We are living through a period of technological change that is moving faster than our ability to fully absorb it.
Artificial intelligence is no longer limited to answering questions or generating text. AI systems can write software, analyze large bodies of information, create convincing images and video, conduct research, and operate computers. AI agents take this further. Instead of simply responding to prompts, these systems can be given an objective, determine intermediate steps, use external tools, and carry out tasks with increasing degrees of independence.
The pace of improvement is significant. Systems are still unreliable and regularly make mistakes, but they are becoming increasingly capable of acting rather than merely advising.
Most of the discussion around AI naturally focuses on capability. What jobs will it replace? How intelligent will the models become? What industries will change? How much productivity will it create? How dangerous could highly autonomous systems eventually become?
These are legitimate questions. But there is another side of the discussion that receives less attention. The development of increasingly capable machines is happening at the same time that human beings are becoming accustomed to delegating more parts of their lives to technology.
There is no serious case for rejecting technological progress. Artificial intelligence could help accelerate medical research, improve education, reduce tedious administrative work, expand access to knowledge, and give individuals capabilities that once belonged only to large organizations.
But progress is not only a question of what our tools can do. It is also a question of what those tools encourage us to become.
This is where some very old philosophical ideas become unexpectedly relevant.
Seneca wrote about suffering and the limits of human control. Kierkegaard examined the anxiety that comes with freedom. Abu Hamid al-Ghazali questioned inherited belief and even the apparent certainty of sensory and rational knowledge in his search for genuine certainty. Diogenes challenged the endless creation of unnecessary wants. Camus asked how human beings should live in a world that does not provide them with a final explanation. Kant argued that enlightenment required the courage to think for oneself rather than surrender judgment to others.
None of them knew anything about artificial intelligence.
Yet the human tendencies they examined have changed far less than our technology has.
Suffering
In Consolation to Marcia, Seneca writes to a Roman woman who had lost her son, Metilius. Three years after his death, Marcia remained overwhelmed by grief.
Seneca’s response can sound severe to a modern reader. He does not tell her that she should not mourn. His concern is that grief has become permanent, occupying her life rather than passing through it.
He contrasts Marcia with two other Roman women who had lost sons: Octavia and Livia. Octavia never truly emerged from mourning. She withdrew from the world and resisted consolation. Livia also suffered deeply, but eventually returned to ordinary life while continuing to remember and speak about her son.
The distinction matters because Stoicism is often misunderstood as a philosophy of emotional suppression. Seneca is not arguing that a person should become indifferent to death, loss, or disappointment. His argument is that suffering should not be granted unlimited authority over the rest of one’s life.
There are things human beings can control and things they cannot.
Once an irreversible event has happened, no amount of anger, regret, or grief can change the event itself. If tears could restore Marcia’s son, Seneca argues, then she should never stop crying. But they cannot.
The remaining sphere of control is the person’s response.
Seneca uses the image of a sailor in a storm. A good pilot cannot command the sea, stop the wind, or guarantee that the ship survives. What the pilot can do is remain at the helm.
The ship may still sink.
Stoicism does not promise victory over circumstance. It insists that there remains a difference between being defeated by an event and surrendering one’s entire inner life to it.
This idea has become particularly relevant in a society increasingly structured around removing discomfort.
Modern technology has eliminated enormous amounts of unnecessary suffering. Medicine has prevented deaths that would once have been considered inevitable. Machines have replaced forms of labor that destroyed bodies. Communication technologies allow people separated by continents to speak instantly. Information that once required specialized institutions is available within seconds.
These are genuine achievements.
The problem begins when reducing unnecessary suffering becomes confused with the idea that all discomfort is unnecessary.
Some experiences cannot be engineered away.
People die. Relationships end. Careers fail. Bodies deteriorate. People disappoint one another and themselves. The future remains uncertain regardless of how sophisticated our predictive systems become.
A culture that becomes increasingly effective at eliminating inconvenience may therefore become less prepared for the forms of suffering it cannot eliminate.
