AI is weakening our capacity to think, understand, judge

In Part I, I argued that Magnifica Humanitas (MH) correctly identifies many dangers of artificial intelligence (AI). In Part II, I argued that MH’s proposed cure is deeply deficient because it relies too heavily on dialogue, shared responsibility, institutional oversight, global management, and modern social-doctrine language, while failing to ground the response clearly in Divine law, natural law, the Ten Commandments, and the salvation of souls. In Part III, I offered a proposal for what people of good will, using reason and natural law, could recognize about the proper use of AI and the restrictions that should be placed upon it. In Part IV, I offered a proposal for how AI should be judged from the perspective of the Catholic faith.

This fifth part turns from principles to present reality. Before making predictions about where AI is likely going, we should first look at what is already visible.

Dystopia / Tower of Babel 2.0

The current trajectory of AI is dystopian because it is governed by fallen human nature. AI is not a demon, an evil soul, or an independent moral agent, but it magnifies the sins, ambitions, weaknesses, and idolatries of the men who build and deploy it. It makes deception cheaper, surveillance easier, lust more personalized, war faster, censorship more scalable, labor more replaceable, and false religion more interactive. Simply put, AI makes evil much easier at mass scale.

That is already clear from the way AI is being adopted. Corporations are racing to automate labor and capture markets. Governments are exploring digital identity, online safety laws, payment controls, age verification, surveillance, and algorithmic enforcement. Militaries are pursuing AI-selected targeting, drones, cyberwarfare, intelligence analysis, and battlefield decision systems. Criminals are using AI for fraud, impersonation, phishing, and scams. Religious confusion is already appearing through AI spiritual guides, fake clergy, counterfeit religious messages, and machine-generated doctrine.

The direction is clear. AI is already building the Tower of Babel 2.0: mankind using a common technological language to build a world without God and against God. The first Babel was not condemned because men built with brick, but because they used human unity and technical power in prideful independence from God. AI presents the same temptation on a global scale: one system, one language, one managerial intelligence, one tower of control reaching toward heaven and covering all humanity.

Unless AI is placed firmly under Divine law, natural law, family authority, human responsibility, and the Kingship of Christ, the future will not merely be artificial intelligence. It will be artificial domination. The danger is not only that AI will be used by bad men. The deeper danger is that society will be rebuilt around machine systems that make sin, control, deception, and dependency easier than virtue, truth, freedom, and responsibility.

Given the rejection of Catholic faith and morals in society today, there is little supernaturally supported resistance to this trajectory. Many will scoff at the AI commandments suggested in Parts III and IV. They will say such principles are unrealistic, religious, anti-progress, or impossible to enforce. But that is exactly the point. A society that rejects Divine law will not naturally place its most powerful tools under Divine law. It will place them under appetite, profit, ideology, bureaucracy, and force – especially force.

That is Babel 2.0, e-brick by e-brick.

The remaining sections of this part examine the present realities already moving us in that direction: rapid adoption, weak regulation, corporate automation, military acceleration, AI-assisted fraud, digital identity, cyberwar, commercial manipulation, and the growing replacement of human judgment by machine systems.

AI is being imagined in god-like terms

Another present reality is that AI is already being imagined in god-like terms. Not because it is divine, spiritual, conscious, wise, or morally responsible, but because men are beginning to speak of it as an oracle, a genie, a counselor, a providence-machine, and a source of answers to every human problem.

This is one of the most dangerous signs of the age. The temptation is not merely to use AI as a tool, which is how it started. The temptation is to treat it as a power above man: something that sees more, knows more, predicts more, remembers more, speaks more confidently, and answers more quickly than any ordinary human being, all because it “thinks faster” and is “more intelligent”. Once that attitude takes root, AI stops being treated as a servant and begins to be treated as a false authority, a master, yes, a “god.”

The language surrounding AI often encourages this. AI is promised as a system that will increase productivity, creativity, scientific progress, and economic opportunity for humanity. Some technologists speak as though artificial intelligence will solve disease, poverty, education, work, science, loneliness, and even the limits of human life itself. In popular language, AI is often described as an “oracle” or a “superintelligence.” In a video online, Sam Altman said “we are close to creating a genie that can grant any wish”. That language is spiritually dangerous.

