Project Asimov (Working Draft)

A guide to pressing ethical matters in artificial intelligence technologies.

Contents

A research proposal.

A literature review.

A sample demo.

General notes.

Methodology notes.

Framework notes.

A draft outline for Part I.

Part I: Ethics for AI Practitioners

With artificial intelligence (AI) permeating into military , law enforcement , financial , healthcare and transport applications, what was once the fancy of science fiction authors has been made real enough to both save and endanger livelihoods and lives. In light of what is at stake, there is a growing need and a moral imperative for AI researchers, engineers and practitioners to understand the ethical issues associated with the technology. This project aims to illustrate these issues through real-world case studies and design a set of ethical best practices for practitioners implementing AI solutions. Specifically, the project will focus on the following issues - algorithmic bias , explainability and privacy . We will define these concepts in an online guide containing references to real-world cases and interactive examples. The project will also look at how ethical considerations can be factored into the design and implementation of AI solutions, including references to important mitigations that are currently available .

Part II: AI & Ethics for Non-AI-Practitioners

Technological advancements have generally led to improvements in peoples' lives. Simultaneously, these advancements are accompanied by increasing complexity and opacity that belie understanding. In particular, the layperson's understanding of recent artificial intelligence (AI) technologies often conflates personal experiences with science fiction and media hype . This project first aims to provide a primer to AI concepts for the general public and act as a reference guide to common AI terminologies found in the media, marketing materials and solution proposals. The project also aims to raise awareness about possible problems with AI technologies - less about Terminator and HAL9000, more about fairness and privacy . This culminates in an online website that serves as a playbook for surviving in an AI-intensive society, including explorable explanations and links to real-world examples. Finally, the materials should help policymakers, commercial stakeholders and the general public pose critical questions about social and ethical aspects of AI and how these might be factored into the design of such solutions.