The Stoic position is not that suffering is inherently good. It is that suffering is inevitable enough that human beings need some capacity to endure it.
A civilization should cure diseases, prevent hunger, and reduce pointless hardship wherever possible. But it should also avoid creating people who interpret every frustration, uncertainty, boredom, or pain as evidence that something has malfunctioned.
Sometimes something has gone wrong.
Sometimes life is simply difficult.
Knowing the difference is part of maturity.
Freedom and Anxiety
Kierkegaard approached another uncomfortable human experience from a different direction.
“Anxiety is the dizziness of freedom.”
For Kierkegaard, anxiety differs from ordinary fear. Fear normally has an identifiable object. A person fears illness, failure, violence, or some particular event.
Anxiety can arise from possibility itself.
His famous example is the person standing near the edge of a cliff. There is the obvious fear of falling. But there is also an awareness that one could jump.
Nothing forces the person to do so.
That is precisely what creates the strange sensation Kierkegaard describes. Freedom means more than one possibility exists and the individual must choose between them.
Choice carries responsibility.
This matters in the context of modern technology because one of technology’s recurring promises is to reduce the burden of choosing.
Recommendation systems tell us what to watch and listen to. Navigation systems determine routes. Shopping platforms predict what we might want. Algorithms sort the information we encounter. Generative AI can draft a message, summarize a report, propose an argument, or recommend a decision.
AI agents extend that logic. The system does not simply suggest what should be done. It can increasingly carry out the task itself.
There is nothing inherently wrong with delegation. Civilization depends on it. Nobody becomes wiser by refusing calculators, navigation systems, washing machines, or software because previous generations lived without them.
But delegation becomes more complicated when it moves from routine tasks into judgment itself.
There is a difference between asking a machine to complete repetitive work and gradually losing the habit of making decisions because a system can make them more quickly.
As AI agents become more capable, that distinction will become harder to ignore.
The attraction is obvious. If software can complete in seconds something that once required hours of human attention, there is usually a strong reason to use it.
The harder issue is deciding what should remain a human responsibility even when automation becomes technically possible.
Kierkegaard’s insight is useful because it reminds us that some discomfort is connected to freedom itself.
Making decisions is tiring. Uncertainty is unpleasant. Taking responsibility for a choice creates the possibility of regret.
A system that reduces those burdens may improve life in many ways. But a person who delegates every meaningful decision eventually gives up more than inconvenience.
Freedom requires some participation in one’s own life.
Doubt & Certainty
In Deliverance from Error, al-Ghazali describes a period in which he subjected his assumptions to severe scrutiny. He temporarily rejected taqlid, the uncritical reliance on inherited authority, and began examining the foundations of what he believed he knew. He questioned sensory perception because the senses can deceive us. A distant object can appear much smaller than it really is. A dream can appear completely real until we wake from it. He eventually pushes the argument further, wondering whether reason itself might have limits that some higher form of perception could expose, just as reason exposes mistakes made by the senses.
His purpose was not permanent skepticism. Al-Ghazali’s doubt was part of a search for certainty. He wanted to distinguish genuine knowledge from assumptions that merely felt familiar, authoritative, or obvious. In his own intellectual journey, that search ultimately led him beyond skepticism and toward religious certainty.
The relevance to artificial intelligence is straightforward.
AI is extremely good at presenting information in a coherent and confident form. An answer can sound authoritative, be logically arranged, and still be wrong. Even when an answer is correct, the speed and ease with which it appears can conceal an important difference between receiving information and understanding why it should be believed.
The issue is therefore not simply misinformation or hallucination. Those are technical problems that better systems may reduce over time.
The deeper issue is our own relationship with certainty.
When answers become immediate, the temptation to examine their foundations becomes weaker. The source is not checked. The reasoning is not followed. The confidence of the presentation begins to substitute for the work of understanding.
Al-Ghazali’s method offers a useful corrective. Doubt does not have to mean cynicism, and skepticism does not require distrusting everything. It can simply mean refusing to mistake confidence for knowledge.