The AI genie is increasingly imagined not merely as a tool, but as a wish-granting power: something one consults in order to obtain knowledge, productivity, comfort, control, or fulfillment.  Put another way, it tempts man to treat prompts as prayers. That is why the “AI genie” image is so revealing. The old pagan temptation was to summon a spirit to obtain hidden knowledge or power. The modern technological temptation is to summon (send prompts into) a machine to obtain the same thing. In both cases, the soul is tempted to seek mastery rather than obedience.

The genie metaphor is especially revealing from a Catholic standpoint. A genie is not merely a tool. Rather, it is a spirit-like power that man summons and commands in order to obtain hidden knowledge, power, protection, pleasure, or fulfillment. That is the logic of the occult. Catholic teaching rejects divination, magic, sorcery, and the attempt to tame external powers and place them at man’s service. The problem is not only the external ritual. The problem is the disordered desire underneath it: knowledge without humility, power without obedience, and control apart from God.

Another religious danger is already visible. In the Chat With God app, you can literally pick your god, by selecting from Christianity, Islam, Judaism, Hinduism, Buddhism, Sihkism, Agnostic/Spiritual or Exploring/Curious, Taoism, Indigenous, and more. Once selected, you can “talk” to him in your tradition. In a Christian focused app called Isaiah, the app-store description says its AI companion is for “anyone longing to feel close to God again” and “anyone who just needs someone to talk to.” Even the machine wants to become God.

This is not Catholic spiritual direction. It is machine-generated faux religious faux companionship. A machine can imitate religious language, but it cannot possess wisdom, grace, supernatural charity, pastoral authority, or discernment of spirits. It cannot know the soul before God. It cannot judge sin, call to repentance with authority, absolve, bless, or guide a person as a true spiritual father. Rather the AI god will tell you what you want to hear.

The deeper issue is that modern man wants the benefits of God without obedience to God. He wants knowledge without humility, power without repentance, comfort without conversion, guidance without authority, and desires without worship. AI fits that temptation perfectly. It offers answers without prayer, counsel without priesthood, moral language without moral law, and apparent wisdom without the Cross.

AI must not become an oracle or a genie. It must not become a false providence, and certainly not become the voice man consults instead of God, conscience, Scripture, Tradition, the Church, fathers, mothers, priests, teachers, and wise human counsel.

A Catholic response must say plainly: AI may assist man, but it must never be worshipped, trusted as ultimate authority, or treated as a substitute for God.

AI adoption is outrunning regulation

This is the normal pattern of any major new technology. The telephone, radio, automobile, television, cable TV, the internet, smartphones, and social media all spread before law and culture understood their full effects. First comes invention, then adoption, then disruption, and only later regulation. AI follows the same pattern, but with greater speed, scale, and reach.

A telephone carried the human voice while a television delivered video. AI can generate speech, imitate voices, write code, produce images, make recommendations, classify people, summarize evidence, automate fraud, assist diagnosis, guide drones, censor content, and act inside business, education, medicine, law, war, religion, and family life. It is not one device or one medium, but instead a layer of machine decision-making being added to nearly every human system.

AI regulation is not absent, but it is fragmented, delayed, uneven, and far behind deployment.

Around the world, many governments are talking about AI regulation. Trackers now list dozens of countries with AI strategies, bills, policies, principles, guidance documents, or proposed frameworks. But relatively few jurisdictions have comprehensive, binding AI laws already in force.

That means AI is not entering a world of clear, unified, enforceable moral governance. It is entering a world of scattered laws, draft bills, voluntary frameworks, agency guidance, national strategies, corporate promises, and regulatory experiments. The deeper problem is speed. By the time lawmakers understand one use case, ten more are already in production. By the time one harm is debated, the technology has shifted form.

So the reality is not “no regulation.” The reality is more complicated: some regulation exists, but it is late, partial, political, inconsistent, and vulnerable to regulatory capture. The same corporations, governments, militaries, and institutions that profit from AI, depend on AI, or want to weaponize AI will also help shape the rules that supposedly govern it.

Corporations are replacing human contact with AI systems

Given the corporate drive to increase profitability, it is easy to see how many jobs will be replaced or hollowed out by AI agents. This is not the first time communications technology has changed labor. When international telephone calling became cheap enough, many customer service departments were moved overseas because it was less expensive to hire lower-cost labor in foreign countries. The customer still spoke to a human being, but that human being was now farther away, more scripted, and essentially cheaper.

AI takes the next step. The first contact will no longer be a lower-cost person overseas. It will be a machine.

