Artificial intelligence (AI) is entering more and more areas of our lives. Each new application raises pressing ethical issues. AI pessimists are worried about potential abuses. AI optimists are hopeful about potential benefits. Both are correct, in our view. To show why, we will survey some good news and the bad news about safety, privacy, justice, and responsibility in AI. Then we will propose ways to make AI more moral by building human morality into AI systems and AI companies. This talk summarizes the main points in our recent book with Vincent Conitzer.

Duke-Margolis is hosting a virtual public meeting entitled Understanding Bias and Fairness in AI-enabled Healthcare Software. This meeting will convene stakeholders across disciplines for conversations on ways in which bias can affect artificial intelligence (AI) in healthcare software and how to promote fairness in artificial intelligence software, including methods to test for and prevent bias throughout the development process. The meeting will also discuss whether AI itself can play a role in reducing existing biases in the health care setting. Among the expert speakers will be a diverse array of computer scientists, bioethicists, statisticians, anthropologists, and federal regulators.

The emotion of disgust can be partitioned into three distinct functional domains: pathogen disgust, sexual disgust, and moral disgust. Using adaptationist logic, we propose that disgust first evolved to mediate the avoidance of disease-causing agents, and then was co-opted as new selective pressures arose to guide decisions regarding mating behavior and, ultimately, other social interactions. We discuss findings from our fMRI study investigating the possible neural correlates of these proposed domains. Specifically, our study explored whether: (i) pathogen, sexual, and moral disgust activate common neural systems, and (ii) these three domains also entrain separate cognitive and behavioral systems specific to their respective evolved functions. Fifty male participants completed a set of surveys, and afterwards were scanned while performing a memory task that presented neutral statements, statements describing pathogen related acts (pathogen disgust), statements describing incestuous acts (sexual disgust), and

statements describing non-sexual socio-moral transgressions (moral disgust). Conjunction analyses indicated that pathogen, sexual, and moral disgust indeed activate common neural systems, and planned comparisons provided evidence that each functional disgust domain also has additional, unique neural correlates. Self-report data revealed distinct patterns of reactions to pathogen, sexual, and moral disgust providing additional support for our proposed model. We will discuss these and other related findings, as well as consider the implications our data have for the study of morality.