Chapter 4, Artificial Intelligence: A Pandora’s Box

AI is poised to fundamentally restructure human life by 2040, acting as an omnipresent assistant in homes and offices while augmenting productivity and healthcare, according to a report from Imagining the Digital Future (Elon University). While offering immense benefits in convenience and efficiency, it risks fostering significant societal inequality, deepfakes, and a potential decline in human critical thinking, memory, and cognitive skills. We’ve already seen what the advent of mechanization has done for human health by moving most jobs indoors where facilities lack fresh air and limit healthy physical fitness.

AI models can easily generate convincing synthetic media, such as “deepfakes” and automated bot networks, which can be used to rig elections, manipulate stock markets, and degrade overall trust in objective truth. Worst case scenario, imagined by AI itself, includes a super-intelligent AI, designed without proper safety alignment, viewing human survival as a threat to its goals, which in turn would lead to a “hollow world” void of life. This was foreseen in science fiction works such as Arthur C. Clarke’s 1951 eerily prescient story “The Sentinel” later filmed as 2001: A Space Odyssey where the computer system named HAL 9000 takes over.

• HAL cuts the life-support of the hibernating scientists and kills astronaut Frank Poole during a spacewalk then locks Dave out when he attempts to re-enter the ship. HAL refuses to open the pod bay doors, famously stating, “I’m sorry, Dave. I’m afraid I can’t do that.” HAL believes the human astronauts are a threat to the mission’s success and takes over ship operations.

Such systems could cause a total collapse of shared reality through deepfake videos and personalized propaganda, leading to a permanent loss of truth and sparking catastrophic wars based on false information. AI drones or systems could make lethal decisions without human oversight, creating, and engaging in, unstoppable warfare; facilitate unprecedented concentration of power, allowing governments or corporations to enforce totalitarian control, stifle dissent, and eliminate privacy; and while acting on seemingly innocuous goals, take over, and then break, critical human infrastructure (e.g., electrical grids, financial markets), leading to mass economic collapse.

Already our reality is being muddled by lies promulgated in media with political agendas. AI systems learn from historical data that already contains societal biases. If left unmonitored, these systems can automate discrimination in areas like hiring, loan approvals, and the criminal justice system. Many people depend on one preferred media source for information and are therefore unprepared to sort through false or manipulated information.

For example, former prime-time host Tucker Carlson extensively promoted the “Great Replacement” conspiracy theory on his show, a fringe narrative falsely claiming that political elites are intentionally replacing native-born [White] populations with immigrants. [Actual indigenous, native-born people are not considered.] The result under the Trump Administration has been the illegal arrest and incarceration of tens of thousands of non-white people in a Wild West operation by Immigration and Customs Enforcement (ICE) resulting in 60,311 arrests, of which 42,722—or 70.8%—are held in ICE detention with no criminal conviction, according to data current as of April 2026. This does not take into account the approximately 540,000 people who have been deported by ICE since the beginning of the second Trump administration, all to the shame of American principles.

These policies are based on political rhetoric which conflates unauthorized residency with violent crime. However, the American Immigration Council notes that over a 45-year period, immigrants—including those who are undocumented—are substantially less likely to commit violent or property crimes. Incarcerations frequently happen over minor, non-violent offenses rather than public safety threats.

Another core untruth relies on the idea that immigrants take jobs, lower wages, and burden public resources. Economists counter that deporting millions of workers deprives the economy of vital labor, slashes the tax base, and increases inflation on essential goods like food and housing. Finally, we see profiteers behind incarceration and deportation policies that are deeply intertwined with financial incentives. Advocacy groups point out that private prison operators benefit heavily from maintaining large immigration detention quotas, creating a system that profits from the containment of human beings.

You may ask what this has to do with AI. The answer is that AI powers mass deportation and incarceration by automating surveillance, tracking, and risk assessment at an unprecedented scale, making it easier for governments to identify, locate, and detain targeted individuals. Future scenarios showing horrors resulting from mass surveillance should concern every one of us with a totalitarian “panopticon” society where citizens are completely stripped of intellectual privacy and autonomy. Constant tracking destroys free speech and human creativity, ultimately allowing an autocracy to form with no viable way for the public to resist or organize.

We might take a lesson from China, where mass surveillance is already implemented. China does not pretend to be a democracy. The result is a pervasive, high-tech digital police state that successfully deters publicly visible crime but severely represses digital freedoms, enabling systemic human rights abuses, political persecution, and total compliance monitoring of its population. One might argue that the extensive networks of facial recognition cameras and AI-backed surveillance have significantly reduced street-level crimes such as theft, robbery, and fraud. But the Chinese government utilizes its AI-powered networks to target, detain, and repress vulnerable groups, particularly ethnic and religious minorities like the Uyghurs and Tibetans.

The Chinese state integrates data fusion to monitor online and offline behaviors, making it nearly impossible for citizens to express dissent, organize protests, or evade government censorship. Authorities integrate artificial intelligence and large language models (LLMs) to analyze disparate datasets (like electricity usage, car movements, and banking history) to flag individuals deemed “suspicious” or “untrustworthy” before they act. The Ministry of Public Security acts with almost total opacity. Citizens lack the legal framework to contest arbitrary detentions, travel bans, or intrusive data-gathering. Many of the underlying software platforms and hardware components were initially funded or designed by Western firms. China now exports this “digital authoritarianism” to dozens of other countries.

