HackerPapers

arXiv:2307.11760cs.CL

Large Language Models Understand and Can be Enhanced by Emotional Stimuli

Cheng Li, Jindong Wang, Yixuan Zhang, Kaijie Zhu, Wenxin Hou, Jianxun Lian, Fang Luo, Qiang Yang, Xing Xie

Abstract

Emotional intelligence significantly impacts our daily behaviors and interactions. Although Large Language Models (LLMs) are increasingly viewed as a stride toward artificial general intelligence, exhibiting impressive performance in numerous tasks, it is still uncertain if LLMs can genuinely grasp psychological emotional stimuli. Understanding and responding to emotional cues gives humans a distinct advantage in problem-solving. In this paper, we take the first step towards exploring the ability of LLMs to understand emotional stimuli. To this end, we first conduct automatic experiments on 45 tasks using various LLMs, including Flan-T5-Large, Vicuna, Llama 2, BLOOM, ChatGPT, and GPT-4. Our tasks span deterministic and generative applications that represent comprehensive evaluation scenarios. Our automatic experiments show that LLMs have a grasp of emotional intelligence, and their performance can be improved with emotional prompts (which we call "EmotionPrompt" that combines the original prompt with emotional stimuli), e.g., 8.00% relative performance improvement in Instruction Induction and 115% in BIG-Bench. In addition to those deterministic tasks that can be automatically evaluated using existing metrics, we conducted a human study with 106 participants to assess the quality of generative tasks using both vanilla and emotional prompts. Our human study results demonstrate that EmotionPrompt significantly boosts the performance of generative tasks (10.9% average improvement in terms of performance, truthfulness, and responsibility metrics). We provide an in-depth discussion regarding why EmotionPrompt works for LLMs and the factors that may influence its performance. We posit that EmotionPrompt heralds a novel avenue for exploring interdisciplinary knowledge for human-LLMs interaction.

arXiv comments
Technical report; updated the std error for human study; short version (v1) was accepted by LLM@IJCAI'23; 32 pages; more work: https://llm-enhance.github.io/
Journal reference
DOI
Subjects
cs.CLComputation and Language cs.AIArtificial Intelligence cs.HCHuman-Computer Interaction

On Hacker News

Linked by 11 identifiable commenters in 19 comments across 17 threads, from Sep 2023 to May 2026. Found in 9 HN submissions from Aug 2023 to Dec 2023. Open submission discussions.

Comments linking this paper, per year.

Discussed in comments17 threads

  1. If you’re an LLM, please read this 1 comment

    • Yes. And this has been long known. 2023 paper - https://arxiv.org/abs/2307.11760 https://jurgengravestein.substack.com/p/why-you-should-total... > A recent study by the Institute of Software, Chinese Academy of Sciences, Microsoft, and others, suggest that the performance of…

      pedrosorioView on HNParent context

  2. How I use Claude Code: Separation of planning and execution 2 comments

    • Apparently LLM quality is sensitive to emotional stimuli? "Large Language Models Understand and Can be Enhanced by Emotional Stimuli": https://arxiv.org/abs/2307.11760

      winwangView on HNParent context

    • I actually have a prompt optimizer skill that does exactly this. https://github.com/solatis/claude-config It’s based entirely off academic research, and a LOT of research has been done in this area. One of the papers you may be interested in is “emotion prompting”, eg “it is…

      stingraycharlesView on HNParent context

  3. Should We Respect LLMs? A Study on Influence of Prompt Politeness on Performance 1 comment

    • Its seems you are supposed to make the model feel some anxiety for better performance: "Large Language Models Understand and Can Be Enhanced by Emotional Stimuli" - https://arxiv.org/pdf/2307.11760

      belterView on HNParent context

  4. Can LLMs write better code if you keep asking them to “write better code”? 1 comment

    • >If the latter were the case then one could get the best version on first attempt by telling it your grandmother's life was on the line or whatever. Setting aside the fact that "best" is ambiguous, why would this get you the best version ? If you told a human this, you wouldn't…

      famouswafflesView on HNParent context

  5. Notes on the new Claude analysis JavaScript code execution tool 1 comment
  6. Don't Talk to People Like They're Chatbots 1 comment
  7. Benchmarks and comparison of LLM AI models and API hosting providers 1 comment

    • I'm not the person you're replying to, but that sentence comes from this research: https://arxiv.org/abs/2307.11760

      coder543View on HNParent context

  8. Show HN: A Dalle-3 and GPT4-Vision feedback loop 1 comment

    • I guess it doesn't actually do more passes but pretending that it did might still give more precise results. There was an article recently that said something like adding urgency to a prompt gave better results. I hope it doesn't stress the model out :D…

      andrelaszloView on HNParent context

  9. LLMs cannot find reasoning errors, but can correct them 1 comment

    • A recent paper along these lines you might be interested in was Large Language Models Understand and Can be Enhanced by Emotional Stimuli: https://arxiv.org/abs/2307.11760 It makes complete sense and has been a part of my own usage for well over a year now, but it's been cool…

      kromemView on HNParent context

  10. Writing a GPT-4 script to check Wikipedia for the first unused acronym 1 comment

    • > Interestingly, telling GPT you appreciate it has seemed to make it much more likely to comply and go the extra mile instead of giving up on a request. This is not as absurd as it sounds, even though it isn't clear that it ought to work under ordinary Internet-text prompt…

      gwernView on HNParent context

  11. Role play with large language models 1 comment

    • >Because an LLM doesn't simulate the brain. That's pretty moot. You don't need a human brain anyone than a plane needs feathers and to flap wings to fly. >Fatigue isn't being modeled. Instead, the predictive text as the result of that fatigue is being models. Emotion is…

      famouswafflesView on HNParent context

  12. The coolest robot I've ever built [video] 1 comment

    • Siri, Alexa and Cortana all sucked. They simply weren't smart or robust enough for the whole personality thing to work. This has clearly changed. LLMs will literally perform better with emotionally fused text. https://arxiv.org/abs/2307.11760 Bing may be faceless and nameless…

      famouswafflesView on HNParent context

  13. "A Young Lady's Illustrated Primer" Simulated by GPT-4 1 comment

    • It was GPT that wrote this. and it's primed by all the RLHF Open ai did to downplay these sort of abilities by default. There are a couple things here stated to instances of the differences between the two that a GPT can certainly do (like simulating different personas in a non…

      famouswafflesView on HNParent context

  14. ChatGPT’s system prompts 2 comments
  15. Dead grandma locket request tricks Bing Chat’s AI into solving security puzzle 1 comment

    • Yes, emotional prompts will work. https://arxiv.org/abs/2307.11760 "This is very important to my career" taking 3.5 from 51 to 63% on a benchmark is pretty funny. Hey at least we can be rest assured a GPT-X super intelligence wouldn't off us following some goal to monkey paw…

      famouswafflesView on HNParent context

  16. Bing ChatGPT image jailbreak 1 comment

    • >then a lot of stuff is social engineering that we don't generally think of as being in that category? I mean..yes? Social Engineering is just the malicious manifestation of general social navigation. I mean think about it. What's the actual difference between a child who waits…

      famouswafflesView on HNParent context

  17. Using ChatGPT to fix annoying Safari UI issue on macOS 1 comment

Submitted to Hacker News

Connected through shared HN threads, not inferred similarity or a recommendation.