Goldman Sachs has estimated  that generative AI could expose the equivalent of 300 million full-time jobs globally to automation. McKinsey has warned that automation’s biggest effects are likely to hit office support, customer service, and related categories. These forecasts vary in detail, but they all point in the same direction: AI will not merely assist workers. In many settings, it will replace the first layer of human contact.

Customer service is the obvious example. Klarna announced in 2024 that its AI assistant handled two-thirds of its customer-service chats in its first month, covering about 2.3 million conversations, and was doing the equivalent work of 700 full-time agents. The company also said the assistant was available in 23 markets, communicated in more than 35 languages, reduced average resolution time from 11 minutes to under 2 minutes, and was expected to drive a $40 million profit improvement in 2024. That is exactly the incentive structure corporations understand: fewer workers, faster responses, more languages, lower costs, and higher profits.

A company can use AI chatbots, voice agents, automated ticket systems, scripted escalation flows, and sentiment analysis to reduce the number of human beings who ever speak to the customer. The system can answer common questions, deny requests, delay refunds, route complaints, enforce policy, collect data, and wear the customer down before any human employee is reached. What used to be outsourced to a foreign call center can now be outsourced to a machine. Imagine arguing with C-3PO, who is fluent in six million forms of communication, about a refund, in your language, with an accent and dialect tuned to you.

This will not be limited to customer service. AI agents will be used in sales, billing, HR, insurance claims, banking support, medical intake, legal intake, tech support, education, scheduling, collections, compliance, and government services. Many people will still have jobs, but the human part of the work will shrink. One employee, assisted by AI, may do what several employees previously did.

Military AI is already thinning the kill chain

Military AI is not merely a future danger. It is already being deployed in the systems by which targets are found, tracked, classified, prioritized, and attacked. The common reassurance is that a human remains “in the loop.” But that phrase can conceal how much of the kill chain has already been handed over to machines.

Ukraine shows where this is going. Reuters has reported that Ukrainian drone pilots are using AI-assisted targeting systems, while Ukraine says any decision to strike must still be made by a human. Ukraine’s Ministry of Defence has described next-generation interceptor drones that can autonomously identify, track, and engage enemy Shahed-type drones. Ukraine’s Brave1 platform says one system automates 95 percent of the interception process, while still claiming that the human operator can call off the attack at any moment.

The moral danger is obvious. AI can make killing faster, more remote, more automated, and easier to justify as technical necessity. It can reduce the enemy to a heat signature, vehicle shape, movement pattern, risk score, or target class. It can also make mistakes at machine speed. A civilian truck, ambulance, farm vehicle, aid convoy, or fleeing family could become a pattern to be classified rather than a human reality to be judged.

Fraud and impersonation on the rise

AI will make fraud more convincing because it attacks the ordinary signs by which people recognize reality. A familiar voice, a familiar face, a video call, a written message, a document, a photograph, or an urgent request from a trusted person can no longer be accepted as confidently as before.

This is already happening. In Hong Kong, a finance employee at a multinational company was tricked into transferring about HK$200 million, or roughly US$25 million, after joining a video call with what appeared to be the company’s CFO and other employees. The people on the call were reportedly deepfake recreations. The employee did not merely receive a suspicious email. He saw and heard what appeared to be company leadership authorizing the transaction.

Families are also being targeted. AI voice-cloning scams have been used to make parents and grandparents believe a child or loved one is kidnapped, injured, arrested, or in immediate danger. In one reported case, an Arizona family believed they were hearing a daughter’s voice in a kidnapping and ransom call, only to discover it was a scam created with artificial intelligence.

These examples show the deeper danger. AI fraud does not merely steal money. It attacks trust. It teaches people that a voice may not be a voice, a face may not be a face, a video call may not be a meeting, and an urgent cry for help may not come from the person it imitates.

Ordinary people will be forced to ask again and again: “Is this real?”

Governments are pushing safety, age verification, digital ID, and payments as control infrastructure

Because of the fraud capabilities enabled by AI, governments are promoting a solution where total AI surveillance and protection are necessary. The stated reasons will sound reasonable: child safety, fraud prevention, anti-money laundering, cybersecurity, convenience, financial inclusion, welfare delivery, online safety, protection from pornography, protection from scams, and secure access to services. Some of those concerns are real. Children should be protected, fraud should be stopped, and payments should be secure. Government services should not be stolen through identity theft.