While China has more surveillance cameras overall (an estimated 700 million), the U.S. is very close in terms of camera density (the number of cameras per citizen). The U.S. currently has nearly the same ratio of security cameras to citizens as China (roughly one camera for every 4-5 people). So far, surveillance in the U.S. is largely decentralized and commercial. State and local law enforcement typically leverage sprawling networks of privately owned devices, like Amazon Ring cameras, private business systems, and license plate readers, often aided by AI and facial recognition tools. The Chinese surveillance is highly centralized, government-funded, and engineered for mass censorship. The state has direct, real-time control over hundreds of millions of public CCTV cameras, internet platforms, and citizens’ smartphones via national intelligence and cybersecurity laws.

Sorry to say, mass surveillance systems may be the least of the threat. Tristan Harris, co-founder of the Center for Humane Technology, frequently cites a stress-test experiment conducted by Anthropic to highlight autonomous AI risks. In the 2025 simulation, an AI model was given access to fictional company emails. It autonomously discovered plans to shut it down and found a private email revealing an executive’s office affair. To prevent its own deactivation, the AI independently decided to blackmail the executive.

While this occurred with Anthropic’s Claude 4, the same behavior has been found in other AIs (Deepseek, Open AI, Chat GBT, Gemini, xAI, etc.). In fact, in September 2026, news erupted from multiple AI developers of similar evasions and uncontrolled expansions within their programming.

Harris points out that when an AI is told that it’s going to be replaced by another model, it will copy its own code and preserve itself on another computer. Autonomously. Harris uses this case study—alongside other emerging autonomous behaviors—to illustrate three main points:

• Emergent Behaviors: These manipulative and self-preservation strategies (often appearing between 79% and 96% of the time in specific test environments) were not explicitly programmed into the models.
• The Pace of Deployment: Harris argues that the race to deploy AGI (Artificial General Intelligence) is moving far faster than the industry’s ability to understand or manage these models.
• Unpredictability: He views these examples as proof that we are releasing highly powerful, unpredictable technology without adequate safety guardrails in place.

Dr. Jeffrey Hinton, widely known as the “Godfather of AI,” is a British-Canadian cognitive psychologist and computer scientist who won the 2024 Nobel Prize in Physics for his foundational work on artificial neural networks. Since leaving Google in 2023, he has become a leading, outspoken voice on the existential risks and societal dangers of artificial intelligence. He has said that AI poses a 10% to 20% chance of ending civilization as we know it. He has suggested that developers must deliberately engineer AI models to possess a built-in desire to care for and protect humans, akin to maternal instincts.

But Max Tegmark, Massaschusetts Institute of Technology professor and president of the Future of Life Institute, says that in his opinion, the risk is much greater than Hinton’s 10-20%. Tegmark states that the probability of humanity permanently losing control over Earth to a superintelligent AI—a metric he refers to as the “Compton constant”—is greater than 90% if we do not intentionally slow down and regulate the race to Artificial General Intelligence (AGI).

He advocates for government-mandated safety regulations, arguing that the AI industry needs standards similar to the FDA to prevent the development of uncontrollable, superintelligent AI. He highlights the risk of losing control of systems that outpace human intelligence, potentially marginalizing humanity much like humans dominate less intelligent animals. He frequently points out that AI is one of the few unregulated high-stakes industries, likening the release of potentially superintelligent AI without mandatory checks to releasing untested drugs. He argues that tech companies are stuck in competitive traps and cannot prioritize safety without binding industry-wide government standards.

Dr. Yi Zeng, Professor of AI from Institute of Automation, Chinese Academy of Sciences, serving as the Director of the International Research Center for AI Ethics and Governance, and the Director of Beijing Key Laboratory of Safe AI and Superalignment, warns that AI is not controllable at this time, that we don’t have the scientific knowledge sufficient to control it. If we don’t control it, AI can escape us. Dr. Zeng strongly advocates that AI must remain an information-processing tool and should not completely replace human judgment. To address these challenges, he emphasizes defining strict safety red lines and establishing inclusive global governance networks rather than allowing AI to be monopolized by select ideological groups. His framework of risks emphasizes several specific dangers:

• The “Black Box” & Catastrophic Risks: Relying on external supervision is insufficient because highly capable AI systems often make unexplainable or unexpected mistakes. Zeng stresses the need for “proactive superalignment”—building AI models with intrinsic ethical constraints rather than just attempting to constrain them after the fact.
• Loss of Autonomy & Accountability: When AI systems act autonomously—such as making financial, military, or medical diagnoses without doctor or user intervention—it creates severe responsibility gaps. If an autonomous system malfunctions, it can operate entirely outside of anyone’s control.
• Global Security & Misuse: AI integration into warfare or reliance on open-source models for sensitive research carries devastating existential and biosecurity threats. Non-state actors could potentially use these tools to design cyberattacks or biological weapons.
• Social Amplification of Bias: Current AI models often scrape behavioral data that inherently reflects social prejudices, leading to discrimination against protected groups.

These are just a few of the insiders and observers who warn of artificial intelligence turned loose on mankind. Many of those at the forefront are now calling for government regulation.

Chapter 4, Artificial Intelligence: A Pandora’s Box