KOHb - Getty Images
Capital One AI partnerships aim to build trust, grow talent
Capital One's relationships with the National Science Foundation and several universities seek to improve AI's responsiveness and create new metrics to measure its progress.
Capital One is expanding its ecosystem of AI research partners, seeking to pool resources to advance the technology, cultivate STEM talent and glimpse the future capabilities of AI applications.
The financial services firm last month launched a partnership with the National Science Foundation, part of a $100 million effort the NSF announced to boost AI research in the U.S. Capital One plans to invest $5 million over five years in the arrangement, which is the latest in a series of science and technology alliances for the company and its first with a federal research agency after previously teaming up with academic entities.
"It is a new direction in terms of directly participating with the government," said Prem Natarajan, Capital One's chief scientist and head of enterprise AI. "But it is the logical next step in the overall framework we have come to adopt."
The Capital One AI relationship fits into the NSF's broader funding strategy.
"Partnerships with the private sector are an important part of NSF's investment model," an agency spokesman said.
The spokesman said many NSF investments, especially those focused on AI and other emerging technologies, are forward-looking and aim to stay ahead of educational trends, user needs, industry demands and global competitiveness. Working with partners across all sectors sharpens the NSF's investment focus, providing "insight into real-world challenges" and keeping research in touch with practical applications, he added. Partnerships also provide access to research resources for AI workforce and talent development.
Natarajan also cited the importance of cross-sector alliances.
"I think the continued U.S. success in technology, in general but especially AI, requires a multisector partnership," he said. "No single organization has all the talent it needs to overcome the hardest AI challenges."
Building an AI alliance strategy
The NSF arrangement is part of the bank's AI alliance framework, which consists of strategic partnerships akin to its pact with the federal agency and more narrowly focused collaborations with research institutions.
The strategic relationships kicked off in March 2024, when Capital One partnered with Columbia University to launch the Center for AI and Responsible Financial Innovation. Capital One will invest $3 million to back the center, which aims to accelerate research, education and responsible AI use in financial services.
Capital One has since entered similar alliances with the University of Illinois Urbana-Champaign and the University of Southern California. In addition, the company has launched smaller, targeted academic relationships with Carnegie Mellon University; the University of Maryland, College Park; MIT; New York University; the University of North Carolina at Chapel Hill's MURGe-Lab; and the University of Virginia.
Seeking to 'systemically advance AI'
Capital One's NSF partnership stands out among its other relationships for the scope of the public-private initiative.
"With NSF, we are trying to systemically advance AI," Natarajan said. "It's literally everything from material sciences and economics and social sciences to computational sciences, physics, biology and chemistry -- the whole gamut of human exploration and discovery."
That perspective lends itself to envisioning future technological developments.
"One of the best ways to anticipate what might be coming is to partner with institutions like NSF," Natarajan said.
As for specific AI research areas, Natarajan said he finds the NSF's focus on AI assistants particularly exciting. The NSF AI Research Institute on Interaction for AI Assistants, which is led by Brown University and known as ARIA for short, aims to pursue next-generation technology for use in mental and behavioral health applications. Trust, empathy and personalization are critical in this field, but "current AI systems fall short," according to an NSF research abstract.
Natarajan said developing AI's ability to understand human emotions -- and respond appropriately -- supports the mental healthcare objective and broader applications.
"I think even though the overall [ARIA] framework is about mental and behavioral health, their core mission is developing AI that is trustworthy, capable and context aware," he said. "The way we want humans to gain more trust in AI is by increasing the performance of AI, but also by making AI more responsive to humans."
For example, an AI assistant with a higher emotional quotient might be able to initiate conversations with users who appear confused, prompting them to elaborate on their sentiments.
"Those kinds of interactions are particularly crucial in the mental and behavioral health aspect, but they are also important in many other applications of AI," Natarajan said.
He said the research on AI assistants could be applied to many industry-specific use cases. Examples might include language models underpinning generative AI assistants for Capital One employees or capabilities for customer-facing agentic AI offerings, he added.
Developing new AI metrics
Another potential outgrowth of the NSF collaboration is creating new metrics for assessing AI systems, a key consideration as organizations seek to measure the value of their investments. ARIA, for instance, seeks to develop evaluation metrics, and Natarajan anticipates other new AI research institutes being set up as part of the NSF's initiative will do the same. In addition to ARIA's focus on AI assistant technology, four more NSF AI Research Institutes will cover materials science, machine learning, drug discovery and the use of AI in STEM education. They increase the total number of research institutes doing AI work to 29.
"We are in this phase where there's a lot of work to be done even in designing the right metrics," Natarajan said. "What is the measurement that tells us that things are really progressing?"
Natarajan stated that the metrics to consider include methods for measuring trust, performance and an AI system's capability to interact "in the moment" with humans. He pointed out that while current metrics can evaluate an AI system's proficiency in solving math problems, there is a significant gap when it comes to well-established benchmarks for assessing how AI manages social interactions.
Natarajan said the arrival of effective metrics has historically fostered innovation:
"Computer science, machine learning [and] general science have advanced when we've had good metrics that we use to measure progress."
Along with metrics, the NSF AI Research Institutes are expected to produce data sets to test, measure and benchmark the results of their research and experimentation, he added.
Expanding tech talent pools
Richard Fairbank, Capital One's chairman and CEO, referred to the company as a "technology company that does banking" during its second-quarter earnings call in July.
Accordingly, the bank aims to attract skilled employees in fields such as AI. Natarajan said he believes Capital One's NSF partnership and university relationships help accomplish that. The best tech talent wants to join a "vibrant, inquisitive community" working to advance the state of the art, he said. Research alliances signal the existence of such an environment, he noted.
"These partnerships actually communicate to the talent that we want to hire," Natarajan said, emphasizing that this message is also critical for retaining existing talent.
Capital One's relationships also benefit its research partners, according to Natarajan. He said collaborating with the bank's scientists and engineers as they address financial industry challenges can help researchers direct their efforts toward finding impactful solutions to problems.
From a broader perspective, research partnerships contribute to strengthening STEM talent in the U.S., Natarajan said.
"We are trying to enrich the talent pool, especially the AI talent pool," he said. "We think this is the highest-leverage way to do it. The force multiplication of the partnership with NSF is just immense."
The NSF spokesman cited workforce development as a "central mission" for the agency's AI research institutes. Its approach is to fund a significant portion of the nation's foundational AI research and workforce initiatives, aligning with White House priorities to invest in AI education at all levels, he said.
The Trump administration's America's AI Action Plan, which states near-term policy goals, calls for federal agencies, including the Departments of Commerce, Education and Labor, as well as the NSF, to "prioritize AI skill development as a core objective of relevant education and workforce funding streams."
John Moore is a writer for Informa TechTarget covering the CIO role, economic trends and the IT services industry.
