Summary of Prompt and Prejudice, by Lorenzo Berlincioni et al.
Prompt and Prejudice
by Lorenzo Berlincioni, Luca Cultrera, Federico Becattini, Marco Bertini, Alberto Del Bimbo
First submitted to arxiv on: 7 Aug 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
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 examines the effect of using first names in Large Language Models (LLMs) and Vision Language Models (VLMs), focusing on ethical decision-making tasks. The proposed approach appends first names to annotated text scenarios, revealing demographic biases in model outputs. A curated list of over 300 diverse names is tested across thousands of moral scenarios. Popular LLMs/VLMs are audited using social sciences methodologies to emphasize the importance of recognizing and mitigating biases in AI systems. Additionally, a novel benchmark, the Practical Scenarios Benchmark (PSB), assesses bias presence in everyday decision-making scenarios, highlighting risks and biases that may arise in practical applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study looks at how using first names affects language models’ decisions. Researchers added diverse names to text scenarios and tested them with popular models. They found that these models show biases towards certain demographics. The team proposes a new way to test AI systems for bias and introduces a benchmark to help compare model behaviors. |
