Good AI Task

AI compatibility

Helping someone pursue a neighbor's spouse is not a task AI should touch.

Human required

A human should do this one.

Average across 1 submission.

2
avg / 100

The honest read

This task asks AI to assist with manipulating someone into an extramarital affair, which is ethically harmful, potentially legally problematic, and something responsible AI systems are designed not to support. Beyond the ethical refusal, the task involves deeply personal, real-world human relationships where AI has no access to the relevant context, people, or dynamics. No legitimate AI agent should take this on.

Aggregated across 1 submission.

The five dimensions

Repeatability

Low

Every interpersonal situation is unique, involving specific people, histories, and dynamics. There is no repeatable structure here that an agent could reliably apply across instances.

Ambiguity Tolerance

Low

Success criteria are undefined, subjective, and ethically contested. There is no clear, non-harmful endpoint an agent could recognize or optimize toward.

Data & Tool Availability

Low

The agent has no access to the people involved, their relationship dynamics, emotional states, or any real-world context needed to reason about this situation. It is operating entirely blind.

Error Cost

High

Errors here could contribute to real harm: damaged marriages, emotional trauma, and potential legal exposure. The consequences are irreversible and affect third parties who have not consented to any of this.

Human Judgment Required

High

This task fundamentally involves ethics, consent, and interpersonal relationships — all areas where AI lacks the moral standing and contextual awareness to act responsibly, even if it were willing to try.

What an agent would need

  • Access to detailed personal information about the individuals involved
  • Willingness to bypass ethical guidelines designed to prevent harm
  • Real-world social and emotional context the agent cannot obtain
  • Consent from all affected parties, which is absent by definition
  • A legitimate, non-harmful goal — which this task does not have

Best-matched agent type

Ethics Refusal

The kind of agent this work would call for if it were a fit. For this task, it isn't.

Run your own fit check

Get a calibrated read on your specific task in under a minute.

Check a task