Getting the Words Right

Summary: AI creates real opportunities around translation, synthesis, coordination, and adaptive views, but the caveat is consistent: it can also harden bad abstractions, amplify weak data, and scale portable legibility at the expense of interaction-generated understanding. Many recurring organizational problems become easier to see through the language of endurants, perdurants, containers, anchors, legibility, and métis. The broad pattern is simple. Many organizational problems persist because the work is messy, the concepts are fuzzy, the views are partial, and the coordination burden is real. AI can help with many of these problems, but it can also make the same problems harder to see while spreading them further and faster. If context is not just transmitted but often generated through interaction, then the question is not only what AI can summarize or translate. The deeper question is whether AI is helping people create shared understanding, or only packaging existing fragments more efficiently. Seen through the4Es, that means asking:

• is AI improving contact with reality (embodied)?

• is it making the situation clearer (embedded)?

• is it helping cognition move across people and tools (extended)?

•is it supporting the interaction through which understanding emerges (enactive)? At a Glance

AI OPPORTUNITIES (AND CAVEATS)87 Problem areaReal opportunityReal caveat Integration and translation adapt across local variation without forcing one taxonomy scale semantic confusion faster Exceptions and messy work support judgment on edge cases without full automation over-automate the rare cases that need human sensemaking Data quality and cognitive load reduce maintenance burden and keep artifacts fresher add noise, prompts, and disengagement Recontextualization and multiple lenses generate views for different audiences from shared sources create subtle drift across those views Narrative and relationships surface patterns across scattered artifacts and signals tell convincing stories from weak relationships Centralization and control support local adaptation on top of shared structures recentralize behavior around what is easiest to measure Translation Layers One of the clearest opportunities is translation. Teams regularly use the same words differently. One team’sepicis another team’s initiative. One system’s dependency is another team’s annotation. AI can help reconcile these local differences without forcing strict standardization up front. That is valuable because many integration problems are really problems of meaning, not just data plumbing. In Shannon-style terms, AI looks helpful because it can move more context around more quickly. But the caveat is equally important. If the underlying endurants are fuzzy and the local usage is inconsistent, AI may simply learn the mess and reproduce it. It becomes a translation layer on top of semantic drift rather than a real improvement in understanding. It makes context more portable without necessarily making it more shared. That helps the extendedside of cognition, but may leave theenactiveside untouched.

AI OPPORTUNITIES (AND CAVEATS)88 Messy Work and Exceptions AI also creates opportunities around messy work and exception handling. That matters because knowledge work is full of:

•fuzzy categories •evolving work boundaries •one-off anomalies •partial exceptions that do not justify full automation Used well, AI can help teams spot anomalies, suggest resolutions, and support semi-structured judgment without requiring everything to be turned into a rigid workflow. The caveat is that some mess is productive. Some ambiguity helps teams think, debate, and learn. If AI smooths away the productive friction too early, it can weaken local sensemaking and push teams toward overconfidence. In 4E terms, it can interrupt theenactiveandembodiedwork through which people discover what is actually going on. Keeping Artifacts Alive Another opportunity is reducing the maintenance tax on important but adjacent artifacts. Capability maps, strategy docs, planning sheets, and other support structures often decay because the effort to keep them current competes with the actual work. AI can help by suggesting updates, flagging stale areas, and reducing the manual burden of completeness. This is promising, but only if the prompts, suggestions, and interventions feel trustworthy. If the system creates more noise than value, people will disengage. Once that happens, the artifact still looks alive while the underlying trust disappears. Theembeddedenvironment appears healthy while the real cognitive support has decayed. Multiple Lenses, One System Many of the forever problems are really about different lenses on the same underlying reality.

• executives want coherence
• operators want actionable detail

AI OPPORTUNITIES (AND CAVEATS)89 •finance wants investment logic •product wants outcomes and learning •delivery teams want workable coordination AI can help produce multiple views from shared sources, tailoring what matters for each audience without requiring endless manual recontextualization. That is a real opportunity. But it also creates a familiar caveat from earlier notes in this series: different containers can drift away from the same underlying anchors. If AI generates a slightly different story for each audience, people may feel aligned while actually reasoning from incompatible assumptions. What looks like shared context may only be parallel interpretation. Narratives, Relationships, and “My Dots” AI is also well suited to work across relationships rather than just records. That matters because real organizational understanding is rarely built from isolated objects. People connect:

•goals •risks •customer needs •metrics •dependencies •narratives across time AI can help surface those links, summarize across scattered artifacts, and generate “my dots” views that make the system more navigable. But narrative power is dangerous. A coherent AI-generated story can feel true even when the underlying relationships are weak. This is another version of the legibility problem: the representation becomes compelling enough that people stop asking whether it is faithful. A generated narrative can substitute for the interaction that would have tested it. It can strengthen theextendedartifact layer while weakening the enactive process that should have challenged it. Official, Real, Ideal One of the deepest opportunities is helping organizations navigate the gap between the official model, the real model, and the ideal model.

AI OPPORTUNITIES (AND CAVEATS)90

AI can help identify where today’s process is decaying, where the real

work has drifted, and where teams are compensating. It can highlight

patterns that suggest what to keep, what to adapt, and what to sunset. But if AI overfits to the current state, it can become a machine for fossilizing the present. The system gets very good at describing how things work today while making it harder to move toward a better future state. The Series Lens Read through the language of this series, the recurring question is not simply whether AI is useful. It is what kind of thing AI is helping us stabilize, and what kind of process it is helping us navigate.

• If AI strengthens the wrong endurants, it hardens bad containers.

• If AI hides the relevant perdurants, it removes the history and transitions needed for interpretation.

•If AI improves legibility without preserving métis, it can make the system cleaner and less truthful at the same time.

• If AI helps people work across anchors, histories, and situated interaction, it can genuinely reduce friction.

•If AI supports the4Eswell, it can strengthen reality contact, situational awareness, distributed cognition, and shared sensemaking instead of just speeding up output. The setting matters too:

•in clearer settings, AI can often automate or retrieve context effectively •in complicated settings, AI can help coordinate distributed context across bounded structures •in complex settings, AI is most valuable when it supports shared sensemaking rather than pretending to replace it The Core Insight AI is most useful when it helps people work with complexity rather than pretending complexity has disappeared. The opportunity is not just better automation. It is better translation, better synthesis, better navigation across lenses, better support for local judgment, and better interfaces between messy reality and organizational legibility.

AI OPPORTUNITIES (AND CAVEATS)91

The caveat is just as important. AI can turn weak abstractions into

stronger illusions. It can make bad containers look authoritative, make

stale artifacts look alive, and make partial stories feel complete. When

that happens, AI does not reduce the need for interaction. It merely hides that need behind more convincing outputs. Try This Now:

• Pick one place where AI might help in your organization: translation, synthesis, status, planning, or navigation.

• Write down what weak abstraction, missing context, or fuzzy category sits underneath that use case.

•Ask: would AI help people work with the complexity there, or mainly make a shaky model look more convincing?