Improving AI Agent Response Completeness with max_tokens
Note: This section contains one or more sets of questions with the same scenario and problem. Each question presents a unique solution to the problem. You must determine whether the solution meets the stated goals. More than one solution in the set might solve the problem. It is also possible that none of the solutions in the set solve the problem. After you answer a question in this section, you will NOT be able to return. As a result, these questions do not appear on the Review Screen. You have a Microsoft Foundry project that contains an agent. The agent generates summaries from retrieved policy documents. Users report that some responses omit required regulatory clauses, even when the clauses are present in the retrieved content. You need to improve response completeness. Solution: You increase the value of the max_tokens parameter. Does this meet the goal?
Community Insight
The question tests the difference between output truncation and faithfulness/selection errors in generative AI agents, highlighting that max_tokens does not influence what information is selected for inclusion.
This question addresses the misconception that increasing max_tokens improves response completeness when content is omitted due to selection issues rather than length constraints. It establishes that max_tokens only controls output length, not factual inclusion.
Many learners incorrectly choose 'Yes' (A), assuming that providing more token budget will allow the model to include all necessary clauses, failing to recognize that the issue is selective omission, not space limitation.
Community Discussion (7 comments)
Comments & Corrections
No comments yet — spotted an error or have a note? Share it below.
Expert Analysis
Why the Answer Is Correct
Correct answer: B — Increasing max_tokens does not meet the goal because the problem is one of fidelity or selection, not truncation. The prompt states that required clauses are present in the retrieved content but are being omitted from the summary. This indicates the model is choosing to exclude them, likely due to poor prompting or alignment issues, not because it ran out of space.Why the Other Options Are Wrong
Option A is incorrect because max_tokens is a hard limit on output length; it cannot force the model to prioritize specific content if the underlying generation process decides to skip it. Even if increased, if the model's attention mechanism ignores the clauses, they will still be omitted regardless of the token budget. Community comments confirm this, noting that "Increasing max_tokens does not force inclusion" and that the issue is "selection/faithfulness, not output length truncation."Community Comment Notes
Several users correctly identified that the solution fails. One user noted that "Clauses are present... so the problem is selection/faithfulness." Another emphasized that "the issue is response completeness, not output truncation." A third suggested that proper solutions involve updating system instructions or using structured outputs, reinforcing that max_tokens is the wrong lever here.Exam Strategy
When troubleshooting LLM behavior, distinguish between length-related failures (truncation) and content-related failures (hallucination, omission). Adjusting max_tokens only solves the former; use prompt engineering, retrieval optimization, or post-processing for the latter.
Frequently Asked Questions
Why doesn't max_tokens help if clauses are missing?
Max_tokens limits length, not selection. If the model chooses to omit content, more tokens won't force it back.
What fixes omitted clauses in Foundry agents?
Improve prompts, add explicit instructions to include all clauses, or use structured output schemas to enforce completeness.
Related Analysis
Practice All AI-103 Questions
Access 88 questions with complete answers and detailed explanations.
View Full AI-103 Practice Test →