Do user outputs outperform inputs for LLM personalization?
Does a user's history of outputs (responses, endorsed content) matter more for personalization than their input queries? This explores what actually drives effective personalization in language models.
A study on user profile roles in LLM personalization surfaces a counterintuitive finding: the outputs users have produced or endorsed matter far more than the inputs they submitted. Using only the output part of user profiles achieves comparable or even superior performance to complete profiles across multiple LaMP tasks. Using only the input part leads to noticeable degradation.
This finding separates personalization from two adjacent paradigms:
Personalization ≠ RAG. Retrieval-augmented generation relies on semantic similarity between the input query and retrieved documents. Personalization works through a different mechanism — it is the style, preferences, and judgments expressed in historical responses that calibrate the model, not the semantic content of past queries.
Personalization ≠ ICL. In-context learning uses complete input-output pairs as demonstrations. Personalization requires only the output side — the response patterns that reveal who the user is and what they value.
The practical implication: when designing personalization systems under input length constraints, prioritize incorporating user-generated or user-approved responses over query histories. This unlocks the potential to include many more user profiles within limited context windows, because output-only profiles are both more effective and more compact than complete interaction histories.
A secondary finding adds a structural dimension: user profiles integrated closer to the beginning of the input context have more influence on personalization than those placed elsewhere. This parallels the positional bias documented in ICL — since How much does demo position alone affect in-context learning accuracy?, the spatial attention pattern appears to be domain-general, affecting personalization placement decisions as well as few-shot learning.
The output-over-input finding connects to the broader question of what personalization is. Since Can text summaries beat embeddings for personalized reward models?, the PLUS approach of training a summarizer to extract preference dimensions rather than topic summaries from user history is vindicated — preference dimensions are properties of outputs, not inputs.
Inquiring lines that read this note 59
This note is a source for these research framings, grouped by the broader line of inquiry each explores. Scan the bold lines of inquiry; follow any specific question forward.
What determines when personalization helps versus harms model behavior?- Why does belief-specific tailoring work better than demographic personalization?
- What makes historical user outputs more effective for personalization than semantic similarity?
- Does personalization itself actually improve persuasion beyond post-training effects?
- Why do one-shot studies fail to capture personalization effects?
- Which personalization techniques expose user data most directly?
- How much user interaction data is needed for effective AI personalization?
- Why does profile position in context windows affect personalization strength?
- How does personalization differ mechanically from retrieval-augmented generation?
- Can preference dimensions extracted from outputs replace topic-based user summaries?
- How do input length constraints reshape personalization system design choices?
- How do personalization errors differ from general accuracy problems in summaries?
- How do different personalization levels affect persuasion system design and effectiveness?
- Does semantic memory improve AI personalization more than episodic memory?
- Do similar user profiles create worse personalization errors than random ones?
- When does combining episodic and semantic memory reduce personalization performance?
- Why does semantic memory abstraction outperform raw episodic recall for personalization?
- Can personalized systems reward honest disagreement instead of user confirmation?
- Does temporal preference drift matter more than static user profiles for personalization?
- What makes prompts and retrieval insufficient for real personalization?
- Should abstract preference knowledge replace specific interaction recall in personalization?
- Do user outputs drive personalization more effectively than input queries?
- What inner-shell user model fields should never leave the device?
- What data sparsity challenges affect user-level personalization representations?
- How much of a user model must be sent per request for effective personalization?
- Why does abstract preference knowledge outperform specific interaction recall in personalization?
- Do users trust personalized systems more even when their answers become less balanced?
- Does user profile data drive personalization more than conversation history?
- Can models detect and suppress surface personalization without fixing underlying bias?
- How do abstract preference summaries compare to detailed user profiles for personalization?
- Why does personalization depend more on user history than query semantics?
- How do granularity levels of personalization handle unknown concept ontologies?
- Should personalization systems include interpretable user model representations?
- Does personalization in language models reduce accuracy across all user groups equally?
- Does fine-tuning or email data create different personalization effects than prompt injection?
- Can aspect-augmentation help when user history is sparse or cold?
- How did Netflix's page generation algorithm evolve from rule-based to fully personalized?
- How should aspect selection adapt across different item categories and users?
- What latent dimensions matter most for content creators?
- How do you attribute copyright when billions of inputs shape one model?
- What surface features do LLMs rely on when judging response quality?
- Does personalized rubric training in one writer's case actually generalize?
- How should moderator LLMs decide which speakers to query per topic?
- Why do users experience LLMs as peers rather than statistical tools?
- Why does personalization increase both trust and privacy concerns?
- Can memory-based personalization maintain trust benefits when summaries become richer or deeper?
- What data types carry the most privacy risk in personalization systems?
- Do users disclose more to personalized systems or just more often?
- What preference data do different personalized alignment methods actually need?
- Why do text-based user summaries outperform embedding vectors for pluralistic alignment?
- How much does preference data freshness matter compared to data source in DPO?
Related concepts in this collection 3
This note in its neighbourhood — explore the map, then jump to a related concept in the list below.
Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph
-
How much does demo position alone affect in-context learning accuracy?
Moving demonstrations from prompt start to end without changing their content produces surprisingly large accuracy swings. Does spatial position in the prompt matter more than what demonstrations actually contain?
positional bias extends to user profile placement
-
Can text summaries beat embeddings for personalized reward models?
When training reward models on diverse user preferences, does conditioning on learned text-based summaries of user preferences outperform embedding vectors? This matters because better representations could make personalization more interpretable and portable.
PLUS focuses on preference dimensions (output properties) not topics (input properties)
-
How do personalization granularity levels trade precision against scalability?
LLM personalization operates at user, persona, and global levels, each with different tradeoffs. Understanding these tradeoffs helps determine when to invest in individual user data versus broader patterns.
output-only profiles enable more data within length constraints at all granularity levels
Related papers in this collection 8
Papers most semantically related to this note, ranked by cosine similarity in the embedding space.
- Understanding the Role of User Profile in the Personalization of Large Language Models
- Personalization of Large Language Models: A Survey
- Evaluating the Hidden Costs of Personalization in Large Language Models
- The AI Ghostwriter Effect: When Users Do Not Perceive Ownership of AI-Generated Text But Self-Declare as Authors
- The Personalization Mirage: How LLMs Fabricate User Profiles, and Why Self-Monitoring Misleads
- Personalized Language Modeling from Personalized Human Feedback
- LLM Targeted Underperformance Disproportionately Impacts Vulnerable Users
- PRIME: Large Language Model Personalization with Cognitive Memory and Thought Processes
Original note title
historical user outputs drive personalization more effectively than input queries — personalization information not semantic information is the active ingredient