1. To what extent is bias a major problem in today’s AI systems? What effect is that bias having?

Although often unintentional, algorithmic bias often comes from training data due to historical inequality of human experts, when the data is not representative of the actual population, or from improper selection of a combination of features. The problem of bias can be prevented by the operationalization of principles such as fairness and justice. Analyzing the training data with these principles within a process framework can prevent these unwanted consequences.

2. Companies need to commit to “creating and implementing AI responsibly and ethically.” How is this process evolving?

The recent rise of Artificial Intelligence technologies has sparked a proliferation of policies and practices intended to ensure the ethical development of Artificial Intelligence. However the variety of different ethical issues and potential outcomes of AI have led to an extensive yet somewhat ill-defined discourse surrounding the implementation of ethical guidelines. This discourse has left AI technologists with a lack of clarity over ethics and its relation to AI technologies. To counter this phenomenon, technologists will need clarity to the issues surrounding the development, implementation and use of AI, by categorising these issues as outcomes and linking them to specific ethical principles that can be operationalized.

3. What are steps that companies can take to improve the ethical foundation in their AI systems?

First companies will need to have a clear understanding of the ethical principles that need to be upheld, then it will be necessary for them to have a systematic way to translate theory into practice. At Aiforgood Asia we call this “ethics as a service”. This is the process of operationalizing ethics into all the different phases of development and implementation from how the technologies are designed, where they are deployed, who they are sold to; and how they are applied. Companies can start by establishing an independent multi-disciplinary ethics board to oversee the implementation and distribute responsibility throughout all levels of the organization and then ensure the enforcement of selected processes and tools.

4. The future: If we look several years ahead, what can we expect to see for the future of AI and ethics?

As the public becomes more increasingly aware of the potential negative outcomes that can arise when ethical principles are not implemented in AI design, we will see a variety of Companies respond. Some will create ethics boards, and hire ethics and compliance officers. Unfortunately, most of these efforts will not be effective, as industries and companies try to get ahead of policy governance. This flood of ethics “Bluewashing” are essentially public relations exercises that serve to distract consumers from the ethical damage the companies are causing. An example of this is when academia and academic research on AI ethics are used to lend credibility to the tech industry, which lobbies hard to avoid restrictive legal regulations. I worry that these tactics will water-down the effectiveness of implementing ethics and confuse the public if people are not more aware of this practice.

– Jesse Arlen Smith – President, Aiforgood Asia