AI The utilisation of Artificial Intelligence (AI) can lead to a range of beneficial outcomes for organisations, such as improved efficiency, accuracy and customer satisfaction. However, the misuse of AI has been highlighted by Amazon’s finding that their AI-based human resources algorithm was discriminating against female applicants for technical roles. Additionally, AI-driven mortgage financing has been observed to unfairly disadvantage Latinx and Black borrowers due to their perceived increased risk, and predictive software for predicting future criminals has been seen to show bias towards Black Americans.
The potential risks associated with using Artificial Intelligence (AI) that fails to consider diversity and justice can be highly detrimental. It is therefore essential that professionals in the business world take the necessary steps to ensure that technology remains unbiased and unprejudiced. Without proper oversight and monitoring of the technology, the potential biases currently present may become ingrained in the systems.
Despite the lack of a unified set of regulations, businesses must strive to ensure the ethical and transparent utilisation of Artificial Intelligence. Fortunately, a significant 80% of CEOs are prepared to take measures to enhance AI accountability. We recommend the following strategies to achieve this.
Create a Working Definition
It is essential to come to a consensus on what is meant by ‘ethical AI’ before continuing the discussion. A clear and succinct definition should be agreed upon. Several organizations have made public statements regarding their plans in this area. Microsoft, for example, have stated their aim is to “advance AI driven by ethical values that put people first”, listing justness, dependability, safety, privacy, security, openness, transparency, and responsibility.
Rob High, Chief Technology Officer at IBM Watson, has identified three key pillars of ethical AI in the video below. Companies, their customers and other interested parties must be able to trust that the AI is operating in an ethical manner. The public response to the Cambridge Analytica incident on Facebook serves as a reminder that any misuse of data will not be tolerated.
Transparency is the quality of being able to identify the source of data and ensuring robots are appropriately trained. It is essential for individuals to have control over how their data is used, which is the fundamental principle of privacy.
Promote Learning
Once the definition of ethical AI has been agreed upon, it is important to share it with key stakeholders such as business partners, workers and consumers. As Forbes states, “everyone throughout the business has to understand what AI is, how it can be used, and the ethical considerations associated with it.” Executives and other decision makers can use resources such as the World Economic Forum’s AI C-Suite Toolkit to explore complex topics such as how to create an AI-friendly culture within a business, and what expertise is necessary for business leaders to ensure successful AI implementation.
It is essential for the success of any education programme to appoint a leader from the governance team to construct the curriculum, to encourage staff to be involved in practical learning and to assess participants regularly to ensure they understand key concepts.
Install a System of Governance
Establish a team of AI professionals to develop and maintain ethical AI systems. Diversity in terms of race, gender, economic background and sexuality could assist in minimising certain problems related to prejudice in AI. Team members should encompass a variety of sectors, comprising of business owners, customers and policy makers. Prior to any other action, a discussion about ethical AI and related issues such as data privacy, prejudice and explainability (the capacity to explain the steps undertaken by an algorithm in detail) should take place.
The next steps involve analysing the potential risks posed by the company’s AI data and establishing systems, guidelines and checks to ensure AI practises are effectively monitored. As previously mentioned, there are limited regulations in place for ethical AI use. Organisations should, however, make it a priority to adhere to existing regulations. For example, the OECD’s standards for artificial intelligence outline how businesses can employ AI for the benefit of society. Establishing trustworthy AI is reliant upon international collaboration, which is one of the key components.
As the AI industry advances at a rapid pace, it is essential for organisations to develop a business and institutional AI ethics strategy that is in line with their original aims of fairness and transparency. In order to ensure compliance with any new regulations, regular reviews must be conducted to guarantee that these goals are still being met. Furthermore, as AI becomes more ubiquitous, businesses must invest in human resources and be prepared for the consequent changes in the labour market.
Consist with Current Methods
Any successful business undertaking must either align with an organisation’s existing value system or modify it to address the question, “What is the purpose of this?” Transparency, openness and a sense of collective responsibility are all desirable traits to have in place. All of these concepts are applicable to the ethical nature of artificial intelligence. It is time to evaluate whether there is an absence of shared values or an overarching structure for valuing work at the organisation.
Rather than treating ethics as something to consider at the end of product development, it is important to include it from the outset. Although it may not be easy or convenient, taking this approach is highly beneficial. Examples of the measures that can be taken include researching data collection, introducing new controls, setting up approval systems, and conducting regular reviews of processes and practices.
Embrace Openness
Companies committed to ethical AI should be transparent about their approaches. According to a recent article published by the World Economic Forum, “Organisations should be particularly clear about the data being used, how it is being applied and why.” There may be opportunities for businesses to be more open about their technology usage. For example, they can reveal that a chatbot is an automated system, rather than simulating a human.
One way to demonstrate transparency is to seek the opinion of experts in the field on the effectiveness of your methods and policies. This could include involvement in peer groups, communication with legislators and the production of informative materials such as white papers, blog posts and articles.
