Trust in the Age of Mediated Reality

We often speak about trust as if it were a single thing: trust in a website, a platform, an institution, an expert, or a piece of information.

Today, a systemic crisis of trust is emerging - and intensifying - on several levels at once. We are losing confidence not only in what institutions tell us, but also in the infrastructure that delivers information and in the procedures by which claims are judged to be true.

These layers are connected, but they are not identical. Each raises a different question:

1. Infrastructural trust: Am I actually seeing the source I intended to access?
2. Institutional and social trust: Who organizes, filters, interprets, and validates the information I receive?
3. Epistemic trust: Why should I consider this claim justified?

The central problem is not simply that people trust too little or too much. It is that the conditions under which trust becomes reasonable are becoming harder to see.

1. Infrastructural Trust: Can I Reach the Intended Source?

Infrastructural trust concerns the technical systems that stand between a user and a source of information.

These systems include:

• DNS and routing;
• internet service providers and resolvers;
• deep packet inspection and filtering;
• certificates and certificate authorities;
• hosting providers and content-delivery networks;
• platform-level restrictions on content availability.

At this level, the basic question is deceptively simple:

Am I seeing the actual source I tried to reach, or an altered, redirected, filtered, or substituted version of it?

Digital systems are often presented as neutral channels. We type an address, click a link, or open an application, and assume that the result corresponds to our intention. But that assumption depends on a long chain of technical and institutional arrangements.

A domain name must be resolved. Traffic must be routed. Connections must be authenticated. Certificates must be accepted. Content must remain available. Each stage introduces a potential point of intervention.

Some interventions are obvious: a website is blocked, a domain is redirected, or a service is unavailable. Others are more difficult to detect. A connection may appear normal while the route, response, or available content has been altered.

Infrastructure therefore creates a foundational layer of trust. Before asking whether a claim is true, we need to know whether we have reached the source that made the claim.

If the answer is uncertain, every subsequent layer becomes unstable.

2. Institutional and Social Trust: Who Organizes Visibility?

Even when the connection is authentic, another problem remains.

A user may genuinely receive information from the platform, search engine, newsroom, university, government agency, or social network they intended to use. Yet the information they see may still be the result of selection, ranking, filtering, moderation, personalization, or interpretation.

This is not only a problem of institutional trust. It is also a problem of social trust: trust in other people, communities, professional groups, and networks of collective judgment.

Institutional and social trust concern the actors who organize, produce, circulate, and validate information:

• search engines and digital platforms;
• news organizations and editorial institutions;
• universities and expert bodies;
• government agencies and professional associations;
• journalists, researchers, and public intellectuals;
• online communities and social networks;
• colleagues, peers, friends, and members of one’s own social groups.

The question at this level is not simply whether the content has been technically substituted. It is whether the people and organizations shaping the information environment are acting according to interests, assumptions, or group loyalties that remain hidden.

Am I seeing what is most relevant - or what is most advantageous to the intermediary and its community?

Social trust adds another dimension. Many of our beliefs do not come directly from institutions. They come through people we know, groups we identify with, and communities whose judgments we rely on.

We often trust a claim because it is endorsed by someone perceived as competent, honest, or familiar. We may accept information because it circulates within a community that shares our values. Conversely, we may reject a claim not because its evidence is weak, but because it comes from an out-group.

This creates a double vulnerability. Institutions can shape what becomes visible, while communities can shape what becomes socially acceptable to believe.

A platform may rank content according to opaque rules. A professional community may exclude inconvenient evidence. An online group may reward conformity and punish dissent. A trusted friend may pass along a claim that they have not independently examined.

In each case, trust operates through relationships - not only through formal organizations.

Mediation is therefore both institutional and social. Institutions select and frame information, but people and communities interpret, endorse, contest, and redistribute it.

This does not make trust irrational or avoidable. No one can personally verify every fact. Social cooperation depends on relying on others. The problem arises when social proximity is confused with competence, group loyalty with truth, or institutional status with impartiality.

When users do not know how visibility is produced - or how beliefs circulate within their communities - they may confuse prominence with importance, familiarity with reliability, popularity with credibility, and solidarity with evidence.

At this level, the question becomes:

Who is shaping what I see, whom do they represent, and why should I trust their judgment?

3. Epistemic Trust: Why Should I Believe the Claim?

The deepest layer concerns the grounds on which we accept a statement as justified.

Epistemic trust is trust in sources, arguments, evidence, and procedures for reaching conclusions.

