Trust and governance: Navigating AI ethics and communication

The podcast emphasizes the critical role of trust and governance in navigating AI ethics, particularly in enterprise settings. Kelly Combs from KPMG discusses her experience in AI governance since the launch of KPMG's framework in 2018, focusing on building AI capabilities and identifying functional use cases. Nathaniel highlights the necessity of operationalizing AI for companies, stressing practical aspects of AI transformation.

Companies currently face uncertainty regarding AI regulatory frameworks, especially in the US, where no federal legislation exists. Some states have begun implementing regulations emphasizing auditability, transparency, and bias considerations in high-risk scenarios. The EU AI Act is noted as a conservative framework outlining responsibilities for high-risk AI use cases. Organizations need to design governance frameworks that align with their values while adapting to evolving regulations, with key components including AI policy development, clear roles and responsibilities, stakeholder engagement, and technical controls.

Governance frameworks must differentiate between internal and external AI use, with internal applications focusing on productivity in lower-risk areas and external applications requiring stricter governance due to higher risks. The approach to governance should be proportional to risk levels, ensuring foundational expectations are established for all AI use. The podcast also addresses the importance of risk management in relation to data and external impacts, advocating for resource allocation based on exposure levels.

Key stakeholders, including CIOs and heads of data science, are identified as drivers of AI capabilities, aiming to empower employees while ensuring responsible AI use. Education and upskilling are emphasized as essential components of governance, with KPMG developing tailored digital and data literacy pathways. The phenomenon of "secret cyborgs," where employees use generative AI tools without reporting, highlights a gap between available tools and workplace offerings. Governance is positioned as a solution to these challenges, rather than a hindrance.

There is a misconception that rules limit abilities, but organizations with clear expectations can encourage experimentation and effective use of AI. A shift in mindset is necessary to view AI positively, as it will be integral wherever people and technology intersect. Educating employees about the omnipresence of AI is crucial for fostering acceptance and innovation. Governance should be seen as an enabler, providing standardized tools for experimentation and creating a safe environment for innovation.

Establishing a clear strategy for AI is vital for bridging the trust gap between employees and leadership. Organizations should communicate their strategic roadmap and address employee concerns about AI's impact on jobs, emphasizing that AI will augment human work rather than replace it. Familiarity with AI tools is crucial for employees to remain relevant in the changing job landscape.

AI affects all functions within an organization, necessitating a cross-functional strategy group that includes technologists and business sponsors. Governance must involve risk management across legal, data, and HR functions, with clear roles defined for managing AI-related risks throughout its lifecycle. The strategic group is responsible for decisions on technology vendors, deployment speed, and effectiveness measurement, ensuring that strategy and risk partners are connected for informed decision-making.

Governance is an ongoing, iterative process that must evolve with advancements in technology, particularly with generative AI. Organizations need to customize their governance structures to align with their specific risk appetites, values, and AI strategies. Mature organizations focus on identifying high-risk areas, federating governance frameworks into practical tools, and establishing frameworks to measure fairness, safety, and security in AI. As AI continues to integrate into various contexts, governance will need to adapt to emerging risks and foster a culture of learning and experimentation.