CFOs Must Act Now to Capitalize on Rapid AI Advancements

(SeaPRwire) –   Good morning. Artificial intelligence is advancing rapidly, yet numerous organizations have not yet determined who is responsible for converting this progress into tangible business results.

During ’s Modern CFO dinner in San Francisco last Thursday, sponsored by Deloitte and ServiceNow, Melissa Valentine, a senior fellow at the Stanford Institute for Human-Centered Artificial Intelligence, told CFOs that their opportunity to lead the creation of AI value is closing quickly.

Valentine referenced a recent Harvard Business Review piece authored by the founders of the Return on AI Institute, highlighting survey data that emphasizes this opportunity. The survey revealed that only 2% of C-suite executives indicated CFOs were tasked with deriving value from AI. However, in instances where CFOs held this responsibility, 76% reported significant value generation, outperforming other departments. Laks Srinivasan, a co-author of the report, noted that finance leaders are uniquely equipped to define, assess, fund, and measure AI projects, and then implement that structure throughout the organization.

Valentine, who is also a tenured associate professor of management science and engineering at Stanford’s School of Engineering, addressed the finance executives, stating that CFOs have a strategic chance to spearhead AI initiatives if they are prepared to quantify the benefits and accept accountability. She contended that generative AI is transitioning from an experimental stage into systematic measurement—a domain familiar to CFOs. While strict accountability was premature two years ago, she noted that it is now indispensable.

Regarding the implementation of guardrails, Valentine highlighted a recent event where Anthropic accidentally leaked internal source code for its Claude coding tool, providing a rare public look at how leading AI labs secure their models. She emphasized the idea of “harness engineering”—the infrastructure built around models to ensure safety and usability, which includes secondary AI systems intended to monitor primary models. Her recommendation to CFOs was to examine this architecture, as leaders need to verify that the ecosystem surrounding a model is sufficiently robust to be governed, monitored, and trusted at an enterprise level.

This instance supported a wider theme in Valentine’s address: the criteria for safe, production-grade AI differ significantly from those for routine employee experimentation. She delineated a clear contrast between two distinct types of AI transformation. The first starts at the frontline, with staff utilizing tools like Gemini or NotebookLM to find practical uses via trial and error. The second is driven from the top down, where production-grade scenarios require strong data infrastructure, engineering precision, and governance. While both are important, each necessitates a separate operating model.

The key lesson for financial executives is that accountability for AI is evolving into a core competency for CFOs. As AI shifts from a novelty to an operational necessity, the leaders who enforce discipline will be the most successful in securing its value.

Sheryl Estrada
sheryl.estrada@.co