Responsible AI at Kyndryl
As a trusted technology services company running mission-critical infrastructure and applications, Kyndryl has a unique perspective on Artificial Intelligence’s (AI) practical implications. With a client-centric, vendor-agnostic approach, Kyndryl dedicates resources to recognize and understand the potential business benefits of AI weighed against applicable regulatory, ethical requirements and associated legal risks of AI.
Our ongoing investment in AI best practices, thoughtful selection of AI tools and platforms, secure training of AI models with the right machine learning algorithms and secure technical guardrails, demonstrates our commitment to staying at the forefront of technological innovation. We recognize that AI is “more than algorithms and models” and are committed to upleveling and strengthening the skills of our employees who enable and develop AI solutions. You can have the best engine in the world, but with no wheels or steering, the vehicle is worthless.
As the world’s largest IT infrastructure services provider, we use our data expertise and solutions to help our enterprise customers to implement and derive benefits from AI at scale to achieve specific business outcomes. In doing so, we are committed to adopting a risk-based approach to AI deployment and development by identifying, evaluating, and mitigating risks across AI-Applications lifecycle.
We believe the AI system is only as strong as the data used to train the solution. Your data quality and more importantly, your data strategy is critical to the insights you will gather from your AI system. For it to be used to its full potential, appropriate data is key.
Our Responsible AI approach at Kyndryl entails adhering to the following fundamental commitments:
Transparency and Explainability: AI solutions will be designed to provide visibility into the decision-making processes underlying the models, as well as the nature, source and quality of the data sets used to train the AI models. The nature and function of AI solutions should also be explainable to establish user trust.
Detect and Reduce Bias: We require that when using or developing AI solutions, we consider, measure and take steps to reduce potential bias to promote fairness and minimize bias in AI design and usage.
Safe and Secure: AI solutions must incorporate security by design principles and evaluate and monitor performance to protect against harm to individuals, groups of individuals, and environmental or reputational damage to our enterprise and our customers.
Accountability: We have implemented a governance program to evaluate and oversee AI usage within our enterprise and with our partners and customers to contribute to a trustworthy and accountable AI ecosystem that respects human rights, the rule of law and democratic principles.
Education and training: We commit to thorough training and clear guidance for those utilizing AI technologies, in an effort to ensure adherence to the above-noted principles and promote responsible innovation.
It is through adherence to these fundamentals that Kyndryl will foster the right safeguards and processes to ensure responsible use of AI internally and with our customers.