Beyond Principles Video Series
In this series, Steve Mills, BCG GAMMA's chief AI ethics officer, joins fellow experts and leaders to discuss how organizations can turn responsible AI principles into practice.
When done right, responsible AI isn’t simply a framework to mitigate risk. It’s a way to deliver transformative business impact. BCG works with companies to make responsible AI a reality.
As artificial intelligence plays a more central role in business and society, so does the need for ensuring its responsible use. The financial and reputational risks of AI failures are real, but they can be managed if organizations take the right steps.
Our philosophy is twofold: first, responsible AI should not be viewed strictly as a defensive maneuver—but also as a source of value. Second, the principles of responsible AI offer an important starting point, but they must be translated into action.
Companies do not need to choose between protecting customers with responsible AI and boosting the bottom line. Companies can both realize the transformative power of AI and amplify its benefits by developing AI in a responsible manner. Time and again, BCG research has shown that companies with a strong sense of purpose and a focus on total societal impact outperform their peers. Incorporating an organization’s authentic purpose and AI ethics into the core of AI will:
Companies that recognize the need for responsible AI usually begin by developing principles to guide their actions. But they sometimes struggle to move from high-level principles to tangible changes in the way they build AI solutions. We call this the responsible AI gap. Companies must bridge this gap if they want to unlock the true potential of responsible AI, and they can do so by taking BCG’s six simple steps.
When we work with clients to implement the six steps of responsible AI, our perspective is informed by the following beliefs:
Don’t just philosophize, operationalize. Many companies have developed responsible AI principles, but very few have changed how they operate. Bridging the gap between principles and actions is our focus.
Play offense, not defense. When companies focus strictly on risk mitigation, they miss the upside potential of responsible AI. Our holistic approach helps companies understand how to realize the sources of value from AI ethics.
Look at systems, not algorithms. Most frameworks around artificial intelligence and business ethics focus exclusively on algorithms, but problems can arise anywhere in the AI value chain—from data collection through decision making. Algorithms and data constitute only 30% of AI; the remaining 70% comprises business processes. Ignoring that 70% means ignoring a significant source of concerns about responsible AI. Our approach considers the entire AI system, end to end.
Start from a position of strength. We use our proprietary Responsible AI Organizational Maturity Assessment to help companies evaluate their organizations and analyze their readiness for responsible AI. We provide a heat map of strengths and weaknesses in such key areas as data and privacy, fairness and equity, and social impact.
In this series, Steve Mills, BCG GAMMA's chief AI ethics officer, joins fellow experts and leaders to discuss how organizations can turn responsible AI principles into practice.
As part of our broader commitment to responsible AI, BCG has developed a set of tools that help companies achieve their business goals and preemptively address the repercussions that AI can have on individuals and society. Companies can use these free tools to ensure their AI programs are, in fact, responsible.
AI can benefit society in many ways. But given the amount of energy that’s needed to support AI’s computing requirements, these benefits can come at a high environmental price. CodeCarbon is a lightweight software package that can be seamlessly integrated into a company’s Python code base. The software estimates the amount of carbon dioxide that will be produced by the cloud computing resources needed to execute the code. It then shows developers how they can lessen emissions by optimizing their code or by hosting their cloud infrastructure in regions that use renewable energy.
Many business users treat AI as a technology black box that produces positive business results—but does so in “mysterious” ways. FACET, which BCG designed around the leading Python package scikit-learn, helps human operators understand advanced machine learning models. Now able to “open the box,” data scientists and business users can make decisions that save money, maximize yield, retain customers—and ensure that the end result is, indeed, responsible. FACET can be applied to a broad range of use cases, from improving patient outcomes to optimizing complex supply chains.
In BCG's latest survey, 55% of the organizations are less advanced than they believe.
To earn the public’s support, government use of advanced analytics must include stakeholder input, proper controls, regular reviews, and contingency plans for lapses.
Dataset bias can lead to serious ethical problems in even the most well-meaning AI models. How can practical techniques, such those used as facial recognition, help reduce this potential harm?
Principles abound for socially responsible artificial intelligence. Here’s how to put them into action.
Our deep commitment to Responsible AI goes far beyond just principles. Here’s how we put our Responsible AI principles into practice.
Key performance indicators help organizations guide their AI projects to successful, responsible, and meaningful results. Here are five KPI categories to keep your eye on.
