mlsanityData to 7 Oct 2026

Can Hallucinations Be Useful? Solving Multi-Hop Questions With SLMs By Chaining System-I/II Reasoning

arXiv 2605.27596 ↗First posted 26 May 2026cs.CLcited by 0

On the map: LLM reasoning → Hallucination detection and mitigation

Abstract (arXiv)

Recently, there has been increased interest in Small Language Models (SLMs), which are fast, show good performance, and have lower hardware demands than large language models (LLMs). However, SLMs hallucinate more frequently than LLMs, impacting their ability to solve complex multi-step reasoning problems as early mistakes cascade to the final response. To address this, existing works think-first followed by iterative retrieval to reduce hallucination. We argue that the think-first strategy is not always necessary as we find that: (i) SLMs are often accurately confident in their initial answer and, (ii) hallucinations can actually be beneficial for honing in on the true answer. As such, we position our work as an inversion of this strategy, i.e., answer first-reason later. We propose a cognitively-inspired framework where the model is first allowed to quickly answer the question (System-I (zero-shot)) and then resorts to deeper thinking (System-II) based on evidence retrieved from a knowledge source using the initial hypothesis. By combining System-I and System-II style thinking, we show that our method can outperform prior work that takes the traditional think-first route on various multi-step question-answering benchmarks.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Hallucination detection and mitigation, one of 17 directions in LLM reasoning. Builds on 15 of its 64 references that are in the map. First posted on arXiv 26 May 2026.

Builds on

15 of 64 references.

References · 15

  1. 2025Qwen3 Technical Reportcited by 8,195
  2. 2023Retrieval-Augmented Generation for Large Language Models: A Surveycited by 2,185
  3. 2025Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Modelscited by 1,175
  4. 2025Kimi K2: Open Agentic Intelligencecited by 523
  5. 2025R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learningcited by 322
  6. 2025Agentic Retrieval-Augmented Generation: A Survey on Agentic RAGcited by 242
  7. 2025A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?cited by 215
  8. 2025Graph-R1: Towards Agentic GraphRAG Framework via End-to-end Reinforcement Learningcited by 50
  9. 2025Guiding Reasoning in Small Language Models with LLM Assistancecited by 13
  10. 2026Reasoning in Trees: Improving Retrieval-Augmented Generation for Multi-Hop Question Answeringcited by 13
  11. 2025Test-Time Scaling in Reasoning Models Is Not Effective for Knowledge-Intensive Tasks Yetcited by 10
  12. 2024Small Language Models for Application Interactions: A Case Studycited by 5
3 more
  1. 2026Finding What Matters: Anchoring Context Knowledge with Evolving Indices for Iterative Retrievalcited by 2
  2. 2026D2Plan: Dual-Agent Dynamic Global Planning for Complex Retrieval-Augmented Reasoningcited by 1
  3. 2026SEEK: Steering LLM Reasoning for RAG via Internal Reasoning Sketchescited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Think More, Hallucinate Less: Mitigating Hallucinations via Dual Process of Fast and Slow Thinking
  2. 2025Don't Let It Hallucinate: Premise Verification via Retrieval-Augmented Logical Reasoning
  3. 2026Mitigating Prompt-Induced Hallucinations in Large Language Models via Structured Reasoning
  4. 2025Auditing Meta-Cognitive Hallucinations in Reasoning Large Language Models
  5. 2025Improving the Reliability of LLMs: Combining CoT, RAG, Self-Consistency, and Self-Verification
  6. 2025Mitigating Hallucination in Large Language Models (LLMs): An Application-Oriented Survey on RAG, Reasoning, and Agentic Systems
  7. 2025Are Reasoning Models More Prone to Hallucination?
  8. 2026PRISM: Probing Reasoning, Instruction, and Source Memory in LLM Hallucinations