Summary of Generative Models, Humans, Predictive Models: Who Is Worse at High-stakes Decision Making?, by Keri Mallari and Julius Adebayo and Kori Inkpen and Martin T. Wells and Albert Gordo and Sarah Tan
Generative Models, Humans, Predictive Models: Who Is Worse at High-Stakes Decision Making?
by Keri Mallari, Julius Adebayo, Kori Inkpen, Martin T. Wells, Albert Gordo, Sarah Tan
First submitted to arxiv on: 20 Oct 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- 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 research paper examines the suitability of large generative models (LMs) for high-stakes decision-making tasks, specifically recidivism prediction. The authors analyze three popular LMs, considering factors beyond accuracy, such as agreement with human predictions or existing predictive models. Experiments are conducted to evaluate how different information types influence LM decisions and test techniques designed to increase accuracy or mitigate bias. The findings suggest that current LMs are not suitable for these tasks due to their limitations in handling imperfect, noisy, and biased data. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large generative models (LMs) are being used to make important decisions, but are they really good at it? This paper looks at three popular LMs and how well they can predict whether someone will commit a crime again. The authors not only check the accuracy of these models but also see how well they agree with what humans or other machines already know. They even test different ways to give the models more information, like adding photos, to see how it affects their decisions. The results show that LMs are not yet ready for this kind of responsibility. |
