PwC and NBCUniversal: Bold and Human AI Strategies

Humans at the Center of AI Initiatives

In the current context of deploying AI agents, it is imperative not to simply hand over the keys to these technologies but to ensure that every initiative remains under human direction. Scott Likens, Global Chief AI Engineer at PwC, emphasized the importance of this approach by stating, "Stop being the human in the loop. The human is the loop." This philosophy involves focusing on the needs of the end user and working backward to determine the most appropriate AI tool. Lasherelle Morgan, Senior Vice President of AI Innovation and Acceleration at NBCUniversal, recommends starting with easily repeatable processes and data, beginning by identifying pain points. She advises, "Don't just introduce an AI tool. Ask yourself: 'What is your problem?' 'What do you spend five hours of your day on?'"

Experimentation as a Lever for Innovation

Large-scale experimentation is essential to determine where AI can truly deliver benefits. Scott Likens from PwC explained that the company deploys AI-based experiments in cycles of one or five days. The goal is to encourage leaders to think about the possibilities offered by AI, rather than focusing solely on cost savings of 2% or 3%. "All this talk about tokens started a few months ago, and now there is suddenly a focus on costs with AI," he noted. To succeed, this requires a radical shift in mindset, especially among middle managers who are not accustomed to such short work cycles. Likens clarified that while leaders and new employees may be open to these changes, the real challenge lies in this "frozen middle" of experts and managers reluctant to change their habits.

Cleaning Up Defective Processes

For those eager to integrate AI, it is crucial to understand that technology will not automatically resolve blocked processes. Morgan warns, "You need to have clean data and a clear workflow from start to finish. You literally need to take a pen and paper and write down the process, and show me who owns that process." AI excels at highlighting inefficiencies, but it requires a solid foundation to be effective. At PwC, AI innovation and development rely on a well-planned data foundation established before AI becomes central. Likens emphasized that the challenge is to provide context to the data, often held in employees' heads, and to collect this tacit knowledge to improve processes. The goal of PwC's AI efforts is "the collection of tacit knowledge, telemetry, what agents are doing, which then feeds back into the collection of tacit knowledge."

Governance and Security in AI

At NBCUniversal, governance and safeguards are essential elements of the AI integration process. The level of oversight depends on the potential risk to the organization, which Morgan refers to as the potential "blast radius." "With a use case as simple as an agent tool that pre-schedules my lunch on my calendar, it's a low risk, we don't need a human in the loop," Morgan explained. In contrast, for more sensitive applications, such as automatically sending messages to consumers, the risk is higher and requires increased oversight. NBCUniversal's governance process includes admission forms that help Morgan's team track and measure the potential impact of AI and agents on the organization. At PwC, the responsibility for AI is centralized around "one percent of the organization - these are the specialized AI engineers." These engineers establish standards, create reliable and safe frameworks, and carry out the necessary builds. Then, 10% of the organization, spread throughout the company, are practical builders who develop solutions for clients, directly understanding an industry or function.