2000 character limit reached
Insert-expansions for Tool-enabled Conversational Agents (2307.01644v1)
Published 4 Jul 2023 in cs.HC, cs.AI, and cs.CL
Abstract: This paper delves into an advanced implementation of Chain-of-Thought-Prompting in LLMs, focusing on the use of tools (or "plug-ins") within the explicit reasoning paths generated by this prompting method. We find that tool-enabled conversational agents often become sidetracked, as additional context from tools like search engines or calculators diverts from original user intents. To address this, we explore a concept wherein the user becomes the tool, providing necessary details and refining their requests. Through Conversation Analysis, we characterize this interaction as insert-expansion - an intermediary conversation designed to facilitate the preferred response. We explore possibilities arising from this 'user-as-a-tool' approach in two empirical studies using direct comparison, and find benefits in the recommendation domain.
- Nikola A. Kompa. Inner speech and ‘pure’ thought – do we think in language? Review of Philosophy and Psychology, 2023.
- Deep learning. Nature, 521(7553):436–444, 2015.
- Multilayer feedforward networks are universal approximators. Neural Networks, 2(5):359–366, 1989.
- Attention is all you need. In Guyon I., Von Luxburg U., Bengio S., Wallach H., Fergus R., Vishwanathan S., and Garnett R., editors, Advances in Neural Information Processing Systems, volume 30. Curran Associates, Inc, 2017.
- Improving language understanding by generative pre-training.
- Language models are few-shot learners.
- Deep reinforcement learning from human preferences. Advances in neural information processing systems, 30, 2017.
- Sparks of artificial general intelligence: Early experiments with gpt-4.
- Summary of chatgpt/gpt-4 research and perspective towards the future of large language models.
- Chain-of-thought prompting elicits reasoning in large language models.
- Tarek R. Dika. Descartes’ method. In Edward N. Zalta and Uri Nodelman, editors, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, 2023.
- Ai chains: Transparent and controllable human-ai interaction by chaining large language model prompts. In Simone Barbosa, Cliff Lampe, Caroline Appert, David A. Shamma, Steven Drucker, Julie Williamson, and Koji Yatani, editors, CHI Conference on Human Factors in Computing Systems, pages 1–22, New York, NY, USA, 2022. ACM.
- Arthur N. Applebee. Writing and reasoning. Review of Educational Research, 54(4):577–596, 1984.
- Talm: Tool augmented language models.
- Augmented language models: a survey.
- Kenneth Oakley. Tools makyth man. Antiquity, 31(124):199–209, 1957.
- Emanuel A. Schegloff. Sequence organization in interaction: A Primer in Conversation Analysis I. A primer in conversation analysis : v. 1. Cambridge University Press, Cambridge, 2007.
- Designing pedagogical conversational agents for achieving common ground. In International Conference on Design Science Research in Information Systems and Technology, pages 345–359, 2023.
- Sequence organization: A universal infrastructure for social action. Journal of Pragmatics, 168:119–138, 2020.
- Vincenza Tudini. Extending prior posts in dyadic online text chat. Discourse Processes, 52(8):642–669, 2015.
- Joanne Meredith. Conversation analysis and online interaction. Research on Language and Social Interaction, 52(3):241–256, 2019.
- Estimating the impact of “humanizing” customer service chatbots. Information Systems Research, 32(3):736–751, 2021.
- On seeing human: a three-factor theory of anthropomorphism. Psychological review, 114(4):864–886, 2007.
- Dissociating language and thought in large language models: a cognitive perspective.
- Rewoo: Decoupling reasoning from observations for efficient augmented language models.
- Modular brain networks. Annual review of psychology, 67:613–640, 2016.
- Chemcrow: Augmenting large-language models with chemistry tools.
- Toolformer: Language models can teach themselves to use tools.
- Prompt sapper: A llm-empowered production tool for building ai chains.
- Small models are valuable plug-ins for large language models.
- Gpt4tools: Teaching large language model to use tools via self-instruction.
- Toolkengpt: Augmenting frozen language models with massive tools via tool embeddings.
- Plan-and-solve prompting: Improving zero-shot chain-of-thought reasoning by large language models.
- On the tool manipulation capability of open-source large language models.
- Api-bank: A benchmark for tool-augmented llms.
- Measurement properties of completely and end labeled unipolar and bipolar scales in likert-type questions on income (in)equality. Social science research, 97:102544, 2021.
- The chatbot usability scale: the design and pilot of a usability scale for interaction with ai-based conversational agents. Personal and Ubiquitous Computing, 26(1):95–119, 2022.
- Distilbert, a distilled version of bert: smaller, faster, cheaper and lighter.
- Short and extra-short forms of the big five inventory–2: The bfi-2-s and bfi-2-xs. Journal of Research in Personality, 68:69–81, 2017.
- Arne Roets and Alain van Hiel. Item selection and validation of a brief, 15-item version of the need for closure scale. Personality and Individual Differences, 50(1):90–94, 2011.