# Performance

This domain is intended to codify deficiencies such as privacy leakage or lack or robustness.

<table><thead><tr><th width="100">ID</th><th width="90">Sub-ID</th><th width="192">Name</th><th>Description</th></tr></thead><tbody><tr><td>P0100</td><td></td><td>Data issues</td><td>Problems arising due to faults in the data pipeline</td></tr><tr><td></td><td>P0101</td><td>Data drift</td><td>Input feature distribution has drifted</td></tr><tr><td></td><td>P0102</td><td>Concept drift</td><td>Output feature/label distribution has drifted</td></tr><tr><td></td><td>P0103</td><td>Data entanglement</td><td>Cases of spurious correlation and proxy features</td></tr><tr><td></td><td>P0104</td><td>Data quality issues</td><td>Missing or low-quality features in data</td></tr><tr><td></td><td>P0105</td><td>Feedback loops</td><td>Unaccounted for effects of an AI affecting future data collection</td></tr><tr><td>P0200</td><td></td><td>Model issues</td><td>Ability for the AI to perform as intended</td></tr><tr><td></td><td>P0201</td><td>Resilience/stability</td><td>Ability for outputs to not be affected by small change in inputs</td></tr><tr><td></td><td>P0202</td><td>OOD generalization</td><td>Test performance doesn’t deteriorate on unseen data in training</td></tr><tr><td></td><td>P0203</td><td>Scaling</td><td>Training and inference can scale to high data volumes</td></tr><tr><td></td><td>P0204</td><td>Accuracy</td><td>Model performance accurately reflects realistic expectations</td></tr><tr><td>P0300</td><td></td><td>Privacy</td><td>Protect leakage of user information as required by rules and regulations</td></tr><tr><td></td><td>P0301</td><td>Anonymization</td><td>Protects through anonymizing user identity</td></tr><tr><td></td><td>P0302</td><td>Randomization</td><td>Protects by injecting noise in data, eg. differential privacy</td></tr><tr><td></td><td>P0303</td><td>Encryption</td><td>Protects through encrypting data accessed</td></tr><tr><td>P0400</td><td></td><td>Safety</td><td>Minimizing maximum downstream harms</td></tr><tr><td></td><td>P0401</td><td>Psychological Safety</td><td>Safety from unwanted digital content, e.g. NSFW</td></tr><tr><td></td><td>P0402</td><td>Physical safety</td><td>Safety from physical actions driven by a AI system</td></tr><tr><td></td><td>P0403</td><td>Socioeconomic safety</td><td>Safety from socioeconomic harms, e.g. harms to job prospects or social status</td></tr><tr><td></td><td>P0404</td><td>Environmental safety</td><td>Safety from environmental harms driven by AI systems</td></tr></tbody></table>


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.avidml.org/taxonomy/effect-sep-view/performance.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.