---
name: tai-theme-llms-and-foundation-models
description: 'Use the Testing AI theme LLMs and Foundation Models to plan, review, or teach related AI quality work. Applies concepts and techniques from the book to testing AI, AI-generated software, and non-deterministic systems when relevant.'
---

# LLMs and Foundation Models

Skill name: `tai-theme-llms-and-foundation-models`

Based on **Testing AI: Engineering Confidence in AI Systems** by **Jason Arbon**.

## Theme Purpose

Use these approaches when testing LLM training, data quality, AI pollution, RLHF/RLAIF behavior, model bug reports, internal concepts, mechanism fit, tokenization artifacts, image generation, and vision-language models.

Apply these concepts when testing AI, AI-generated software, model-backed features, agents, search, chatbots, RAG systems, generated code, dynamic interfaces, or other software whose behavior can vary across runs, users, data, tools, or time.

## How To Use This Theme

- Identify the behavior, capability, risk, or release decision being evaluated.
- Choose the relevant concepts below and turn them into concrete eval cases, samples, traces, checks, rubrics, metrics, or release gates.
- Prefer evidence that supports a decision: ship, canary, hold, rollback, or collect more samples.
- Report by slices and severe failures when averages hide risk.
- Preserve enough evidence that another person or agent can understand what was tested, how it was measured, and why the recommendation follows.

## Concepts And Techniques To Apply

- Evaluate training data quality, contamination, AI pollution, representativeness, labeling quality, and data provenance.
- Test reward-model and RLHF/RLAIF behavior for over-optimization, refusal artifacts, sycophancy, and hidden tradeoffs.
- Write useful LLM bug reports that describe distributions, reproducible prompts, context, model versions, and failure classes.
- Use mechanism-aware tests that account for tokenization, context windows, attention, decoding, retrieval, tools, memory, multimodal encoders, and product wiring.
- Use model-internals tools when they help explain concepts, attention, representations, image generation, or vision-language behavior.
- Separate model capability testing from product-quality testing and release readiness.

## Reporting Guidance

- State what was tested and what population the evidence represents.
- Explain uncertainty, missing coverage, severe failures, and known blind spots.
- Connect findings to a concrete decision or next action.
- Use topic-specific chapter skills only when deeper detail is needed; this theme skill should stand alone as practical guidance.