AI governance is so crucial for the growth of modern organizations that tools are already arriving to manage it. We spoke to Guru Sethupathy, founder of FairNow, about what AI has in store for the market.
To find out, we spoke to AI experts across the world to discover the truth within the AI buzz. In the next part of this series, we spoke to Guru Sethupathy from FairNow.
To find out more about what the market is saying about AI, download our AI survey results:
Guru was an amateur chess player in high school when IBM's Deep Blue defeated Gary Kasparov. That seminal moment drove him to begin an AI journey that led him to enroll as a computer science major at Stanford.
Since then, Guru has conducted academic AI research, advised Fortune 100 companies on AI for McKinsey. Before FairNow, he led AI and governance as a senior executive at Capital One.
Guru is positive about the potential of AI technology to revolutionize the business world. However, he also understands the risks involved in leveraging the technology.
By ensuring fairness, transparency, and accountability, however, organizations can build protective guardrails and innovate responsibly. That's why everyone who's excited about AI should be thinking about governance.
To help, Guru started FairNow to build trust, increase adoption, and lower the risk of AI. He knows that governance can be a painful process, but the FairNow platform aims to simplify governance with automations, centralization, and intelligence.
Guru has unique expertise in AI governance, so we knew he would be the perfect person to talk to about it. First of all, let's hear his thoughts on why AI governance is so important.
"However, with potential comes risks. These risks encompass regulatory compliance, and critical business concerns like data privacy, security, bias, reliability, and performance. A good AI Governance program helps organizations realize the value potential of AI while managing and mitigating the risks."
"First, many organizations don’t understand what a good AI governance program entails. Many questions arise:
"This list can become very intimidating to organizations that are new to formal governance structures.
"However, even for the most informed stakeholders, AI governance can still be challenging to implement. AI technology often spans multiple business units and functions, making centralized risk management and mitigation (essential for consistency and visibility) an extensive coordination effort.
"Successfully managing AI governance on a day-to-day basis requires cross-organizational buy-in, ensuring all models are inventoried, and risk factors are monitored in alignment with the organization's standards."
"The growing importance of AI Governance will necessitate the following:
"In fact, a recent analysis found that the number of Fortune 500 companies citing AI as a potential risk has surged by over 473.5% over the last year, with more than 55% now flagging concerns. Good governance should help organizations get comfortable adopting technology and innovate in areas they might have previously avoided.
"One key takeaway is that good governance should be risk-weighted. This means that all AI applications should undergo risk assessments, with low- to medium-risk applications requiring only minimal governance.
"Equally important is the need for clarity and consistency in AI governance programs. By clearly establishing policies, procedures, controls, and human-in-the-loop accountability, organizations can build confidence in high-risk AI applications among regulators, customers, and employees.
"Governance of high-risk AI applications can be incredibly time-consuming if done manually and without proper organization. Disorganized or inefficient AI governance programs can hinder innovation, creating barriers rather than opportunities.
"This is where intelligent, centralized, automated governance technology can provide tremendous value."
"As a result, they have already had to implement model risk management or model governance programs, and have become leaders in this space. A key lesson is that SR-11-7 and model governance at financial institutions offer a good template for other organizations and sectors to learn the key pillars of good governance.
"That being said, even at these organizations, there are significant opportunities to innovate as governance is still often manual, complex, and time-consuming. Another important lesson for organizations looking to adopt new AI technologies, especially those who haven't yet begun exploring governance, is that it is far easier to implement AI with a solid governance foundation in place than to adopt AI governance strategies after the fact.
"In my experience, retro-actively applying governance principles often requires switching AI providers, rebuilding models, and making other costly adjustments. Organizations embarking on their governance journey for the first time should prioritize standardizing, simplifying, and automating governance using technology as much as possible to avoid these challenges."
"These stakeholders not only help shape the governance program but also play vital roles within the accountability framework. For example, data scientists might be responsible for inventorying AI applications and metadata, as well as creating documentation; business owners could handle approvals; and legal/risk/compliance can provide a second line of review.
"While these examples are illustrative, the specifics of accountability will vary by organization."
"For example, the technology team can firewall certain websites and restrict access to external applications, requiring approval before use. Finally, the technology function should maintain log files of interactions with external AI applications and check that those interactions (sharing personal data or company secrets) do not run afoul of internal policies."
"The governance solution should be a single point where all stakeholders can come together to manage various aspects of governance, including inventorying and risk management, compliance tracking and evidencing, workflow, and documentation. If necessary, the platform should integrate with relevant tools and technologies so the customer doesn’t have to duplicate work or information.
"Companies that implement governance through a hodgepodge of existing tools end up creating a lot of manual work, and things can fall through the cracks. That is why a centralized, fit-for-purpose governance solution is highly valuable.
"It should have embedded logic and capabilities that enable it to pro-actively share insights, notifications, and triggers. That way, stakeholders will not need to rely on their own memory to perform governance actions.
"Today, governance is extremely manual and time-consuming. However, many aspects—such as testing, documentation review, and control verification— can and should be automated, saving substantial time and costs while ensuring precision and compliance.
"The FairNow platform (hyperlink:https://fairnow.ai/) is centralized, integrated, intelligent, and leverages automation, bringing the power of technology to governance."
To find out more about what the market is saying about artificial intelligence (AI) governance, download our AI survey results: