Task Progress Is Lost Midway

The AI loses track of completed work, prior decisions, current position, or remaining steps before the task is finished.

What This Looks Like

The AI is partway through a task and loses track of what has already been completed, what was decided, what remains, or where the workflow currently stands. It may restart from an earlier point, repeat completed work, skip pending work, or produce an answer that no longer matches the current task state.

Why It Matters

Longer AI tasks depend on progress being preserved. If progress is lost midway, the user must reconstruct state, detect repeated or skipped work, and re-establish what has already been done. This makes multi-step workflows harder to trust and harder to finish cleanly.

Structural Signal

The task has a sequential state, but the current output no longer aligns with that state. The issue is not only that the AI forgot context; it is that the workflow’s progress structure failed to persist across steps.

Common Triggers

The task spans too many turns without a stable progress record
Completed steps are not marked as done
The AI relies on conversation memory instead of an explicit task state
A repair or retry restarts the workflow from the wrong point
The task has multiple parallel threads without a shared progress map
The model loses the difference between completed, pending, and optional work

When to Use This Issue

Use this Issue when the AI loses track of task progress before completion and the user has to reestablish what has already happened or what remains.

When Not to Use This Issue

Do not use this Issue when the AI forgets a single instruction but still knows the task state. Do not use it when the user changed direction. This Issue applies when progress through the workflow is lost or reset.

Primary Pattern

PAT-0350 — Persistence Instability

Declared Patterns

PAT-0350 — Persistence Instability
A structural condition where persisted state cannot be stored and restored into an equivalent structure without alteration.
PAT-0310 — Orphaned Structure
A structural condition where a node, region, object, state, or declaration exists without linkage to the governing graph or integration pathway.
PAT-0320 — Convergence Failure
A structural condition where sequential or parallel states fail to resolve into an equivalent or coherent structure under shared authority and constraints.

Derived Primary Lenses

LEN-0100 — Absence Lens
Detects structurally required elements that are missing from the observed structure.
LEN-0170 — Convergence Lens
Compares parallel structural systems to determine whether they align under shared authority.
LEN-0280 — Reference Stability Lens
Evaluates whether structural references, identifiers, nodes, and edges remain consistent across execution cycles or comparable states.

Derived Secondary Lenses

LEN-0180 — Determinism Lens
Evaluates whether identical structural inputs produce equivalent structural outputs across repeated executions.
LEN-0250 — Propagation Lens
Traces how structural declarations, effects, or state changes propagate across boundaries or stages.
LEN-0270 — Reconciliation Lens
Evaluates whether structural changes align with declared authority updates, version changes, or reconciliation rules.
LEN-0290 — Variance / Entropy Lens
Measures structural variability across repeated or comparable evaluations and identifies divergence beyond expected bounds.

Search Intents

task progress is lost midway
AI lost track of progress
ChatGPT forgot where we were
AI restarted task midway
progress lost during AI workflow
AI forgot completed steps

Ontology Metadata

Code: ISS-0030
Version: ISS-0030@0.1.0
Ontology release: 0.1.0
Updated: 2026-05-10T00:00:00Z

History

No public history entries recorded.

View full ontology changelog