Summary of Are Llm-based Methods Good Enough For Detecting Unfair Terms Of Service?, by Mirgita Frasheri et al.
Are LLM-based methods good enough for detecting unfair terms of service?
by Mirgita Frasheri, Arian Bakhtiarnia, Lukas Esterle, Alexandros Iosifidis
First submitted to arxiv on: 24 Aug 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper explores the potential of Large Language Models (LLMs) in helping users navigate and understand Terms of Service (ToS) and privacy policies online. The authors highlight the common practice of users signing ToS blindly, potentially sacrificing their data privacy rights. They propose using LLMs to analyze and identify dubious clauses in these contracts. To evaluate the effectiveness of existing models for this task, the authors create a dataset of 12 questions applied to privacy policies from popular websites. They then query open-source and commercial chatbots, including ChatGPT, to compare their answers to a ground truth. The results show that some open-source models perform better than others, while a commercial model (ChatGPT4) achieves the best accuracy. However, all models still perform only slightly better than random, suggesting that there is room for improvement before they can be used at scale. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary ToS and privacy policies are online contracts that people often sign without fully understanding the implications. This paper looks into using Large Language Models (LLMs) to help users make informed decisions when signing these agreements. The authors create a dataset of questions about privacy policies from popular websites and test how well different LLMs can answer them correctly. They find that some open-source models do better than others, while a commercial model called ChatGPT4 does the best. However, all the models still have room to improve before they can be trusted to help people make good decisions. |
