Credit: Anton Vierietin / Shutterstock
Across industry sectors, there’s growing excitement — and apprehension — about how roles and responsibilities will be shaped by artificial intelligence.
On the one hand, there is enormous potential for AI platforms and tools to help improve individual and enterprise performance, deliver cost savings, increase productivity, foster more intelligent and actionable insights, and improve customer interactions through the delivery of personalized services.
On the other hand, there is unease about AI’s possible impact on skills and jobs. There are also concerns that the technology exposes organizations to unprecedented business risks, such as data privacy and security breaches, unintended bias and discrimination, and the propagation of false information.
With so much at stake, organizations need to strike a balance between the race to innovate and a more moderate strategy for redesigning work patterns for AI.
“Generative AI shouldn’t go hand in hand with a gold rush mentality,” said Prem Natarajan, chief scientist and head of enterprise AI at Capital One, speaking on a panel at the 2024 MIT Digital Technology and Strategy Conference. “If this technology is truly enduring, it’s important to take your time … and figure out how to do it thoughtfully and responsibly. You have to build the scaffolding to bring everybody along.”
The panel also included executives from ING and Google and was moderated by MIT Industrial Performance Center executive director Ben Armstrong.
6 elements of a successful strategy
The panelists advocated for a measured approach to AI that considers organizational readiness while putting systems and processes in place to mitigate serious business risks and ensure success.
While there is no single blueprint, a comprehensive strategy for effectively redesigning work for AI should include the following, the experts advised:
Modernizing the data and technology infrastructure. AI is fed by massive amounts of data, which in turn requires significant amounts of scalable computing and storage horsepower. Before getting started, organizations need to take stock of their infrastructure and identify limitations. To maximize AI workflows, firms should modernize with a cloud-native approach and streamline their technology stacks so data and services are no longer fragmented across platforms and core systems.
“AI brings to life the value that is resident in your data,” Natarajan said. “If you don’t have your data act together, that’s the first thing to do.”
Upskilling and educating talent. Unlike with other initiatives, wholesale reliance on systems integrators to implement and maintain AI isn’t a sustainable plan. Companies need to assess the knowledge levels of internal talent and provide learning systems and training resources to help employees get up to speed, particularly on critical skills like engineering generative AI prompts. To help upskill its employee base, Google has created a short-form course called AI Essentials that was designed to cover the basics, according to Lisa Gevelber, chief marketing officer of Google in the Americas.
“These tools are amazing when you know how to use them,” said Gevelber, founder of Grow With Google, the company’s initiative to create economic opportunity for all. “But it is garbage in, garbage out.” To promote individual productivity with AI, Gevelber recommends the “four C’s”: clarity on AI parameters and guardrails, critical metrics for evaluating success, community for sharing best practices, and credentials for rewarding self-paced learning.
Embracing an ecosystem approach. Don’t work in isolation; instead, collaborate with a broad set of technology partners, advised Marco Eijsackers, global head of the CIO office for European bank ING. For example, ING has brought in partners such as Google, Microsoft, Adobe, and Spotify to leverage data and specific capabilities to simplify the developer experience or elevate customer interactions using AI. “It only works if the whole system works,” Eijsackers said. “Otherwise, it’s just another tool.”
Taking a test-and-learn approach. The use cases that most fully leverage AI’s capabilities can also be the most complex. Target smaller opportunities to get people comfortable, solicit feedback on early designs, and iterate accordingly, the executives agreed.
Keeping humans in the loop. Mitigating risk is extremely important, especially in highly regulated industries like finance. At Capital One, that means maintaining human oversight of AI processes, whether related to customer interactions or software development.
“Pretty much everything we do will be with a human in the loop for the foreseeable future, until we have the telemetry and observations that tell us AI is working well enough to automate more,” Natarajan said.
Thinking beyond the hype. Rather than operating from a sense of urgency, take a values-based approach, the panelists advised. Calibrate development and deployment cadences to specific business opportunities and the overall readiness of the company to embrace new ways of working. “People do want to change and embrace AI, but it’s not the holy grail,” Eijsackers said. “You have to have your act together.”