It asks:

• Why should this claim count as established?
• How can we distinguish an argument from a rhetorical technique?
• What matters more: expertise, data, personal experience, or majority agreement?
• How should uncertainty be represented?
• How should we respond to conflicting evidence?
• What would count as a reason to revise our belief?

These questions cannot be answered by technical authentication alone.

A document may be authentic and still be false. A respected expert may be mistaken. A study may be methodologically sound but limited in scope. A personal experience may be genuine but insufficient to support a general conclusion. A majority may be wrong, while an unpopular minority position may later prove correct.

The challenge is not merely to find trustworthy people or institutions. It is to understand the procedures by which trust is earned, maintained, challenged, and revised.

This is why the crisis of trust is also an epistemological crisis. We are increasingly unsure not only whom to believe, but what believing responsibly should involve.

The central question is:

What exactly gives us the right to consider a claim worthy of trust?

That is no longer merely a technical question. It is an epistemological one.

Yet technology plays an active role even here. It determines which evidence is accessible, which sources are visible, which arguments are amplified, and which forms of reasoning travel most effectively through the information environment.

Substitution and Mediation Are Not the Same

It is useful to distinguish two forms of threat that are often collapsed into one.

Substitution occurs when a user believes they are interacting with one entity while actually interacting with another.

For example, traffic may be redirected, a connection may be intercepted, or a third party may appear to speak on behalf of a website. The fundamental problem is one of identity.

The user asks:

Is this really the source I intended to reach?

Mediation, by contrast, occurs when the user is genuinely interacting with the declared service, but that service selects, filters, ranks, or interprets what the user sees.

The fundamental problem is not identity but selection.

The user asks:

Why was I shown this rather than something else?

And then:

What interests or assumptions shaped that choice?

These two threats require different responses. Authentication and cryptography can help with substitution. Transparency, accountability, pluralism, and institutional checks are more relevant to mediation.

Confusing the two produces false confidence. A secure connection may establish that we are communicating with the correct server. It does not establish that the server has shown us the most relevant material, or that its interpretation is justified.

This can be stated simply:

Cryptography can help answer the question, “Is this really the intended node?” But it cannot answer, “Why did I see this?” or “Why should I believe it?”

The Limits of Technical Solutions

Technology is often expected to solve the crisis of trust by replacing human judgment with verification.

Digital signatures can verify authorship or integrity. Certificates can help authenticate domains. Transparency logs can make certain changes visible. Reproducible processes can reduce dependence on opaque claims. Open data can make analysis more inspectable.

All of this matters.

But verification is not the same as justification.

A cryptographic signature can show that a document was signed by a particular key. It cannot show that the signer was correct. A provenance record can show where data came from. It cannot determine whether the data supports the conclusion drawn from it. An algorithm can be transparent and still encode questionable assumptions.

Technical tools can reduce some forms of uncertainty. They cannot eliminate the need for interpretation.

In fact, they may move trust to a different location. If we rely on a verification system, we must still ask:

• Who designed it?
• What does it verify?
• What does it leave out?
• Who maintains it?
• What happens when it fails?
• Which assumptions are built into its categories and standards?

Trust does not disappear. It becomes distributed across protocols, institutions, maintainers, auditors, and users.

Can Technology and AI Solve the Crisis of Trust?

Can technology and AI solve these problems sufficiently to address the current crisis of trust?

Yes - but only if they operate simultaneously at both extremes of the problem: at the deepest epistemological level and at the most basic infrastructural level.

It is not enough to build more secure networks if users still cannot understand why a claim is persuasive. Nor is it enough to provide better argument analysis if users cannot be confident that they are interacting with the intended source.

The response must therefore work in both directions: downward, toward the infrastructure that delivers information, and upward, toward the reasoning by which information is assessed.

1. Make the Structure of Arguments Visible

AI systems should help users see the argumentative structure of a text rather than merely summarize its content.

A text can be analyzed and represented through formal argumentation frameworks such as ArgDown or the Argument Interchange Format, or AIF. These representations can make visible:

• the main claims;
• the evidence supporting them;
• the assumptions connecting evidence to conclusions;
• objections and counterarguments;
• contradictions between claims;
• distinctions between facts, interpretations, and predictions;
• degrees of uncertainty;
• unsupported rhetorical moves.

This analysis could be generated locally by small AI models capable of running on relatively modest hardware. Local processing would allow users to inspect texts without sending every document or personal reading history to a centralized service.

At the same time, articles could include “baked-in” argument maps prepared by their authors or editors. In that model, the structure of the reasoning would not be added only after publication by an external AI. It would become part of the publication itself.

