Summary of Language Models Trained to Do Arithmetic Predict Human Risky and Intertemporal Choice, by Jian-qiao Zhu et al.
Language Models Trained to do Arithmetic Predict Human Risky and Intertemporal Choice
by Jian-Qiao Zhu, Haijiang Yan, Thomas L. Griffiths
First submitted to arxiv on: 29 May 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computation and Language (cs.CL); General Economics (econ.GN)
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 proposed approach uses computationally equivalent tasks to enhance the utility of Large Language Models (LLMs) as cognitive models. This involves leveraging tasks that both an LLM and a rational agent need to master for solving a cognitive problem, such as decision-making. The authors demonstrate that pretraining LLMs on ecologically valid arithmetic datasets leads to better predictions of human behavior than traditional cognitive models. This suggests that LLMs can be used as cognitive models if carefully investigated via ablation studies of the pretraining data. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large Language Models (LLMs) are similar to humans in how they behave, which has led researchers to consider using them as models of human cognition. However, there are some challenges to overcome before LLMs can be used this way. In this paper, researchers propose a new approach that helps LLMs become better cognitive models. They do this by looking at tasks that both an LLM and a person need to accomplish in order to solve a problem. The authors then test their approach on decision-making tasks like making risky or long-term choices. By training the LLM on math problems, they show that it can predict human behavior well. |
Keywords
» Artificial intelligence » Pretraining