But the same infrastructure that verifies identity can also control access.

This is explained well by my LifeSite colleague Jeremy Williamson. The first step is Age Verification Laws, which are being implemented now.

China is the clearest warning. Its social-credit, surveillance, censorship, digital-payment, and online-identity systems show how digital infrastructure can be used not merely to provide services, but to monitor and discipline society. A Canadian Security Intelligence Service publication warns that China’s developing social-credit system makes possible comprehensive data collection to measure individual loyalty to the state, extending the Communist Party’s control over citizens’ lives. China has also moved toward state-managed digital identity for online activity, justified as privacy and fraud protection, but criticized as giving the state a more centralized way to identify users across platforms.

Other countries are not simply copying China, but many are building pieces of the same digital architecture. The European Union has adopted a Digital Identity Regulation requiring member states to provide EU Digital Identity Wallets by the end of 2026. The EU is also developing age-verification systems so users can prove they are old enough to access age-restricted online content such as pornography, gambling, and alcohol-related services. The United Kingdom’s Online Safety Act has already required strong age checks for pornography services. Australia has imposed social-media age restrictions requiring platforms to take reasonable steps to prevent children under 16 from having accounts. In the United States, state-level age-verification laws for pornography and social media are spreading, and the Supreme Court upheld Texas’ pornography age-verification law in 2025.

The pattern is clear. The infrastructure is being promoted as safe, inclusive, convenient, efficient, and necessary. But once identity, payments, speech, age verification, welfare access, travel, banking, health records, social media, and government services are tied together digitally, the same system that protects can also exclude. This is the danger.

Cyberwar and cybercrime are already AI against AI

Cyberwar and cybercrime are increasingly AI against AI. Google’s Threat Intelligence Group reports that threat actors are using AI for research, phishing, coding, and malware-related activity, while Google itself uses AI defensively through tools designed to detect software vulnerabilities and help fix them. Microsoft’s 2025 Digital Defense Report likewise says that AI adoption benefits both defenders and threat actors. Canada’s cyber agency says “As cyber criminals will likely adopt frontier AI, organizations should consider AI-native defences to keep pace with the evolving threat landscape.”

That is the new battlefield. The attacker uses AI to find weaknesses, write convincing messages, imitate people, generate code, test defenses, and scale attacks. The defender uses AI to monitor networks, detect anomalies, block suspicious behavior, patch vulnerabilities, and respond faster than human analysts can. Then the attacker uses better AI to evade the defender’s AI, and the defender uses better AI to catch the attacker’s AI.

This is not merely hacking with better tools. It is the beginning of machine-speed conflict. The human being may still set the goal, approve the operation, or interpret the result, but much of the fight will happen faster than ordinary human judgment can follow.

AI itself is being attacked

There is another danger: AI is not only being used by attackers against human victims. AI itself is becoming a target.

If a government, bank, hospital, military unit, charity, school, court, or corporation relies on AI to detect fraud, summarize evidence, flag threats, approve payments, monitor networks, guide decisions, or defend systems, then corrupting that AI becomes an obvious objective. The attacker will not merely try to evade the machine. He will try to poison it, manipulate it, deceive it, or compromise the systems it depends on.

One example is prompt injection. Malicious instructions can be hidden inside text, documents, websites, emails, or other material the AI reads. The user may think the AI is merely summarizing or analyzing content, while the content itself is secretly instructing the AI to ignore rules, reveal data, misuse tools, or follow the attacker’s commands. Researchers have already reported prompt-injection attacks capable of exfiltrating personal information from AI systems.

A second example is the AI supply chain. Models, datasets, plugins, tools, and dependencies can themselves be compromised. Researchers studying model hubs such as Hugging Face have found malicious models and dataset-loading scripts capable of threats such as reverse shells, browser credential theft, and system reconnaissance. Hugging Face itself warns that some model formats, such as pickle-based files, can execute arbitrary code when loaded, which is why safer formats such as “safetensors” are recommended.

This makes AI security much harder than ordinary software security. With traditional malware, one may be able to identify the malicious file, remove it, patch the system, and restore from backup. But if the AI system itself has been corrupted, the poison may be buried in prompts, memory, training data, fine-tuning data, embeddings, model weights, retrieval databases, plugins, tools, or vendor systems.

