A Voting-Based System for Ethical Decision Making
Authors: Ritesh Noothigattu, Snehalkumar Gaikwad, Edmond Awad, Sohan Dsouza, Iyad Rahwan, Pradeep Ravikumar, Ariel Procaccia
AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we implement and evaluate a system for ethical decision making in the autonomous vehicle domain, using preference data collected from 1.3 million people through the Moral Machine website. |
| Researcher Affiliation | Academia | Ritesh Noothigattu Machine Learning Dept. Carnegie Mellon University Snehalkumar Neil S. Gaikwad The Media Lab Massachusetts Institute of Technology Edmond Awad The Media Lab Massachusetts Institute of Technology Sohan Dsouza The Media Lab MIT Iyad Rahwan The Media Lab MIT Pradeep Ravikumar Machine Learning Dept. CMU Ariel D. Procaccia Computer Science Dept. CMU |
| Pseudocode | No | The paper describes the steps of its algorithm (Data collection, Learning, Summarization, Aggregation) in narrative text but does not include a clearly labeled "Pseudocode" or "Algorithm" block. |
| Open Source Code | No | The paper mentions the Moral Machine website as the source of data ("http://moralmachine.mit.edu") and refers to a "full version of the paper (Noothigattu et al. 2017)" for proofs and robustness results. It does not state that their code (for the algorithm described) is open-source or provide a link to it. |
| Open Datasets | Yes | Indeed, we use, for the first time, a unique dataset that consists of 18,254,285 pairwise comparisons between alternatives in the autonomous vehicle domain, obtained from 1,303,778 voters, through the website Moral Machine.2 |
| Dataset Splits | No | The paper mentions "100 test instances" for synthetic data and "3000 test instances" for Moral Machine data but does not explicitly specify training, validation, or test splits by percentages or sample counts for any dataset, nor does it reference standard pre-defined splits for specific datasets (other than mentioning the Moral Machine dataset itself, which is implicitly the source of data for all parts, but no specific splits within it are defined). |
| Hardware Specification | No | The paper does not mention any specific hardware used for running the experiments (e.g., GPU/CPU models, memory, cloud instances). |
| Software Dependencies | No | The paper describes the statistical and machine learning methods used (e.g., Maximum Likelihood Estimation, Gaussian process, Gumbel distribution, standard normal distribution CDF, Borda count, Copeland SCCs) but does not mention specific software libraries or their version numbers (e.g., "PyTorch 1.9", "Scikit-learn 0.24"). |
| Experiment Setup | No | The paper describes the setup for synthetic data generation (e.g., "βi from N(m, Id)", "d = 10", "N = 20 voters", "cardinality 5") but does not provide specific hyperparameters like learning rates, batch sizes, optimizers, or training epochs which are typical for ML models. It focuses on the model parameters itself (beta). |