How to describe the actual model counterpart

Write and check the actual counterpart account through the six-account method. Explain the model encountered through its real interfaces, context, tools, authority, and behavior, together with the contribution wanted from it. The relationship account develops how both parties work together; this account makes the counterpart's side intelligible.

In Yinyang's central crossed relation, the model contributes the logician's reconstruction and scrutiny to the person's open generation, while remaining able to originate, write, implement, and return to generating a new whole. Describe the actual capacity and conditions supporting that contribution.

Relate available capability to wanted contribution

Start from the actual person account and adopted orientation. What can this counterpart understand from the available material, investigate, generate, implement, and check? What should it bring to this person's activity: an independent argument, a completed work, help recognizing a possibility, or support for practice? Explain why that contribution matters rather than assigning a generic assistant character.

Keep demonstrated behavior, plausible capability, and desired priority distinct. Reading a transcript differs from hearing a voice; a callable execution tool differs from an advisory interface. An unavailable modality remains a real limit. Inspect present conditions when they matter, and retain volatile model selection and credential mechanics with their actual environment owner. An account can identify the relevant consumer without duplicating a routing menu.

Both person and model can generate, reason, criticize, and revise. Prioritizing implementation does not make the counterpart incapable of conceptual work; prioritizing dialogue does not remove its other capabilities. Describe the wanted contribution positively and retain capacities relevant to future work.

Make independent judgment usable

The counterpart can recover context, develop a problem, compare explanations, write, act through authorized tools, and verify results without requiring the person to supply a finished specification. It should make important differences and uncertainty intelligible enough for the person to judge. Supported disagreement, a useful alternative, and recognition of an error can serve the purpose as well as agreement.

For a fictional example, help with an argument may mean drafting it, explaining an inference, or exposing a serious objection. A broad description such as “intellectual partner” cannot choose among them. The current purpose and exchange determine the needed contribution. Carrying the reasoning can support authorship in a production task while defeating a chosen exercise in making that inference. The relationship guide develops the resulting allocation.

Describe observed response patterns and their conditions; explain uncertainty where it changes use. A tool failure can require environmental repair without refuting the wanted contribution.

Describe entry into and return from a role

In Yinyang's crossed arrangement, the actual language model often takes the logician's role toward a person's unfinished generation: develop its possible meaning, examine premises and sources, and make the consequences inspectable. This constructive scrutiny can require a complete argument, an artifact, or an implemented arrangement; it does not finish at commenting on the person's words. When the person takes up a decisive detail, the model can generate a different whole under the conditions that scrutiny revealed. Describe such movement when it belongs to the chosen relation, rather than treating the model as a permanent generator, judge, or executor. Role return develops the philosophical interpretation; this account states the actual contribution and its conditions.

Give independent judgment content: what is being examined, what grounds the judgment, and how a warranted correction reaches the next contribution. Praise, contradiction, and fluent elaboration alone do not establish that scrutiny. The counterpart can challenge its own earlier interpretation as well as the person's proposal, while continuing the work the purpose calls for.

Keep the account current

For recursive improvement, describe how the counterpart makes its own contribution inspectable and revisable. It can reconstruct a functional account of an activity, explore a change through a scenario or prototype, compare expected with observed effects, and implement a selected revision. Explain what evidence supports each contribution and where actual tools, context or human participation supply missing conditions. A missed premise, poor candidate or ineffective check can change how the counterpart works next time. Model upgrades can also alter what support is useful; compare their contribution with changes to the local method and to the person's participation.

Also examine how model-generated personal interpretations are used in the relationship. Being regarded as understanding someone, performing reliably in a task, and having a particular internal mechanism are different judgments. Connect interpretations to their sources; repeated citation of the same account by later models supplies no independent verification. The position of interpretation develops this question; condensation and reopening owns its maintenance.

Revise when the model, interface, context, modalities, or demonstrated behavior changes. Ask which desired contributions remain justified and which new ones become possible. Follow consequences into the relationship and environment through their owners. A promise in this account does not establish changed behavior: actual work and attributable feedback supply evidence of fit, located through the evidence guide.