加入 getAbstract 阅读摘要

The Ethical Algorithm

加入 getAbstract 阅读摘要

The Ethical Algorithm

The Science of Socially Aware Algorithm Design

Oxford UP,

15 分钟阅读
7 个要点速记
音频和文本

看看什么内容？

Can humans program computers to behave ethically? In some cases, yes.

自动生成的音频
  • 0.5×
  • 0.75×
  • 1×
  • 1.25×
  • 1.5×
  • 1.75×
1×
自动生成的音频

Editorial Rating

10

Qualities

  • Scientific
  • Eye Opening
  • Visionary

Recommendation

Algorithms decide what ads people see online and whether consumers get loans. These computer-executed recipes acquire more and more power over humans’ lives. Do they wield it fairly? Can humans design algorithms to behave ethically? The message of Michael Kearns and Aaron Roth, computer scientists at the University of Pennsylvania, is yes – sometimes. Kearns and Roth explain when humans can expect algorithms to behave ethically, and in what cases computer science still struggles to accomplish this. They detail the conundrums and solutions in accessible language, even if it can be hard to follow along. This book will give you a good sense of when you can expect algorithms to behave ethically, and in what cases computer science still has its work cut out.

Summary

Algorithms raise questions of privacy, fairness, safety, transparency, accountability and even morality.

Companies and governments accumulate vast amounts of data about us. Computers collect and analyze these according to certain algorithms. This leads to valuable insights and helps provide services. But it can also divulge private information. And an algorithm may take or propose decisions about people that they would consider unfair or immoral.

Some approaches use algorithms that process data according to step-by-step instructions. Others use machine learning which allows the system to evolve. Such systems build internal models that are hard to understand for humans. And while processing data according to their rules, these systems may even arrive at solutions that offend social norms.

Regulations can help to ensure that systems are used according to their purpose. And when developers design algorithms for a purpose, they’re ideally led by the FATE principle: fairness, accuracy, transparency and ethics. However, a successful implementation of FATE can come with a price – the resulting algorithms can be less precise or efficient as compared to a design ...

About the Authors

Michael Kearns, PhD, is a professor of computer and information science at the University of Pennsylvania. He has worked in machine learning and artificial intelligence research at AT&T Bell Labs. Aaron Roth, PhD, is an associate professor of computer and information science at the University of Pennsylvania.

Comment on this summary