Concept dictionary
Model Steering
Model steering is the use of prompts, constraints, examples, controls, or feedback to guide a model toward desired behavior.
- Review state
- Reviewed in POC
- Freshness
- Current
- Version
- 3
Also called
- Model Steering
- Steering
Remember it this waySteering shapes behavior; it does not create judgment.
In plain language
Model steering guides model behavior toward a desired task, style, policy, or constraint. Steering can come from prompts, examples, system instructions, decoding controls, fine-tuning, or feedback.
Why it matters
The steering method affects how durable and reliable the behavior is. A prompt-level preference is different from a trained behavior, and neither guarantees compliance in every context.
Example
A system prompt directs a model to answer as a concise safety reviewer, while examples demonstrate the required format and escalation threshold.
Common misconception
Steering does not prove that a model understands or agrees with the direction. It changes the conditions shaping its output and may fail under conflicting inputs.
Related terms
Sources
- Long Ouyang and colleagues, “Training Language Models to Follow Instructions with Human Feedback”shows post-training methods used to steer language-model behavior toward demonstrated and preferred responses.
- OpenAI, “Model Spec”documents instruction hierarchy and behavioral guidance as system-level steering mechanisms.
Learn more
- Compare [[system-prompt|System Prompt]], [[alignment|Alignment]], [[guardrails|Guardrails]], and [[fine-tuning|Fine-Tuning]] to place Model Steering in the surrounding system.
- Follow the cited source material above for the method, standard, or research context behind this draft.