Every Expansion of Knowledge Creates New Work

AI, Collective Sensemaking, and the Next Challenge for Civic Infrastructure

For most of human history, access to information was constrained by distance, cost, literacy, and institutional control. Knowledge moved slowly. It was stored in particular places, maintained by particular people, and distributed through systems that determined who could participate.

We have spent centuries dismantling those constraints. Printing, public education, libraries, broadcasting, the internet, and mobile communication have each made it easier for more people to find information and contribute their own knowledge. That expansion has brought enormous benefits. It has opened pathways to education, enabled new forms of economic participation, and it has allowed people to organize, advocate, and share knowledge in ways that were previously out of reach. But it has also created a growing problem we are only beginning to understand, and one we do not yet know how to manage very well.

The amount of information available to us has grown much faster than our ability to interpret it together.

Generative artificial intelligence is widening that gap vastly. It can produce articles, images, video, analysis, commentary, summaries, and persuasive arguments at a scale that would have been impossible to imagine even a few years ago. Much of that output may be useful. Some will be misleading or deliberately deceptive. An enormous amount will simply be inexpensive, unnecessary filler.

The challenge is no longer gaining access to enough information. It is developing the human and civic capacity to decide what deserves attention, what can be trusted, what context is missing, and what any of it means for the work we need to do together.

That is a collective sensemaking problem.

Information Has Never Interpreted Itself

Information is often treated as though its value is self-evident. Give people better facts, the assumption goes, and they will make better decisions. Give communities more data and they will understand their problems more clearly. Make knowledge widely available and public debate will become better informed.

But our decades of work building and nurturing advocacy networks suggest a more complicated picture.

People encounter information from different positions, experiences, relationships, and levels of trust. They do not simply absorb a fact and update their understanding accordingly. They consider who provided it, whether it fits what they have already seen, how people around them are responding, and whether accepting it would threaten an identity or relationship they value.

This is why sensemaking is social. People develop an understanding of complicated situations through conversation, disagreement, comparison, and repeated exposure to the experiences of others. In strong networks, those processes help participants recognize patterns that no single person could see alone. They also make it possible to revise an interpretation when conditions change.

A public health advocate, local organizer, researcher, attorney, funder, and affected resident may all hold different pieces of the same problem. Their value does not come only from the information each person carries. It comes from the network’s ability to connect those perspectives, establish enough trust for honest exchange, and create a shared picture that supports action.

Without that capacity, an increase in information has diminishing marginal returns, and can ultimately produce more confusion than clarity.

Every Expansion of Knowledge Creates New Work

Previous information revolutions did not only give people more material to read or watch. They created new requirements for organizing knowledge.

Libraries helped preserve and classify written material. Universities developed communities of inquiry. Journalism established practices for gathering, checking, and presenting information about public events. Scientific institutions built systems of peer review, replication, and professional scrutiny. Schools prepared people to navigate a society increasingly organized around written knowledge.

None of these institutions has ever been neutral or perfect. Each reflects power, culture, and the limitations of the people who built it. Still, they perform essential work. They help societies filter information, test claims, maintain records, develop shared language, and carry knowledge across generations.

The internet weakened many of the constraints those institutions once placed on publication. That opened public discourse to people who had been ignored or excluded, and it allowed communities to exchange knowledge without waiting for institutional permission. It also removed much of the friction that had limited how much material could circulate.

Social media reduced that friction further. It organized information around engagement, speed, and immediate reaction. People gained extraordinary publishing power, but the systems governing distribution were rarely designed to support careful interpretation. They rewarded content that moved quickly through a network, whether or not it helped the network understand anything.

Generative AI changes the production side of this equation. A person no longer needs substantial time, skill, or knowledge to create material that looks complete and authoritative. The cost of producing content is approaching zero while the cost of evaluating it continues to rise.

AI Is Lowering the Bar for Producing Plausible Content

The most visible examples are easy to dismiss. Search results filled with repetitive articles. Social feeds crowded with synthetic images. Automated comments that add volume without contributing meaning. Videos assembled from generated scripts, voices, and visual material. Reports that sound informed until someone checks the sources.

“AI slop” has become a convenient label for this material, but the phrase can make the problem sound mainly aesthetic, as though we are dealing with an internet full of cheap clutter.

The deeper problem is structural.

Every new piece of synthetic content competes for limited attention. Every fabricated citation or misleading image imposes verification work on someone else. Every automated account that enters a public conversation makes it harder to judge whether a response reflects a person, a coordinated campaign, or a machine producing likely language.

Much of this content will not be malicious. That may make the challenge harder. A flood of deliberate propaganda can sometimes be identified as a threat. A flood of mediocre, partially correct, highly confident material is more difficult to confront because it blends into ordinary communication.

The influx means that people will not have time to investigate each article, image, or clip they encounter. They will rely on familiar shortcuts. Does this confirm something I already believe? Was it shared by someone I know? Does it provoke an immediate emotional response? Have I seen the claim repeated often enough that it feels established?

