Based on a conversation with Hrvoje Morić.
We have access to more information than at any other point in human history. Search engines can retrieve millions of pages in seconds. Social media gives almost anyone the ability to publish. Artificial intelligence can digest enormous amounts of information and give us an answer almost instantly.
It may be getting harder, not easier, to know what is true.
That was the question at the center of my conversation with Hrvoje Morić, host of Geopolitics & Empire. Hrvoje has spent years interviewing people across the political spectrum about geopolitics, war, censorship, technology, and power. Our conversation ranged from censorship and deplatforming to propaganda, algorithms, artificial intelligence, climate change, war, and the growing role of digital systems in determining what information people encounter.
We didn't agree on or establish every claim that came up in the conversation. Some of the most interesting questions were precisely the ones where the evidence was incomplete, disputed, or open to multiple interpretations.
That raises a question:
What happens when the systems we depend on to determine what is true are themselves shaping what we are allowed to see?
And: How do we think clearly, when we don't completely trust the information environment around us?
Information Is Not the Same as Knowledge
For most of human history, finding information was not as easy as it is today. If you wanted to understand a complicated subject, you might go to the library, find several books, read competing arguments, examine their sources, and form your own conclusion.
The process may have seemed inefficient. That inefficiency had an advantage:
You had to think.
Imagine walking into a library and finding two books next to one another. One argues strongly for a particular position; the other argues the opposite. You can pull both from the shelf, read them, compare their evidence, and decide whether either argument survives scrutiny.
Modern information systems remove the intermediate step. A search engine ranks the information before you see it. A social-media algorithm decides what appears in your feed. An AI system goes one step further by synthesizing everything into a single answer.
It's is convenient.
Convenience creates a new problem.
Who or what decides which information matteres?
The Algorithm Making Decisions
We may think of censorship as someone banning a book or deleting an article. Information doesn't have to disappear to become invisible. An algorithm can rank something lower. A recommendation system can stop recommending it. A search engine can place it on page ten, instead of page one. A social platform can decide that another piece of content deserves greater distribution.
Nothing needs to be banned.
The information exists.
People simply stop encountering it.
That changes the question. It isn't," Who controls information," it's “Who controls visibility?” Every ranking system necessarily makes choices. Something appears first. Something appears last. Something is recommended. Something isn't.
That doesn't mean every ranking decision is malicious or politically motivated. Ranking is necessary when the amount of available information exceeds our ability to consume it.
Ranking creates power nonetheless. If an algorithm influences which voices receive amplification, then the criteria behind that algorithm can influence what millions of people perceive as important, credible, popular, or fringe.
The Difference Between a Fact and an Interpretation
This becomes important when discussing censorship, propaganda, climate, war, political power, or a contested subject. Several things can become mixed together:
A documented event.
An interpretation of that event.
A hypothesis about why it happened.
A conclusion about who caused it.
Those are not the same.
During our conversation, Hrvoje described Patreon freezing his account and requiring him to remove two interviews from multiple platforms before it would be restored. He declined and closed the account.
It doesn't tell us why it happened. Was Patreon enforcing its own policies? (According to Hrvoje, no.) Were those policies applied consistently?Was the company responding to pressure from somewhere else?Was Hrvoje targeted?
Those are different questions, requiring different evidence. The easiest way misinformation spreads is by moving silently from:
"This happened."
to
"I know why it happened."
Sometimes an explanation prove correct. The evidence required for the second statement is greater than the first. That standard has to apply regardless of whether the claim comes from a government, corporation, journalist, scientist, or independent researcher.
Censorship Creates Another Problem
There is a practical problem with removing questionable information. It doesn't make people trust reliable information more. Sometimes it does the opposite.
Suppose someone encounters a controversial argument.One response is to provide the evidence supporting it, the evidence contradicting it, and the uncertainties that remain.
Another response is:
You aren't allowed to hear that argument.
The second approach leaves an obvious question:
Why not?
Removing an argument doesn't demonstrate that the argument is wrong. It removes the opportunity to examine it. Once people discover that legitimate information has sometimes been suppressed, mislabeled, deprioritized, or later revised, they may begin distrusting everything.
That's dangerous. Healthy skepticism can become indiscriminate skepticism. Instead of believing every institutional claim, someone begins rejecting every institutional claim. Those are mirror images of the same problem: neither requires evaluating evidence.
AI Changes the Problem
This part of the problem became evident after an experience I had while using AI to edit an article. I wasn't asking the AI to evaluate my argument. I was asking it to help with editing. Yet it began inserting qualifications and counterarguments into what I had written. At one point I noticed the system performing what appeared to be an accuracy check.That made me wonder:
Accuracy according to whom?
AI systems have to make choices. They have training data. They have rules. They have ranking mechanisms. They have methods for determining which sources deserve more weight. Those decisions may be reasonable. They are still decisions. The danger isn't that AI gives us false information. The danger is that we stop asking how it arrived at the answer.
The One-Answer Problem
Consider the difference between these two systems.
A search engine says: Here are ten sources.
An AI assistant says: Here is the answer.
