Ethical Artificial Intelligence

Artificial intelligence (AI) products and services play a major role in shaping human decisions in a wide variety of areas of daily life, including financial decisions, e-commerce, healthcare choice, leisure preferences, and professional recommendations.

Major AI products and services include recommendation systems, conversational agents, sentiment analysis, and natural language processing algorithms. These have huge benefits for their users—enhancing the personalization of services, offering highly informed real-time support, ensuring consistent and concurrent engagement across a variety of channels, providing seamless self-services, and forecasting future preferences and decisions. Given the abundant usage of AI products and the growing reliance of both business and customers on their outputs, the trustworthiness and ethics of these products must be thoroughly scrutinized.

Assessing the trustworthiness of AI products is, however, challenging for several reasons. First, the invisible, “black-box” nature of AI operation typically prevents public scrutiny as AI systems are difficult to explain and understand. This opacity may result in biases and discriminatory outcomes that are often difficult to reverse or uncover.

Second, there is currently low public awareness and AI literacy, which may lead to public neglect and disregard of risks associated with AI. Further, the rapid scaled deployment of AI hinders proper scrutiny of AI products in advance of investment or deployment. Lastly, there regulation of AI products is lagging, and unable to provide adequate legal recourse.

The Ethical AI Project provides an innovative and practical methodology for assessing the trustworthiness of AI products, and implements it on a vast number of real-world AI products in a variety of areas, providing actionable findings and recommendations.