Summary of Llms Left, Right, and Center: Assessing Gpt’s Capabilities to Label Political Bias From Web Domains, by Raphael Hernandes and Giulio Corsi
LLMs left, right, and center: Assessing GPT’s capabilities to label political bias from web domains
by Raphael Hernandes, Giulio Corsi
First submitted to arxiv on: 19 Jul 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 research investigates whether OpenAI’s GPT-4, a state-of-the-art large language model, can accurately classify the political bias of news sources based solely on their URLs. The study compares GPT-4’s classifications against Media Bias/Fact Check (MBFC) ratings, controlling for website popularity using Open PageRank scores. Findings reveal a high correlation between GPT-4’s and MBFC’s ratings, indicating the model’s potential reliability. However, GPT-4 abstains from classifying approximately two-thirds of the dataset, tending to avoid rating unpopular websites. The LLM also tends to avoid classifying centrist sources, resulting in more polarized outputs. While this paper suggests that GPT-4 can be a scalable tool for political bias classification, its use should be as a complement to human judgment to mitigate biases. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study looks at if a computer program called GPT-4 can correctly figure out the political leaning of news websites just by looking at their website addresses. The researchers compared what the computer program said with how humans rated those same news sources, and found that they were very similar. However, the computer program didn’t want to say which way it leaned for about two-thirds of the news sites it looked at. It also tended to avoid rating news websites that weren’t super popular or ones that were pretty middle-of-the-road in their political views. This study says that while GPT-4 is a useful tool, people should still use their own judgment when trying to figure out the political leaning of news sources. |
Keywords
» Artificial intelligence » Classification » Gpt » Large language model