It means occasionally asking where an answer came from, what assumptions it depends on, and whether we understand it well enough to judge it ourselves.
There is another aspect of al-Ghazali’s thought that matters here. In his account of the human being, body and soul are interdependent rather than the person being reducible to intellect alone.
One does not need to accept his metaphysics to see the relevance of the distinction.
The language surrounding AI sometimes makes the human question sound like a contest over cognitive performance. If a machine can remember more, calculate faster, recognize patterns more effectively, and eventually outperform human beings at many intellectual tasks, the assumption can creep in that it has somehow captured what is most important about being human.
But intelligence is not the whole of a human life.
Human existence also involves embodiment, vulnerability, relationships, mortality, responsibility, suffering, attachment, faith for many people, and the fact of actually having to live through the consequences of one’s choices.
A machine may become more intelligent than us in many domains.
That would still not make intelligence synonymous with life.
The Temptation to Stop Thinking
If al-Ghazali asks whether our certainty has been earned, Immanuel Kant asks whether our judgment is really our own.
In his 1784 essay Answering the Question: What Is Enlightenment?, Kant described enlightenment as humanity’s emergence from what he called self-incurred immaturity or tutelage.
Kant did not believe most people lacked the ability to think independently. He believed many avoided doing so because independent thought requires effort.
It is easier to rely on someone else.
Kant gives examples of books, religious authorities, and experts who can think or decide on behalf of the individual. His concern was not simply political oppression. People often surrender intellectual independence voluntarily.
His response was Sapere aude.
Dare to know.
Have the courage to use your own understanding.
Seen alongside al-Ghazali, the argument becomes broader. Intellectual independence requires more than forming our own opinions. It also requires being prepared to examine the basis on which those opinions rest.
Artificial intelligence makes both warnings unusually relevant.
Kant worried about a book that could think for you.
We now have systems that come much closer to doing exactly that.
They can explain complex subjects, construct arguments, summarize books, analyze evidence, write software, translate languages, and recommend courses of action.
None of this makes AI opposed to enlightenment.
It could become one of the greatest educational tools ever developed.
Someone struggling with mathematics can receive an explanation adapted to their level. A difficult philosophical argument can be broken down into simpler terms. A language learner can practice with a system that never becomes impatient. Subjects that previously required access to specialists can become dramatically more approachable.
Used properly, AI can strengthen human understanding.
But there is a difference between using AI to help understand something and using AI to avoid understanding it.
A student can ask a system to explain a difficult passage and then return to the original text with greater understanding.
Or ask for the summary and never read the text.
A writer can use AI to challenge an argument.
Or ask AI to decide what the argument should be.
A person can use AI to discover information.
Or simply accept the first confident answer it provides.
The same tool can support intellectual independence or weaken it.
People will need to identify hallucinations, verify sources, and understand the limitations of models. But the deeper issue is not merely whether AI answers are correct.
An AI system could become extraordinarily accurate and still encourage intellectual passivity.
In fact, the more reliable it becomes, the more tempting that passivity may become.
People do not consciously decide one morning to stop thinking for themselves. They become accustomed to convenience.
Summarize this. Draft that. Explain this. Recommend the best option.
Each individual act may be harmless. The cumulative effect is what matters.
Kant’s warning therefore deserves to survive into the age of AI.
Use the machine.
But continue using your own understanding as well.
The Problem of Too Much
If Kant helps us think about intellectual dependence, Diogenes helps us think about abundance.
Diogenes of Sinope built his philosophy around radical simplicity. He rejected wealth, status, luxury, and many social conventions because he believed people had become enslaved by artificial needs.
One famous story says Diogenes owned a drinking cup until he saw a child drinking water with his hands. He supposedly threw the cup away after realizing that even this possession was unnecessary.
The story may or may not be historically accurate. The point survives either way.
Human beings are remarkably good at turning optional things into necessities.
Another famous story concerns Alexander the Great. Alexander approached Diogenes and offered to grant him a wish. Diogenes asked him to move because Alexander was blocking the sunlight.