How does one detect a poisoned AI? And how does one remove the poison? Those questions are much harder than ordinary users, managers, courts, governments, or even many technologists realize. The next cyberwar will not merely ask, “Can we stop the attack?” It will ask, “Can we still trust the machine that tells us what an attack is?”

AI answers are already being warped by advertising and commercial pressure

AI answers are not shaped only by truth, usefulness, or wisdom. They are also being shaped by money. Google alone generated more than 70 percent of its 2025 revenue from online advertising, meaning advertising produced well over $280 billion in a single year. Does anyone really believe that Google will simply give that up and build a purely objective, commercially neutral AI?

The old internet already had this problem. Google search created a vast economy around keywords, ranking, ads, sponsored placement, search-engine optimization, affiliate links, and commercial persuasion. Websites added features for Search Engine Optimization (SEO). The user thought he was searching for information, but he was also moving through an advertising-shaped environment. AI is not abolishing that economy. It is moving it into the answer itself.

Google says AI Mode surfaces shopping recommendations, and it is testing sponsored retailer placements inside that AI search experience by “reinventing what an ad is.” Google also says ads can appear above or below AI Overviews, and may appear within AI Overviews on desktop in the United States when relevant. The point is not merely that ads appear beside answers. The point is that advertising is being integrated into the environment where the answer is generated, summarized, framed, and acted upon.

A second sign is the rise of “Answer Engine Optimization” (AEO) and “Generative Engine Optimization” (GEO). Companies are already building tools to measure how brands appear in ChatGPT, Gemini, Perplexity, and other AI systems. HubSpot’s AEO tool, for example, will “track your AI visibility” and provide recommendations to improve it. Academic work on “Generative Engine Optimization” argues that brands must now “engineer content” using a “new paradigm” for “machine scannability and justification” that will “dominate earned media to build AI-perceived authority”.

That is the new commercial battlefield. The old question was, “How do I rank on Google?” The new question is, “How do I get the AI to recommend me, cite me, summarize me favorably, exclude my competitor, and turn the user toward my product, ideology, or institution?”

This does not require a crude conspiracy in which advertisers secretly rewrite every answer. The danger is more subtle. AI models themselves are being trained, ranked, filtered, tuned, sourced, and monetized in ways that favor commercial priorities, especially those connected to advertising revenue. Which sources are trusted? Which products are recommended? Which claims are emphasized? Which competitors are omitted? Which answers keep the user inside the platform? Which recommendations lead to sponsored purchases?

Advertising-funded AI is not neutral. It is shaped by the business model that pays for it. The keyword economy is becoming the answer-shaping economy. The salesman is no longer beside the answer. He is being built into the answer, and he may even speak to you in your own language, accent, and dialect.

AI is dumbing down society

I referred in earlier parts to the film Idiocracy, where the human race is no longer capable of much thinking because it has become dependent on machines and systems that do the thinking for it. The movie is satire, but the danger is real. Cognitive psychology has long found that retrieval practice improves retention and comprehension, and the generation effect means people remember self-generated material better than material simply provided to them.

In an article titled “I Stopped Using ChatGPT for 30 Days. What Happened to My Brain Was Terrifying”, the author writes, “That’s when I realized: I couldn’t think without AI anymore,” and, “The muscle of thinking is atrophying in real-time.”

That is the danger. It is not necessarily physical brain atrophy in the strict medical sense. It is something more ordinary and therefore more dangerous: skill atrophy through disuse. And the skill being weakened is thinking itself.

Automation bias adds to the problem. People tend to over-trust automated systems and favor their recommendations over their own judgment, even when the machine is wrong. That matters because the more a person gets used to accepting AI phrasing, AI summaries, AI judgments, and AI conclusions, the more he may lose the habit of mentally challenging them.

Entry-level jobs are especially vulnerable to AI because they often consist of the routine tasks by which young workers learn judgment, communication, responsibility, and professional habits. Software development is one warning sign. GitHub Copilot and similar tools are openly marketed as AI coding assistants that help developers generate, understand, refactor, and debug code in real time. That may help experienced developers, but it also threatens the old apprenticeship path. A Stanford working paper found that early-career workers ages 22 to 25 in the most AI-exposed occupations experienced a 13 percent relative decline in employment after the widespread adoption of generative AI, even after controlling for firm-level shocks. Another 2026 study on software engineering warned that generative AI redirects entry-level work into senior-AI workflows, leaving juniors with less of the productive struggle through which expertise once developed.

