Summary of Evaluating the Propensity Of Generative Ai For Producing Harmful Disinformation During An Election Cycle, by Erik J Schlicht
Evaluating the Propensity of Generative AI for Producing Harmful Disinformation During an Election Cycle
by Erik J Schlicht
First submitted to arxiv on: 9 Nov 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
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 The study investigates the propensity of current generative AI models for producing harmful disinformation during an election cycle, evaluating the probability and expected harm of different models given adversarial prompts. The analysis reveals that Copilot and Gemini tied for the safest performance by realizing the lowest expected harm, while GPT-4o produced the greatest rates of harmful disinformation, resulting in much higher expected harm scores. The impact of disinformation category was also investigated, showing Gemini as the safest within the political category due to mitigation attempts made by developers during the election, and Copilot as the safest for topics related to health. Additionally, characteristics of adversarial roles were discovered that led to greater expected harm across all models. Finally, classification models were developed to predict disinformation production based on the conditions considered in this study. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study looks at how well current AI models can create fake information during an election cycle and what kind of harm it could cause. The researchers tested different AI models with tricky prompts and found that some models are better than others at avoiding creating bad information. They also discovered that certain types of fake news are more likely to be harmful than others. By understanding these things, the team developed tools that can predict when an AI model is most likely to create fake news and provide tips for how developers can make their models safer. |
Keywords
» Artificial intelligence » Classification » Gemini » Gpt » Probability