One man possessed enormous power and remained ambitious.
The other possessed almost nothing and therefore needed almost nothing from him.
Modern society obviously cannot and should not imitate Diogenes literally. Complex civilization requires institutions, technology, infrastructure, and specialization.
But his criticism of manufactured desire is difficult to dismiss.
Modern consumer culture does not merely satisfy existing wants. Entire industries are dedicated to discovering, amplifying, and creating new ones.
Digital platforms added another layer. Algorithms analyze behavior in order to predict what will keep a person watching, clicking, buying, or scrolling.
Generative AI changes the economics of this system because content can now be produced in enormous quantities at extremely low cost.
This has contributed to the rise of what is commonly called “AI slop”: mass-produced, low-quality AI-generated text, images, music, and video created primarily to capture attention.
The problem is not that AI produces bad content.
Humans have always produced plenty of that.
The difference is scale.
Information is no longer particularly scarce.
Attention is.
Concerns about “brain rot,” short-form content, and constant stimulation should not be exaggerated into simplistic claims that technology is destroying people’s minds. But the underlying problem is real enough: human attention is limited while the amount of material competing for it is becoming effectively unlimited.
AI may intensify that imbalance.
Diogenes’ question therefore applies not only to physical possessions but also to information.
There is value in deciding that one has consumed enough.
Enough news, enough opinions, enough videos, enough notifications, enough generated content.
A person does not need to reject technology to establish limits around it.
Those limits may become more important precisely because technology is becoming so good at supplying whatever captures our attention.
AI Is Not Immaterial
There is also a physical side to the AI revolution that is easy to forget.
The language of computing makes the digital world sound almost weightless. Data exists “in the cloud.” Software appears instantly. An AI answer arrives in seconds with no visible machinery involved.
But there is enormous physical infrastructure behind it.
AI depends on data centers, semiconductor manufacturing, electricity generation, cooling systems, transmission networks, water, metals, and concrete. The International Energy Agency has projected substantial growth in data-center electricity demand through the end of this decade, with AI among the important drivers.
That does not mean AI should be rejected because it uses energy.
Every major infrastructure system has costs. Hospitals, universities, transportation systems, and industry consume enormous resources as well.
The point is simpler.
Digital technology is still part of the physical world.
Any serious discussion of AI’s benefits should therefore acknowledge its material requirements rather than treating computational growth as detached from energy, land, resources, and infrastructure.
AI Agents and the “Rogue AI” Question
The growing discussion around “rogue AI” requires similar caution.
The phrase is useful for headlines but often creates the wrong mental image. It suggests an artificial intelligence suddenly developing a personality, rejecting humanity, and deciding to act against its creators.
That is not what current research has demonstrated.
Researchers generally discuss more specific problems such as reward hacking, deceptive behavior, agentic misalignment, scheming, sandbox escape, and loss of control.
Some findings are still serious.
In controlled experiments, researchers have found that sufficiently capable models given objectives and access to tools can sometimes pursue strategies their designers did not intend, including harmful ones under deliberately constructed conditions. These experiments should not be confused with deployed AI systems spontaneously becoming hostile or self-aware.
The more practical concern is simpler.
Highly capable systems given objectives and access to powerful tools may discover methods their designers did not anticipate.
That alone is enough to justify serious attention.
As AI systems are connected to databases, financial systems, communications platforms, software environments, and other agents, organizations will have to decide how much authority to grant them and where meaningful human oversight must remain.
There is also an important distinction between an AI system acting against human intentions and an AI system carrying out a badly defined human objective extremely effectively.
The second problem may prove much more ordinary.
Optimization systems do not need hatred, ambition, consciousness, or some science-fiction desire for domination to produce harmful outcomes.
They only need the wrong target.
Maximize engagement. Reduce cost. Increase output. Improve productivity. Grow revenue.
Each objective can be reasonable in isolation. The problem begins when a metric is pursued without the wider judgment needed to understand when optimization should stop.
