Geeta Gurnani, IBM Technology CTO, delivers insight on AI governance imperatives for responsible innovation

Why is AI governance gaining importance as AI adoption accelerates and what are the steps to be taken for good governance?

The progress made with AI, especially with developments in foundation models and generative AI, has the potential for tremendous benefits to businesses and society – but only if it can be trusted. In fact, most enterprises we to have spoken to have said that consumers are more likely to choose services from companies with transparent and ethical AI practices. Therefore, this highlights the importance of organizations having a robust governance mechanism in place if they want to build and use AI responsibly that users can trust.

There are three common threads that companies should look into when taking steps towards AI governance. First is the provenance of data because at the heart every AI model lies vast amounts of data. Businesses need to be able to determine whether the right data was used, where it originated, how it has evolved, and identify any discrepancies in data flows.

Second, AI models must be transparent and explainable, so we can reduce any harmful disinformation or deceptive content. These models require maintenance throughout the AI lifecycle to ensure users always know why and how AI decisions were made.

Finally, ensuring fairness and reducing bias is central to risk mitigation. This is especially true in highly regulated industries, such as financial services and healthcare, where there could be detrimental repercussions and the tolerance is very low for unethical, biased decisions based on incomplete or inaccurate data and models.

What are the challenges faced while tackling responsible AI?

Some of the key roadblocks to responsible AI include the inability to respond to the growing and changing AI regulations and the lack of proper AI governance processes, policies, and tools. Noncompliance with regulations and industry standards can cost organizations both time in supporting audits and millions in fines. Not having the right tools and processes can inadvertently introduce errors into AI algorithms and models.

To help businesses deal with these challenges, last year we released watsonx.governance, a toolkit for AI governance that allows business leaders to direct, manage and monitor their organization’s AI activities. It provides one integrated platform to help businesses manage their AI responsibly at enterprise scale and prepare for a broad range of AI regulations. 

How do we co-shape an ecosystem for responsible AI?

For technology that is as pervasive as AI, building trust in it is a challenge that requires global collaboration. Corporations, governments, academicians and industry bodies have critical roles to play. Its only by working together with collective responsibility and putting in place guardrails throughout the AI lifecycle, that we will be able to establish an effective ecosystem for responsible AI. 

IBM has worked with governments worldwide to advance smart AI regulation that protects people by focusing on the riskiest real-world uses of AI, not the underlying algorithms. We recently helped launch the AI Alliance, a diverse community of AI researchers, creators, and developers committed promoting broader, open, inclusive AI innovation. We also launched a Center of Excellence for Generative AI to help organizations of all sizes, anywhere in the world, embrace a future of effective, ethical AI for business.  

What are your views on the global regulatory frameworks in place?

There is a need for smart AI regulation that provides guardrails for society while promoting innovation. Governments should focus on risk-based regulation, prioritise liability over licensing, and support open-source AI innovation, as well as calling for proactive corporate accountability to ensure that AI is explainable, transparent and fair.