How do we embrace AI with responsibility?

The dawn of artificial intelligence (AI) promised a future of innovation, productivity, and growth.?The momentum has been significant, but there?is apprehension?about ensuring responsible AI usage is visible.?We?are also faced?with its shadow – instances where AI has faltered, with unacceptable consequences for our society. Take, for example, the AI-driven recruitment tool exhibiting gender bias by disproportionately downgrading female candidates' resumes. The case highlighted the importance of transparent and unbiased AI augmentation in hiring. It led to the New York City Council passing a bill that required employers to conduct annual bias audits of AI tools used in hiring decisions.?

Moreover, Generative AI raised the stakes last year due to its creative nature. It can produce inaccurate and irrelevant outputs.

The Problem

Despite AI's potential to contribute to global GDP by $15.7 trillion, more than the current output of China and India combined, its risks are formidable. Governments, the private sector, universities, and other interested stakeholders are increasingly concerned about it and are showing urgency in working together to enhance confidence in using AI. ?The AI principles, adopted by the Office for Economic Co-operation and Development (OECD) and Group of 20 (G20), have served as a benchmark for hundreds of governments and organizations shaping policies on the use of AI. The EU AI Act and the Canada Data and AI Act are putting stringent regulations and financial penalties for non-compliance. The US and UK also have rolled out standards, risk management frameworks, and guidance for the safe development of AI.????

However, the overall progress of establishing the guardrails and taking them to implementation is slower than the fast evolution of this profound technology.

According to?Accenture's research, only?2%?of companies have fully operationalized?responsible AI?across their organizations.?

Many technology companies have released AI principles, but relatively few have made significant adjustments to their operations, as a result - Stanford HAI

We are at a tipping point. Companies need to start turning their intentions into action to avoid missing out on being part of the AI value journey. With every passing day, they are acquiring AI debt.

How do we pivot from Policy to Practice?

Based on years of experience implementing Responsible AI, here is a ten-point approach for implementing Responsible AI. This approach enables the creation and deployment of AI systems that are safe, secure, and trustworthy. TRUST is indeed a cornerstone.

Leadership

• Monitor various regulations, acts, and how the industry evolves with time – and participate in industry forums to learn and share knowledge.
• Create your policy based on your?companies'?core values, evolving regulations, and guidelines.
• Rely on a multi-disciplinary team. It's essential to form a diverse governance board and working group that's cross-functional, involving representation from various functions, including risk, legal, data science, HR, and others.

Risk-based framework

• Create a risk-based framework and a process to filter, screen, and assess risk. Take input from academic institutions, consulting partners & government organizations that are further in the journey, like NIST (National Institute of Standards and Technology)
• Find out how much AI you already use and how responsible that is. Inventory of existing systems and mitigation plan for high-risk systems. Consider this a journey into the unknown, and be agile and nimble in charting your course.

Responsible AI Capability

• Invest in tools, Responsible AI methods & guidebooks, like AIF360, an open-source library containing techniques to detect and mitigate bias in machine learning models.
• Train everyone, the builders and the consumers of AI.
• Build a Center of Excellence (CoE). A core group of experts and evangelists skilled in supporting, educating, and reviewing AI systems. Incorporate design principles that involve friction with AI, i.e., keeping humans in the loop.

Compliance and Measurement

• Set up a regular reporting & compliance system to measure progress and establish trust with your people, customers, and regulators.
• Form an audit capability to generate confidence. Explore the involvement of external agencies and standard bodies.?

In conclusion, it will be a long and exciting journey, and companies should embrace the challenges, learn, and evolve rather than wait for complete solutions.?

"It is not that I'm so smart. But I stay with the questions much longer." ―?Albert?Einstein