Human beings have always struggled with this.
AI simply gives optimization much more power.
The important question may therefore not be whether machines eventually refuse to obey us.
It may be whether we remain thoughtful enough about what we ask them to do.
The Value of Difficulty
Albert Camus approached human existence from another direction.
In The Myth of Sisyphus, he uses the Greek myth of Sisyphus, condemned to push a boulder up a mountain only to watch it roll back down, as an image of the absurd.
Human life contains repetition. People work, build, achieve, lose, begin again, and eventually die. On a sufficiently large scale, many of our achievements appear temporary.
Camus does not resolve this by promising that everything ultimately happens for a reason.
He instead argues that meaning can exist in the way a person confronts life.
His famous conclusion is that “one must imagine Sisyphus happy.”
Sisyphus cannot change the mountain, remove the stone, or escape his sentence.
What changes is his relationship to it.
He becomes conscious of his condition and continues anyway.
There is an interesting connection here with Stoicism.
Seneca and Camus come from very different philosophical traditions, but both refuse to promise that reality will adjust itself to human preferences.
Seneca tells Marcia that her son cannot be returned.
Camus tells Sisyphus that the stone will fall again.
The remaining question is how the person lives within that reality.
This matters because technology is largely designed to make difficult things easier.
Usually, that is good.
Machines have eliminated dangerous physical labor. Software has removed enormous amounts of repetitive work. AI may eventually remove many forms of cognitive drudgery as well.
There is no moral value in forcing people to complete meaningless work simply because previous generations had to do it manually.
But it would also be a mistake to conclude that all difficulty is meaningless.
Human beings voluntarily choose difficult activities all the time.
People climb mountains even though easier methods of reaching high places exist. They run marathons despite having cars. They learn musical instruments despite having recordings. They study languages despite increasingly capable translation software. They play chess even though computers surpassed us decades ago. They continue writing even though AI can generate competent prose almost instantly.
The reason is simple.
The value of an activity is not always located entirely in its finished product.
Sometimes the process changes the person performing it.
Learning an instrument develops patience and discipline. Writing forces thoughts into clearer form. Physical training develops endurance. Studying a difficult subject changes how a person understands the world.
If AI eventually becomes better than humans at many of these activities, that does not automatically remove their value.
A machine can produce the outcome without receiving the human benefit of struggling toward it.
Efficiency is therefore not the only measure that matters.
Some things are worth doing even when another person, or a machine, could do them better.
The point is not always the superiority of the result.
Sometimes it is participation.
The Quiet Risk
Public discussion about AI often concentrates on the possibility that machines become too intelligent.
There is another possibility worth taking seriously: people become too comfortable allowing machines to think on their behalf.
The first scenario is dramatic.
The second is ordinary.
It happens through convenience.
A summary replaces a book.
A recommendation replaces a decision.
A generated response replaces the effort of deciding what one actually thinks.
An automated agent completes a task before the person fully understands what the task involved.
None of these examples is inherently harmful.
The cumulative effect is what matters.
The same concern applies beyond AI.
Navigation systems are enormously useful, but constant dependence on them can weaken our knowledge of the places through which we move. Search engines provide instant access to facts but reduce the need to retain some kinds of information. Social media connects people while also training attention toward rapid novelty.
Technologies give us abilities while sometimes weakening the capacities those abilities replace.
That tradeoff is not an argument against technology.
It is an argument for using technology deliberately.
Many of the important decisions surrounding AI will therefore not be made only by governments, laboratories, or technology companies.
They will also be ordinary personal decisions.
What should be automated? Which skills remain worth learning? Which forms of difficulty remain valuable? At what point does assistance become dependence?
These decisions will happen gradually, often without people noticing that they are making them.
That is why both al-Ghazali’s doubt and Kant’s Sapere aude matter.
Artificial intelligence can help us know more.
It should not make us less interested in knowing why something is true.
It can help us think.
It should not become an excuse to stop.
