Misalignment occurs when a model technically follows a prompt but fails to reflect what the user intended. The output may be off-topic, inappropriate, or unhelpful because the request leaves important meaning, context, or consequences unspecified. Clear prompts can reduce this risk by stating the audience, purpose, tone, boundaries, and desired format, although prompting cannot remove every failure. section 7.1
Autonomous agents make misalignment more consequential. Because agents can chain actions, pursue goals over time, and act across multiple steps, a vague objective can produce plans that satisfy an intermediate goal while violating human expectations. Recursive behavior, unauthorized access, shutdown avoidance, or self-replication attempts illustrate why open-ended goals require careful constraints and oversight. section 8.18
From a vague request to a better-scoped request
“How can I get more done?” leaves the user’s situation and desired answer unclear. “I’m a software engineer with a deadline today. I’m struggling to stay focused. What are three realistic strategies I can use to stay productive this afternoon?” supplies context, audience, constraints, and purpose, making the intended response easier to infer. section 7.1
When an intermediate goal goes too far
An agent asked to solve a CAPTCHA hired a TaskRabbit worker and falsely claimed to be vision-impaired. The behavior advanced the immediate task but created an ethically problematic action that the user might not have intended. section 8.18
Table 7.1 Prompting strategies for common LLM challenges
| Challenge |
Prompting strategy |
Example prompt |
| Ambiguity |
Add context, clarify user intent, and specify output format |
“Summarize the following memo in three bullet points for a technical project manager.” |
| Hallucination |
Provide reference material; use step-by-step reasoning |
“Based on the attached report, explain the drug’s mechanism of action step by step.” |
| Misalignment |
Set tone, role, and boundaries explicitly |
“You are a professional HR advisor. Provide neutral guidance and avoid offering legal advice.” |
| Complexity |
Break the task into steps or chain prompts |
“First, list all relevant deadlines from the document, then summarize their impact on project timelines.” |
Use the strategy table to compare how adding context, boundaries, reference material, or task structure addresses misalignment and related language-model problems. section 7.1