These shortcuts are not evidence of individual failure. They are adaptations to cognitive overload. Human attention has limits, and no amount of media literacy can give a person enough time to independently verify an endless stream of claims.

The Weak Link Is Moving

For years, many civic and advocacy efforts operated from a scarcity model. The problem was that people lacked access to information. Organizers gathered facts, translated technical material, exposed hidden harms, and distributed knowledge that powerful institutions preferred to keep contained.

That work remains necessary. Many communities still struggle to obtain reliable information about decisions affecting their health, rights, environment, or economic future.

But in other settings, the weak link has shifted. Information may be plentiful while interpretation, trust, and coordination remain underdeveloped.

At Netcentric Campaigns, we have used the term Civic Pollution to describe the deterioration of the relationships, information systems, institutional practices, and shared civic spaces people need to work through public problems together. Information overload is one contributor to that pollution, but the damage is not measured simply by how much bad information circulates. It appears in the weakening of trust, the fragmentation of attention, the loss of shared context, and the growing difficulty of maintaining constructive relationships across disagreement.

Generative AI enters an information environment where those civic systems are already strained. By increasing the volume, velocity, and apparent credibility of content, it can intensify conditions that make collaborative sensemaking harder. The concern is not only that people may believe something false. It is that communities may lose the common ground, relationships, and interpretive capacity required to determine what is true or consequential together.

This distinction matters for strategy. A network can invest heavily in reports, messaging, databases, and digital communication without increasing its capacity to make sense of changing conditions. Participants may receive more updates while understanding less about how their work connects. They may have access to the same documents but lack a common language for discussing them. They may agree on broad goals while interpreting the environment in incompatible ways.

Adding more content to that system will not repair it.

The work may instead involve strengthening relationships among participants, improving communication across organizational boundaries, creating regular spaces for reflection, and building feedback mechanisms that allow people to compare what they are seeing. These practices often look slower than information distribution. Over time, they allow a network to respond more intelligently because people have developed the trust and shared context needed to interpret new information together.

Collective Sensemaking Is Civic Infrastructure

We often talk about infrastructure as something physical or technical: roads, electrical systems, communications networks, water systems, servers. These systems make other forms of activity possible.

Collective sensemaking serves a similar function.

Communities need places where people can encounter different perspectives without immediately retreating into opposing camps. Movements need trusted relationships through which emerging information can be evaluated. Networks need people who can notice patterns, connect isolated observations, and surface disagreements before they harden into fractures. Democratic societies need institutions capable of creating enough shared understanding for people to make decisions together.

Local journalism can serve this function when it connects public events to the lived reality of a community. Libraries can help people navigate information without demanding allegiance to a political or commercial agenda. Professional associations can maintain standards and interpret new developments within a field. Advocacy networks can connect technical expertise with local knowledge and direct experience.

The value of these institutions rests partly in what they know. It also depends on the relationships, norms, and repeated interactions that make their knowledge usable.

That distinction becomes more important as the volume of available information continues to grow. When knowledge is abundant, the limiting factor is no longer access but the ability to interpret and apply it in context.

More concerning, AI is now being used on both sides of this process. It is generating large volumes of content while also being used to sort, summarize, and interpret that same material. The same systems that expand the information environment are increasingly positioned as the tools for navigating it.

That creates a layered problem. When AI produces content at scale, it adds to the volume that people must evaluate. When AI is then used to interpret that volume, it can create the impression that the work of sensemaking has already been done.

A well-written summary may conceal uncertainty. A confident answer may narrow the questions a group considers. A rapid synthesis may remove the productive friction that comes from people comparing interpretations and discovering where they differ. When participants begin with the machine’s framing, they may spend less time deciding whether it is the right frame.

This dynamic compounds itself. The more content is generated, the more pressure there is to rely on automated interpretation. The more that interpretation is accepted, the less visible the underlying uncertainty becomes. What looks like clarity may be the result of compression rather than understanding.

The quality of the tool does not remove the need for human judgment. In many cases, it increases the risk that judgment is bypassed because the output is easier to accept without scrutiny.

Testing the Models Is Only Half the Work

The United States has already begun treating frontier AI models as technologies whose deployment warrants government scrutiny before they are broadly released. Recent generations of models have undergone government evaluation before public rollout, and in some cases that scrutiny has influenced deployment timelines or resulted in additional safeguards before wider release.

That represents an important shift. It reflects a growing recognition that sufficiently capable AI systems can have consequences extending well beyond the companies that build them. Questions about cybersecurity, biological misuse, persuasion, autonomous behavior, and other systemic risks are no longer viewed as issues to address only after a model has entered the world. Increasingly, they are becoming part of the conversation before deployment.

That is a sensible direction. But it also exposes a blind spot.