The second is easier. And it changes how we think. When we see ten competing sources, disagreement is visible. When an AI combines those sources into one fluent paragraph, much of any disagreement disappears. Uncertainty disappears. The answer may sound authoritative, even when the underlying evidence is messy. That creates what may become one of the central information problems of the AI era:
The better AI becomes at answering questions, the less incentive these is to investigate the answers. The tool designed to make us smarter could unintentionally make us more intellectually dependent.
What Happens When the Story Changes?
War provides a useful example as alliances and narratives change over time. A government, organization, or armed group portrayed as an ally during one period may become an enemy during another, or the reverse. Consider Orwell's 1984, where the enemy changes while society is expected to accept the current version of events as though it had always been so.
Changing a position isn't itself evidence of deception. Circumstances change. New evidence appears. Alliances shift. Policies change. The question is whether we preserve a historical memory to recognize that the position changed. If yesterday's information quietly disappears beneath today's narrative, we lose the ability to ask:
What changed?
Why did it change?
What evidence justified the previous position?
What evidence justifies the new one?
Those questions matter, regardless of which political party, government, corporation, or media organization is involved.
Predictions Can Mislead Us
One way we judge whether someone understands a system is by looking at their predictions. If they repeatedly predict what will happen before it happens, it seems reasonable to conclude that they understand what is going on. That deserves scrutiny. A prediction isn't independent of the event being predicted.
Consider financial markets. If enough influential people predict a stock market or real-estate crash, people may begin acting on that prediction: Investors sell. Buyers hesitate. Banks tighten lending. Confidence falls. Eventually, the predicted decline may actually occur.
Did the prediction reveal what was going to happen? Or did it help make it happen?
There is another possibility to consider when power is involved. Someone with enough influence may be able to predict an outcome because they, or institutions aligned with them, can influence the conditions that produce it. In that case, saying “They predicted it, therefore they understood what was really happening” would be a mistake.
Prediction can be useful evidence. It isn't proof. We need to ask:
Was the prediction specific and made before the event?
Was the predicted outcome likely?
Could publicizing the prediction have influenced what happened?
Did the person or institution making the prediction have the ability to influence the outcome?
Does the evidence support the explanation they gave for why it happened?
Could another explanation account for the same result?
That leads to the larger problem with information and power. When the people describing reality have the ability to influence reality, we have to evaluate whether their predictions came true and how those outcomes came about
The Same Standard Has to Apply to Everyone
It's easy to demand extraordinary evidence from people we disagree with while accepting weak evidence from people we already trust.
If a government makes a claim, ask for evidence.
If an independent journalist challenges a claim, ask for evidence.
If a scientist makes a claim, ask for evidence.
If someone challenges the scientist, ask for evidence.
If an AI gives an answer, ask where it came from.
If I make a claim, question that too.
Skepticism is refusing to outsource the process of deciding what is true.
What Can We Do?
Suppose information is being manipulated in some cases. Algorithms influence what we see. Suppose governments, corporations, media organizations, advocacy groups, wealthy individuals, political movements, and technology companies attempt, in different ways and for different reasons, to influence public opinion.
We don't need to establish the existence of one centralized system controlling everything before asking what we can do about the vulnerabilities that exist.
Don't depend on one source.
Don't depend on one platform.
Don't depend on one algorithm.
Don't depend on one expert.
Don't depend on one AI answer.
Look for original data when possible. Separate observation from interpretation. Find the strongest argument against your conclusion, and ask what evidence would cause you to change your mind.
Keep track of predictions. Ask whether competing explanations could account for the same.
Preserve your willingness to say: I don't know yet.
The Real Threat May Be Intellectual Dependency
The information problem is often framed as misinformation versus truth. That's too simple.The deeper problem is dependency. If someone else determines what information we encounter, which sources are credible, which questions are acceptable, and what conclusion follows, then we haven't been misinformed - we've stopped thinking.
Artificial intelligence makes this important. AI doesn't merely distribute information, it interprets it for us. That can be enormously useful. I use AI myself. The right relationship with AI isn't:
Tell me what's true.
It is:
Help me investigate what's true.
Show me the evidence.
Show me competing explanations.
Tell me where the evidence is strong.
Tell me where it is weak.
Tell me what assumptions we're making.
Tell me what would falsify the conclusion.
Now let me think.
In a world where information can be filtered, ranked, censored, amplified, generated, and summarized before it reaches us, the most important skill may no longer be finding information.
It may be maintaining the ability to question how we know what we think we know.
Editor’s Note: This article is based on my podcast interview with Hrvoje Morić, recorded and published in October, 2025. The ideas discussed here originate from that conversation. The structure, emphasis, and commentary are my own. Any errors or interpretations should be attributed to me, not to Hrvoje Morić.
About the Author / Host
Daniel Stih is an aerospace engineer, software engineer, indoor environmental consultant, and author of 12 books. Through engineering, investigations, and interdisciplinary research, he explores how people define problems, interpret evidence, and make conclusions when there's uncertainty. Learn more in Why I Think This Way.
Related Articles
When Algorithms Reinforce the First Question
Related Podcast Episodes
ChatGPT Accuracy Guidelines - Accuracy or Misinformation?
Facebook Whistleblower and Big Tech Censorship - Ryan Hartwig