Blog

Everyday Ethics for Artificial Intelligence: a guide for Designers and Developers

  • Francesca Rossi profile
    Francesca Rossi
    4 October 2018 - updated 2 years ago
  • 1 comment
    Total votes: 0

By Francesca Rossi, Member of the High-Level Expert Group on Artificial Intelligence

AI designers and developers hold a vast share of the collective influence of AI. They create systems that will impact millions of people. Ethics must be embedded into the design and development process from the very beginning of AI creation and those ethics must be aligned with the values and ethical principles of a society or the community it affects.

For some time now, IBM’s design team and I have been discussing how to help designers put AI ethics into practice. Last month we published Everyday Ethics for Artificial Intelligence, a digestible and actionable guide for designers and developers, to help them think about several issues in their everyday job of designing and developing a new AI service.

We organized this guide around five main focus areas that align with IBM’s Principles for Trust and Transparency:

1. Accountability: AI designers and developers are responsible for considering AI design, development, decision processes, and outcomes.

2. Value Alignment: AI should be designed with consideration of the norms and values of the user group.

3. Explainability: AI should be designed for humans to easily perceive, detect, and understand its decision process.

4. User Data Rights: AI should be designed to protect user data and preserve the user’s power over access and uses.

5. Fairness: AI should be designed to minimize bias and promote inclusive representation.

Each focus area includes recommended actions, necessary considerations, and questions.

By adopting and practicing the five focus areas covered in this document, designers and developers of AI systems can become more ethically aware, mitigate biases within these systems, add explainability components and instill responsibility and accountability in those who work with AI.

Designers and developers are encouraged to seize opportunities to intentionally put these ideas into practice.

Everyday Ethics for Artificial Intelligence represents the beginning of a conversation: it is an ongoing, interdisciplinary effort. We are releasing the first iteration of the Everyday Ethics guide to allow all who read and use it to comment, critique and participate in all future iterations. You can reach us at edethics@us.ibm.com

Tags: 
Artificial Intelligenceai ethicsAI designAI development

Comments

Log in to post comments
Norbert JASTROCH's picture
Norbert JASTROCH,
05/10/2018 13:15

Contributions like this help to become aware of the multi-faceted issues that relate to applied AI. While the focus herein is on people ("designers"), we should recognize that there are more addressees. It is useful to take into view a threefold: people, organisations, and public administration. 

Given that there is no such thing like a universal ethic which could cover all aspects involved for once and ever, we should take a start - in all modesty - by claiming: 

- For people working on the application of AI related technology, that part of their education should be principles of ethics.

- For organisations (from science, business, administrative bodies) involved in research, development and roll-out of AI, that part of their responsibility must be to establish guiding rules, structures and capacities.

- For public administration, that they set the limits by means of regulation and legislation.

In earlier contributions on this platform I have tried to exemplify these aspects in one or the other case.

I feel myself in full agreement with Simon Rogerson, professor emeritus in Computer Ethics at the Centre for Computing & Social Responsibility, De Montfort University, who states:

"The appliance of ethical sensitivity – rather than compliance with an ethical checklist – must be the way forward." 

in: https://www.researchgate.net/publication/328051332_The_data_self_The_con...

Total votes: 0