system was still the most common political system all over the world. It is more like a local optima of the complex world.

Compared to the centralized totalitarian system, the democracy + capitalism system is less probable while stabler and more complex and prosperous. It is less probable because of many reasons. First, it requires each voter to be mindful and responsible for his vote. Secondly, it is harder to resolve conflicted opinions and form a consensus of public policy direction.

However, the benefit is huge. First, the government system is more stable because there are many local feedback loops to resolve conflicts. Officers in state, county, city and even neighborhood levels are all elected. Local issues can be exposed and addressed locally without going all the way to the president level.

Secondly, though lots of resources are used in reaching consensus, the cost of running and maintaining the whole government is much lower. And less regulations are put on individual freedom. As a result, individuals have more resources to pursue various values other than survival or public official ladder. Though most modern democratic countries are running on capitalism where people mainly pursue money, I would argue that money is the best interaction protocol for a complex system. Most values can be priced in money. If you are good at repairing downpipes while interested in art work, you can sell your labor and buy art work through money. You can also donate money to charitable organizations.

Of course, capitalism has many disadvantages. Many values cannot be priced in money. Tracks that generate short term profits might be overcrowded at the cost of long term stability like pollution. The wealth gap could be enlarged and cause society instability. But in a liberal democratic society, secondary mechanisms such as socialism and environmentalism can be implemented to offset the damage from capitalism.

In essence, the democracy + capitalism system makes the world more complex and prosperous because it offers more diversity in symbiosis.

As stated in [7], the overall tendency of a complex system is toward symbiosis, since that is the most dynamically stable. Maintaining homeostasis often involves competing values and competing opportunities, as well as trade-offs and priorities.” This is even truer in social settings. There are many ways to be alive in this world, but all of them must involve continuing to exist in the future, and existence requires ongoing relationships. Furthermore, based on a simulated complex system, [7] concludes that since symbiogenesis must involve combinations of pre-existing dynamically stable entities, we should expect complex replicating entities to emerge after (and be made of) simpler ones.

The low regulative environment and diverse values in the democracy + capitalism system creates fertile ground for complex symbiogenesis which generates more complex and diverse entities. More people with different expertise and personal values can find their symbiotic position in the whole society. The more interactions (symbiotic opportunity) one has with others, the more diverse desires are created and hence the more intelligence one has. As [7] states, intelligence is an ecology, and the more of a monoculture it devolves into, the less intelligent it will be, and the less interesting life will become. And life only continues to enrich as long as it continues to diversify, and the collective phenomenon of intelligence only grows when diverse sub-intelligences model each other, in the process becoming a greater whole. Therefore with more symbiotic opportunities for individuals, the world becomes more complex and prosperous.

Last but not least, even though the political system might offer more symbiotic opportunities, the complexity of the world still relies on everyone’s passion to make themselves more complex. Though postmodern philosophy renders many traditional life goals meaningless, the one meaningful life goal I find is to pursue complexity of life, adopt diverse values, make long term plans while being flexible to change the plan any time, and interact with the world, the people, as much as possible.

References

1. Melanie Mitchell, Complexity: A Guided Tour,
2. Geoffrey B. West, Scale: The Universal Laws of Growth, Innovation, Sustainability, and the Pace of Life in Organisms, Cities, Economies, and Companies
3. Donella H. Meadows, Diana Wright, Thinking In Systems: A Primer
4. J. Doyne Farmer, Making Sense of Chaos: A Better Economics for a Better World
5. Nick Lane, The Vital Question: Energy, Evolution, and the Origins of Complex Life
6. Max Solomon Bennett, A Brief History of Intelligence: Evolution, AI, and the Five Breakthroughs That Made Our Brains
7. Blaise Aguera y Arcas, What Is Intelligence?: Lessons from AI About Evolution, Computing, and Minds