Authors could publish not only a prose text, but also a machine-readable representation of its argument:

• what is being claimed;
• what evidence is offered;
• which sources are relied upon;
• what remains uncertain;
• which objections are acknowledged;
• what conclusions do not follow from the available evidence.

This would not make arguments automatically correct. An argument map can be incomplete, biased, or badly constructed. But it would make the reasoning more inspectable.

It would shift part of the burden from trusting the author’s authority to examining the relations between claims, evidence, and conclusions.

The role of AI here should not be to replace judgment. It should be to expose structure.

2. Strengthen the Infrastructure of Source Access

The infrastructural layer requires a different kind of intervention.

One possible starting point is support for embedded or “baked-in” DNS-related data in links. Such links could carry signed metadata about the intended domain, relevant DNS records, or the expected destination. A client could use this information to validate or establish the connection without relying entirely on a conventional resolver at the moment the user follows the link.

The aim would be to reduce - or, where possible, eliminate - the need to ask an external resolver to determine where a link should lead.

This would not remove the need for DNS altogether. Domains change, records expire, infrastructure is relocated, and services may deliberately use dynamic routing. Embedded data would therefore need expiration dates, cryptographic signatures, versioning, and mechanisms for safe update or revocation.

A further layer of protection could come from human-verifiable, non-machine-readable values used to derive a temporary session key during the initial connection. If correctly implemented, this could reduce the server’s dependence on the client’s trust in the correct certificate authority. The authentic server could participate directly in confirming the expected certificate or establishing an additional secure channel, rather than leaving the user to choose between no access and access through a potentially substituted certificate.

But the principle is important: a link should be able to carry more information about the identity and expected destination of the source it refers to.

Instead of treating resolution as an invisible step delegated entirely to the network, the system could make more of the source’s identity verifiable at the point of access.

This would help address substitution. It would not, by itself, solve mediation. A user might still reach the correct platform and receive a carefully selected or politically framed version of reality.

That is why the infrastructural solution must be combined with tools that expose argumentation, provenance, selection, and uncertainty.

Two Ends of the Same Problem

These two proposals may appear unrelated.

Argument mapping concerns the interpretation of claims. Embedded DNS metadata concerns the delivery of information. One operates at the epistemological level; the other operates at the infrastructural level.

But they address opposite ends of the same trust problem.

The first asks:

Why should I believe this claim?

The second asks:

Did this information come from the source I intended to reach?

Between them lies the institutional and social layer:

Who selected, framed, transmitted, and endorsed this information?

AI can also contribute here by making ranking, filtering, provenance, and editorial decisions more visible. It can compare sources, identify omitted alternatives, and reveal where a platform’s presentation differs from the underlying body of evidence.

Yet these functions must remain inspectable themselves. A system that explains every institution through an opaque AI model simply moves the problem one level higher.

The goal is not to create a new authority that users must trust blindly. It is to create tools that help users inspect the authorities, procedures, and arguments already shaping their information environment.

From Trust to Inspectability

Technology will not make trust unnecessary. But it can change what trust requires.

A user should not have to trust the entire network in order to establish the identity of one source. They should not have to trust an entire institution in order to inspect the reasoning behind one claim. They should not have to accept a platform’s ranking as a neutral representation of relevance.

The most promising direction is therefore not a world without trust, but a world with more inspectable trust.

Such a system would combine:

• verifiable source identity;
• stronger provenance and delivery guarantees;
• visible argument structures;
• local AI-assisted analysis;
• author-provided machine-readable argument maps;
• explicit uncertainty and counterarguments;
• greater transparency about institutional selection;
• mechanisms for correcting and revising published claims.

The trust crisis cannot be solved by infrastructure alone, because secure delivery does not establish truth. It cannot be solved by epistemology alone, because sound reasoning is useless if the source has been substituted or the evidence is inaccessible.

The answer must reach in both directions at once.

Can technology and AI solve the crisis of trust?

Yes - if they are designed not to demand more blind confidence, but to reduce the amount of confidence that must be given blindly.

That means beginning at both ends: making the path to the source more verifiable and making the structure of the claim more visible.

At the infrastructural level, we need stronger guarantees that information comes from the intended source. At the epistemological level, we need tools that show why a claim is being made and how well it is supported. At the institutional and social level, we need greater visibility into who selects, frames, endorses, and circulates information.

The future of trust will depend less on finding perfectly trustworthy intermediaries than on building systems in which users can inspect identity, selection, evidence, and reasoning for themselves.

The goal is not to eliminate trust.

It is to make trust more local, more explicit, more revisable - and less dependent on invisible intermediaries.