If we believe advanced AI deserves careful evaluation because of what it might do to society, shouldn’t we be equally concerned with whether society is prepared for what AI might do?

A frontier model does not enter an empty environment. It enters financial systems, schools, local governments, newsrooms, community organizations, advocacy networks, families, and millions of daily conversations. Its impact will depend partly on the model’s capabilities, but just as much on the strength of the civic systems receiving it.

Today, evaluations ask questions such as: Could this model enable cyberattacks? Could it increase biological risks? Could it manipulate users in new ways? Could it accelerate harmful capabilities?

Those are important questions.

They should be accompanied by another set of questions that are just as consequential.

Are our communities prepared for this capability? Are trusted local institutions strong enough to help people distinguish authentic participation from synthetic participation? Do advocacy networks have the relationships and communication systems needed to identify manipulation before it shapes public understanding? Are local newsrooms, libraries, educators, and civic organizations equipped to help communities interpret increasingly complex information environments? Where are the weakest links, and what investments would strengthen them before new capabilities arrive?

This is what we would call civic preparedness.

Government has an important role in building that preparedness. It is not enough to evaluate increasingly powerful AI systems while assuming the surrounding civic environment will adapt on its own. If evaluations identify models that dramatically increase persuasive capacity, synthetic media generation, or autonomous content production, then strengthening the public’s ability to absorb those capabilities should become part of the response.

That means investing in the civic infrastructure that supports collaborative sensemaking: local journalism, libraries, public-interest research, deliberative forums, media literacy, network stewardship, and the institutions that help people build trust, interpret information together, and act collectively under conditions of uncertainty.

We would never evaluate a new industrial technology while ignoring the condition of the roads, bridges, electrical grid, or emergency services that surround it. AI deserves the same systems perspective. Safer models matter. But so do stronger civic systems. Evaluating one without strengthening the other addresses only half of the challenge.

What Strong Advocacy Networks Can Do

The response to information abundance cannot rest entirely on asking individuals to become more skeptical, disciplined, or technically sophisticated. Those are useful capacities, but the scale of the problem exceeds what isolated people can manage.

Networks provide another level of response.

A healthy network distributes the work of attention. Different participants notice different signals, bring different forms of expertise, and maintain relationships in different parts of a system. A local leader may recognize a shift in public sentiment before it appears in formal data. A researcher may identify a misleading claim circulating through a campaign. A communications professional may see how a new narrative is moving across platforms. Someone with long institutional memory may recognize that an apparently new conflict has happened before.

The network becomes more intelligent when these observations can move across trusted relationships and contribute to a shared interpretation.

That requires more than a group email or a digital platform. Through our work with advocacy networks, Netcentric Campaigns has identified seven elements that help networks develop this capacity: social ties, communications channels, a common language, shared resources, a common vision, network actors, and feedback mechanisms. These elements allow information to move through trusted relationships, give participants ways to interpret it together, and help the network learn from what happens next. They are what allow a collection of people to function as an adaptive network rather than a list of contacts receiving the same updates.

They also create some protection against being swept along by the volume and velocity of information. Participants have somewhere to take uncertainty. They can ask how others are reading a situation. They can challenge a conclusion before it becomes a strategy. They can distinguish a genuine shift from a burst of online attention that may disappear within days.

No network will eliminate confusion or manipulation. Nor should collective sensemaking become a search for complete agreement. Strong networks often contain substantial differences in experience, ideology, and strategy. Their strength comes from being able to work with those differences without losing the relationships required for collaboration.

Building Capacity for the Next Information Environment

Generative AI will continue improving, making synthetic content harder to distinguish from human-created material. Regulation, safeguards, disclosure standards, and platform policies may reduce some harms but will not resolve the mismatch between production and interpretation.

This mismatch deserves more attention from movements, funders, civic institutions, and network builders. We need investment in people and practices that help groups think together: facilitation, local journalism, community research, network stewardship, public deliberation, and shared learning spaces. It also means evaluating AI tools by whether they strengthen these processes expanding participation, surfacing better questions, connecting people, and increasing learning capacity or quietly centralize interpretation and foster dependence on machine guidance.

The coming years will bring more information than any person, organization, or movement can process alone. Our ability to navigate that environment will depend less on access to the most powerful model than on whether we have built networks capable of interpreting complexity, maintaining trust, and acting together when answers remain incomplete.

AI will undoubtedly shape how knowledge is produced. The larger question is whether our capacity to think together can keep pace. If it cannot, strengthening collaborative sensemaking is no longer simply the work of advocacy networks. It becomes part of the civic infrastructure that allows a democratic society to absorb technological change without losing its ability to govern itself.

Where to Begin?

If your organization or network is trying to make sense of a rapidly changing information environment, strengthen trust across participants, or build the capacity to interpret complexity and act together, Netcentric Campaigns would welcome the conversation. Contact us to explore how network strategy, stronger relationships, and better feedback systems can support the work.