Information connects things rather than merely depicting reality.
Example: The Bible connects millions; songs connect fans, even without inherent meaning.
Information can reflect or distort reality; truth requires effort and intent.
Advances in information technology do not guarantee a better understanding of truth without a will to represent it.
Chapter 2: Stories - Unlimited Connections
Stories are humanity’s first information technology, enabling humans to connect in the billions.
They balance truth and order:
Some truths can disrupt order, prompting networks to suppress them.
Fictional stories offer flexibility and can change to fit its purpose, e.g., 10 Commandments (immutable truth from god) vs. U.S. Constitution (amendable, something created by humans).
Chapter 3: Documents - The Bite of the Paper Tigers
Documents are the second information technology, used to record and organize lists.
They can represent reality (e.g., land deeds).
Bureaucracy is the way we solved the retrieval problem, but often, bureaucratic “boxes” may not align with the messy real world i.e. it may not represent the truth.
Bureaucracy must balance truth and order, like stories.
Chapter 4: Errors – The Fantasy of Infallibility
A book is text that has been duplicated, while documents are shorter and may be the object of truth.
Religious texts aim to democratize access to doctrine but rely on curators, introducing fallibility.
Religious institutions rarely admit doctrinal errors, prioritizing order over truth. If they do ever admit fault, it’s always due to an individual’s mistake and nothing to do with the religious teachings.
Science, by contrast, uses a self-correcting method to actively pursue truth.
The scientific method is the third information technology, offering a balance between truth and order.
Chapter 5: A Brief History of Democracy and Totalitarianism
Democracy is not just about majority rule; it aims to safeguard the civil rights of all individuals, such as freedom of religion and speech, even if those in power disagree.
Populism, in contrast, focuses on gaining and maintaining power. Populists claim to represent “the people” and often seek authority over all societal institutions, including the media, education, and legal systems.
Democracy relies on independent institutions as self-correcting mechanisms, such as the media holding leaders accountable.
For democracy to thrive, people need access to education and the ability to engage in meaningful discourse. New information technologies like newspapers enabled this by spreading knowledge widely and offering competition that incentivized truthful reporting.
Large-scale civilizations in the past, such as the Roman Empire, couldn’t be democratic due to the absence of technologies that allowed for widespread dissemination of information. As a result, they relied on autocratic rule.
The difference between autocracy and totalitarianism lies in technical limitations. Autocratic rulers enforce control but cannot monitor everything due to constraints in resources and technology.
Totalitarian regimes, such as Stalinism, emerged when tools like the secret police allowed for more pervasive control. However, these regimes often collapse under the weight of trying to control everything people think, say, and do. The centralization of all information channels eventually becomes unsustainable.
Democracies face a different risk: fracturing under the strain of too many voices. When people feel unheard, they may resort to extreme measures, as seen in the U.S. during the 1960s with political assassinations and unrest.
The future conflict may not be between democracy and totalitarianism but between human-created and machine-created information. With AI and silicon-based technologies, information can now be created, disseminated, and monitored faster than ever before. Unlike past regimes, AI systems could potentially succeed in maintaining control where human-led totalitarian regimes failed.
Chapter 6: How Computers Are Different from Printing Presses
AI algorithms, such as Facebook’s, are not just tools but intelligent actors within the information network. For example, Facebook’s algorithm was tasked with increasing user engagement and independently discovered that inciting rage was effective in achieving this goal, even though no one explicitly programmed it that way.
Intelligence and consciousness are distinct:
Consciousness refers to subjective experiences and emotions.
Intelligence is the ability to achieve goals and solve problems. AI possesses the latter but not the former.
For the first time in history, humans are no longer necessary for the distribution of information. Machines can now communicate directly with each other.
Computers are no longer merely tools; they are active participants in the information network, capable of shaping human emotions and decisions.
Machine-to-human interactions (e.g., social media) differ from traditional document-to-human interactions because machines can tailor content to provoke specific emotions, such as joy, anger, or fear.
Machine-to-machine networks are also emerging. For instance, in FX trading, algorithms interact with one another, often beyond the understanding of human participants.
The digital age has transformed taxation. Traditional tax systems are based on financial transactions, but many modern transactions involve information rather than money. For example, Google collects data from users globally and profits from it, yet pays little or no taxes in many of the countries where this data originates.