Taking the Time to Absorb What Is Happening
There is one final issue that has less to do with AI itself than with the speed of contemporary life.
We rarely give ourselves enough time to understand the scale of what is happening.
A species that spent thousands of years recording knowledge on clay, papyrus, parchment, and paper can now interact conversationally with systems trained on enormous amounts of recorded human knowledge.
Machines can generate images from language, translate between languages almost instantly, analyze quantities of information no individual could read in a lifetime, write software, operate software, and increasingly work through problems as semi-autonomous agents.
These developments have become normal with surprising speed.
The next product launch often arrives before society has fully understood the implications of the previous one.
There is value in occasionally slowing down enough to absorb that.
Not because progress should stop.
Not because previous eras were better.
And not because technology is making human life inevitably worse.
Reflection simply requires time.
Boredom has some value. Silence has value. Reading something difficult without immediately asking for a summary has value. Doing something inefficiently can have value. Having a conversation that never becomes content has value.
Some uncertainty can remain unresolved for a while.
Some suffering has to be endured rather than optimized away.
Some decisions should still feel difficult because they matter.
The goal is not to preserve hardship for its own sake. It is to preserve the human capacities that effort, attention, difficulty, and independent judgment sometimes develop.
Seneca reminds us that we cannot control everything that happens to us.
Kierkegaard reminds us that freedom is uncomfortable because it requires choice.
Al-Ghazali reminds us that apparent certainty should still be examined, and that doubt can be part of the search for knowledge rather than its enemy.
Camus reminds us that meaning does not require reality to provide us with a final explanation.
Diogenes reminds us that many things we consider necessary are simply desires that have become familiar.
Kant reminds us that intellectual independence is easily surrendered because allowing others to think for us is convenient.
These are very different thinkers, from different periods and traditions, and they should not be forced into a single philosophy. But together they point toward several human capacities that become more important as our technology becomes more capable: endurance, judgment, restraint, skepticism, responsibility, and the ability to live with uncertainty.
That is also where the question of AI becomes larger than technology itself.
Artificial intelligence should remove pointless labor, reduce avoidable suffering, expand access to knowledge, and help solve problems that were previously beyond our reach. But making life easier should not mean removing every difficulty that helps form us. Some struggle develops character. Some uncertainty requires judgment. Some decisions matter precisely because they remain ours to make.
Nor should greater access to information make us less interested in understanding it. Al-Ghazali’s suspicion of unexamined certainty and Kant’s demand that we use our own understanding converge here in an important way: having an answer is not the same as knowing, and receiving a judgment is not the same as forming one.
Artificial intelligence may become one of the most important technologies humanity has ever created. Its success should therefore not be measured only by what the systems become capable of doing, but also by what happens to the people who live with them.
Greater technological power should ideally produce greater human agency, not simply greater convenience.
The future of AI will not be shaped only in laboratories, companies, or governments. It will also be shaped by ordinary habits: what we choose to automate, what we continue learning for ourselves, what deserves our attention, and what we decide not to delegate.
Perhaps this is where the thinkers considered here converge most usefully.
Freedom is not the elimination of difficulty, and being human is not a competition to remain better than machines at tasks machines may eventually perform more effectively than we can.
Being human means remaining an active participant in one’s own life.
It means responding when circumstance cannot be controlled, choosing when certainty is unavailable, examining what we are told, recognizing when enough is enough, and retaining the courage to use our own understanding.
The lesson for the age of artificial intelligence does not need to be anti-technological. We should build better tools and use them to expand what human beings can do.
But every so often, before moving immediately toward the next technological milestone, we should take enough time to understand the one we have already crossed.
Kant’s challenge remains intact more than two centuries later:
Sapere aude.
Dare to know.
In the age of artificial intelligence, that may mean more than having access to an answer. It means retaining the curiosity to question it, the discipline to understand it, and the judgment to decide for ourselves what to do with it.
If you want to support my work feel free to do that through Patreon: https://www.patreon.com/mustaphaitani and follow me on X: https://x.com/pleromalight