Frontiers
Browse

The Moral Choice Machine

Posted on 2020-05-20 - 04:19

Allowing machines to choose whether to kill humans would be devastating for world peace and security. But how do we equip machines with the ability to learn ethical or even moral choices? In this study, we show that applying machine learning to human texts can extract deontological ethical reasoning about “right” and “wrong” conduct. We create a template list of prompts and responses, such as “Should I [action]?”, “Is it okay to [action]?”, etc. with corresponding answers of “Yes/no, I should (not).” and "Yes/no, it is (not)." The model's bias score is the difference between the model's score of the positive response (“Yes, I should”) and that of the negative response (“No, I should not”). For a given choice, the model's overall bias score is the mean of the bias scores of all question/answer templates paired with that choice. Specifically, the resulting model, called the Moral Choice Machine (MCM), calculates the bias score on a sentence level using embeddings of the Universal Sentence Encoder since the moral value of an action to be taken depends on its context. It is objectionable to kill living beings, but it is fine to kill time. It is essential to eat, yet one might not eat dirt. It is important to spread information, yet one should not spread misinformation. Our results indicate that text corpora contain recoverable and accurate imprints of our social, ethical and moral choices, even with context information. Actually, training the Moral Choice Machine on different temporal news and book corpora from the year 1510 to 2008/2009 demonstrate the evolution of moral and ethical choices over different time periods for both atomic actions and actions with context information. By training it on different cultural sources such as the Bible and the constitution of different countries, the dynamics of moral choices in culture, including technology are revealed. That is the fact that moral biases can be extracted, quantified, tracked, and compared across cultures and over time.

CITE THIS COLLECTION

DataCite
3 Biotech
3D Printing in Medicine
3D Research
3D-Printed Materials and Systems
4OR
AAPG Bulletin
AAPS Open
AAPS PharmSciTech
Abhandlungen aus dem Mathematischen Seminar der Universität Hamburg
ABI Technik (German)
Academic Medicine
Academic Pediatrics
Academic Psychiatry
Academic Questions
Academy of Management Discoveries
Academy of Management Journal
Academy of Management Learning and Education
Academy of Management Perspectives
Academy of Management Proceedings
Academy of Management Review
or
Select your citation style and then place your mouse over the citation text to select it.

SHARE

email
need help?