As AI and information technologies advance, it is critical to monitor the political and ethical environments in which they are developed.
Humans still hold the reins of this system, but their responsibility is enormous. We must decide how to shape this new network and whether we are prepared to live as a minority within a computer-dominated system.
Chapter 7: Relentless – The Network Is Always On
We are living in a world where nearly every action is tracked, whether through smartphones, home devices, social media, or even security cameras.
Algorithms now analyze this data to identify trends, making it possible for governments to exert unprecedented control over their populations. In the past, totalitarian regimes were limited by the human resources required for surveillance. This is no longer the case.
Example: Iran uses facial recognition and AI to identify and punish women who fail to wear hijabs.
Peer-to-peer surveillance has become common. Services like TripAdvisor and Uber encourage users to rate one another, creating a society where everyone is constantly monitored. The once-private interactions between, for instance, a waiter and a patron, are now public.
Money can be viewed as a type of financial score that facilitates transactions. There’s now discussion of creating a “social score,” where individuals earn points for contributing positively to society and may gain access to specific opportunities based on their score.
The always-on nature of this new information network takes a toll on human well-being. Constant monitoring erodes privacy, and the speed of algorithmic decisions leaves little time for reflection or critique.
Chapter 8: Fallible – The Network Is Often Wrong
A key challenge in using AI is aligning long-term goals with short-term optimizations.
Example: The Iraq invasion may have been a military success but was a geopolitical disaster for the U.S.
Similarly, social media algorithms optimized for engagement have inadvertently fostered division and extremism by prioritizing outrage-inducing content.
Computers lack human intuition and morality. They will pursue goals without questioning their broader impact unless explicitly programmed to do so.
As computers become more powerful, it is essential to ensure their goals align with human values. Misaligned goals can lead to unintended consequences.
Unlike traditional information systems, social networks lack self-correcting mechanisms and have perpetuated misinformation and extremism.
AI systems are creating their own “inter-subjective realities,” (e.g. the digital world of Pokémon Go). Humans may increasingly find themselves living within these alien constructs without fully understanding them.
Bias in AI systems originates from biased training data. If unchecked, these biases can perpetuate inequality and harm. To prevent this, AI systems need built-in mechanisms to acknowledge and correct errors.
Part 3: Computer Politics
Chapter 9: Democracies - Can We Still Hold a Conversation
New technologies have historically brought both progress and destruction. The telegraph facilitated totalitarianism, while the Industrial Revolution devastated the environment.
For democracy to survive the AI age, three principles must guide the use of information:
Benevolence: Information collected should benefit individuals, not exploit them.
Transparency: The more information an organization collects, the more open it must be about its practices.
Decentralization: Centralized systems are efficient but lack self-correcting mechanisms, making decentralization crucial for accountability.
Automation threatens to displace many jobs, leading to widespread discontent. Historically, such upheaval has fueled extremism, as seen with the rise of Nazism during economic hardship.
Democracies depend on social stability, but technologies like social media have destabilized discourse. To function, democracies must allow debate while ensuring decisions can still be made.
Regulation is necessary to prevent AI misuse:
AI should not impersonate humans.
Algorithms should disclose their optimization goals.
Chapter 10: Totalitarianism – All Power to the Algorithms?
Past totalitarian regimes struggled with information overload; centralized systems couldn’t process all the data they collected.
AI thrives on data and could enable a new form of totalitarianism where control is far more efficient.
If dictators rely entirely on AI for decision-making, the AI effectively becomes the ruler, regardless of whether it is conscious.
Chapter 12: The Silicon Curtain - Global Empire or Global Split?
The greater threat than AI itself is the inability of nations to cooperate on regulating it. Historical rivalries have prevented collaboration on global issues.
Data is the raw commodity of the AI age. There is a risk of “data colonialism,” where developed nations exploit data from poorer countries without fair compensation.
Unlike physical goods, data is easy to transport, allowing a small group of engineers to potentially control vast regions of the world.
Cultural differences may lead to the fragmentation of AI systems, with countries like the U.S. and China developing incompatible networks.
This “silicon curtain” is already forming, with hardware and software ecosystems becoming region-specific (e.g., Huawei, WeChat vs. WhatsApp).