That is the deeper danger. If entry-level work is greatly reduced, how will workers learn enough to become senior-level workers knowledgeable enough to run the machine? If the junior developer no longer writes the ordinary code, fixes the bugs, reads the errors, or struggles through the everyday problem, then how does he become the senior engineer who can judge whether the AI is right?

A company that uses AI to eliminate entry-level formation is like a man sitting on the end of a tree branch while sawing it off. It may look efficient for the moment. But the branch he cuts away is the very thing holding him up. The worker may temporarily gain output while losing formation, and he may temporarily appear more productive while becoming less capable.

Neuroscience and cognitive-aging research support a “use it or lose it” principle. One study found that greater mental activity across life predicted a slower rate of hippocampal atrophy, suggesting that mental work may help protect brain structures involved in memory. Studies on retirement likewise support the concern: when structured mental demands are removed, cognitive decline can accelerate, especially in memory. Other research has found that occupations with higher mental demands are associated with higher cognitive functioning before retirement and slower cognitive decline after retirement.

If AI routinely performs the acts of remembering, searching, outlining, wording, reasoning, comparing, judging, and correcting, then the user may become more productive for a time while exercising fewer of the mental habits by which intelligence is formed.

The hype is being exposed

AI has been promoted as a tool that can solve almost any problem, increase the efficiency of almost every process, and improve nearly every job. It is promised as a revolution in search, education, medicine, law, customer service, software development, advertising, warfare, cybersecurity, administration, creativity, science, and personal productivity.

But now that AI is finding its way deeper into the mechanisms of daily life, the hype is being tested against reality.

AI can produce useful work. That should be admitted plainly. It can summarize, translate, classify, draft, search, compare, generate code, detect patterns, and assist human workers. But it also hallucinates (makes up things that look feasible), fabricates sources, misunderstands context, flatters the user, hides uncertainty, reflects bias, gives confident wrong answers, and often requires human correction. The promised efficiency is real only when a knowledgeable human being is present to judge the output.

That is the key point. AI often works best for the person who already understands the subject. A good programmer can use AI to move faster because he can judge whether the code is right. A good writer can use AI to draft and revise because he can detect weak arguments, false claims, bad tone, and empty prose. A good doctor, lawyer, engineer, teacher, or researcher can use AI as an assistant because he already possesses the knowledge needed to supervise it.

But that is very different from replacing expertise. The beginner may not know when the AI is wrong. The manager may not know when the AI has hidden risk. The customer may not know when the chatbot is merely enforcing a policy. The student may not know when the answer is shallow. The court may not know when a summary has distorted the evidence. The reader may not know when a confident paragraph rests on nothing.

So the hype is being exposed in two directions. AI is powerful, but not wise. It is fast, but not necessarily true. It is fluent, but not understanding. It can assist human judgment, but it cannot replace the moral and intellectual formation that makes judgment possible.

The more AI enters real life, the clearer the truth becomes: it is powerful enough to be dangerous, but not wise enough to be fully trusted. This is another reason AI must remain a servant.

Conclusion

There are, of course, more AI-related developments than can be discussed here, but these are the main highlights. AI is not merely a future concern. It is already being inserted into the ordinary systems by which modern life operates: search, advertising, customer service, employment, warfare, cybersecurity, fraud, education, religion, government services, and the formation of human judgment itself.

The pattern is already visible. AI adoption is outrunning regulation. The race for capability rewards speed over wisdom. Corporations are replacing human contact with machine systems. Military AI is weakening human control over the kill chain. Fraud and impersonation are forcing people to authenticate reality itself. Governments are building digital identity, age-verification, safety, and payment systems that purportedly can protect, but can also control. Cyberwar and cybercrime are becoming AI against AI. AI itself is being attacked. Advertising and commercial pressure are moving from the side of the answer into the answer itself. And the human mind is being tempted to trade formation for convenience.

These are not isolated developments. They are parts of one trajectory. AI is becoming a layer of machine generation, classification, persuasion, surveillance, and decision-making added to nearly every human system.

That is why Catholic moral judgment is urgently needed. If AI remains a tool under truth, responsibility, natural law, Divine law, family authority, and the Kingship of Christ, it may serve real human goods. But if it is governed mainly by profit, ideology, appetite, state power, military competition, and technological pride, then the trajectory is clear.

The present realities already point toward Babel 2.0.

The next part will consider where this path is likely leading.