Deployment of Responsible Artificial Intelligence

Artificial intelligence (AI) is already ubiquitous – powering your phones, watches, cars, refrigerators, and washing machines. The question is whether these have been deployed responsibly, with time and effort taken to verify that they behave as predicted, even at the corner cases. AI, unlike human intelligence, is merely a statistical inference machine that chooses an action based on its calculations of the most probable outcome. With abundant data, AI starts to imitate human behavior, but needless to say; it requires careful management to prevent unintended consequences. Not only for brand reputation but the well-being of individuals, employees, and society. As with any advanced technology, for example, stem-cell engineering, the philosophical considerations of ethics have started to dominate the AI discussions — how independent is too independent for machines? However, the debate is whether humans can ever imbue AI technology with a universally "correct" ethical judgment when our values vary widely by culture.

Numerous instances of AI bias, discrimination, and privacy violations have already surfaced, leaving industry leaders rightly concerned about how to maintain ethical standards as they deploy computationally more intensive AI systems. However, this is not to advocate a moratorium on such AI technologies or whether we can, even if we want to. Instead, we need to make as much effort to build fairness, interpretability, privacy, and security into these systems as we race to build more complex computational machines. It is important to note that ethical decision-making is not just another form of technical problem-solving. Designers and developers hold a share of the collective influence. As such, they must understand the ethical considerations of their work.

Human judgment plays a vital role in the creation of an AI system. We write algorithms, make decisions about their uses, and define the success and failure of the system. At each stage, every individual involved in creating AI is accountable for the system's impact on the world. That's why we need to ensure that the AI system is thoroughly tested, transparent and explainable to clients and employees and has all the ethical considerations in place. A regulatory framework and policies can help ensure that AI technologies benefit humanity. If we can responsibly build AI systems, then the sky is the limit for the technology.

At NTT DATA, we are wary of AI's impact on society and people's activities and decisions. Hence, we've aligned ourselves with the ethical guidelines for trustworthy AI, leveraging our responsible AI framework to support operationalization across the data and AI lifecycle. From data and algorithms to social impact, our vision is to support organizations in defining their analytical strategy in alignment with responsible AI principles. Our AI guidelines attest to this.

Realizing the well-being and sustainability of society

We make sure these AI solutions positively impact society and follow sustainability guidelines.

Co-creating new values by AI

We promote innovation by interacting and co-creating with AI stakeholders.

Fair, reliable, and explainable AI

Users of AI must realize its fairness and trustworthiness.

Data protection

We believe privacy and security are critical to providing AI services. Our AI leverages the principles of security and design to protect personal information.

Contribution to the dissemination of Sound AI

We are working to support society in understanding AI to accelerate its acceptance and improve the literacy of AI users.

As a trusted global innovator, NTT DATA realizes that AI is not neutral unless we consciously design it to be so. As such, we have made it a practice to explore the full ethical, social and legal aspects of AI systems to avoid risks and unintended consequences that may arise from deploying AI technology in society.

Adapted from the 9th NASSCOM Annual Tech Conference on the topic: Deployment of Responsible Artificial Intelligence (August 23, 2022). You can watch a replay here.