How ServiceNow is Building an AI-Powered Culture within HR

The Next Practices Weekly call series has become a well-attended and wide-ranging discussion for HR leaders each Thursday at 11am ET / 8am PT. On this week's call, Judy Albers, i4cp Director of Research Enablement, and i4cp Senior Research Analyst Tom Stone, were joined by special guest Brandon Roberts, Group Vice President of People Analytics and AI at ServiceNow. Here are some highlights from the call:

  • Roberts has worked in People Analytics at Qualcomm and Pinterest prior to joining ServiceNow, where is now Group Vice President of People Analytics and AI. Early in his career he fell in love with using data to making people's work lives better.
  • Roberts was not personally an immediate, early adopter of ChatGPT when it first came out to the public. But after a few months, he dedicated a week to experimenting with it in that every question he had for that week, he input it into ChatGPT to see what it would provide. This was a great learning experience about where it was helpful, and where less so.
  • Early on, HR at ServiceNow arrived at these four questions to drive their AI strategy:
    • How is AI going to change work across the organization? (e.g., Sales, Engineering, etc.)
    • How is AI going to change how the HR function gets work done?
    • How is AI going to change HR tech/analytics and how we use it?
    • How do we ensure ethical, moral, and legal use of AI in all our use cases?
  • Roberts then shared ServiceNow's four-point plan for creating an AI-Powered People Team:
    • Implement AI Operating Model. Create an "AI Factory" in partnership with our People Team Product Model to Supercharge AI and focus on the right use cases aligned to our strategy and guiding principles and with the right governance.
    • Build the Tech Foundation. Develop strategy for AI tech partnering with IT and Product to build best in class solutions, leveraging our own platform whenever possible.
    • Invest Strategically. Invest in thought leadership, talent, required infrastructure and enhanced data sources to augment internal data and accelerate valuable use cases. Develop governance model.
    • Educate and Enable. Support AI education across the organization, identify areas for reskilling and upskilling, and take targeted action to build the workforce of the future (know AI, use AI, sell AI, build AI).
  • They also considered how AI can be used to further enable each stage of the employee lifecycle: Find me, Welcome me, Engage me, Reward me, Grow me, Wish me well. (See slides above for a visual of this lifecycle.)
  • Roberts said that the People Team Operating Model of the future is AI-driven and people-led – a partially centralized model to ensure they are focused on the right things with decentralized execution to drive innovation. They found they were getting a lot of AI use cases and ideas, so they created an Idea App, where each is described in detail: the problem, solution idea, etc. Each is assigned to an appropriate group based on the six phases of the employee lifecycle above. After that group does some review and work on the idea, it goes to HR and Legal Review for legal, ethical, bias, etc., review. The ideas then go to the HR Steering Committee for prioritization, and then ultimately back to the relevant group for execution.
  • Roberts shared that at ServiceNow they look at how AI impacts Work, breaking it into four ways:
    • Automate. Automate or eliminate work previously performed by humans.
    • Augment. Direct support to humans to increase productivity, improve experience, enhance job efficiencies and work output
    • Extend. Performs activities humans are unable to perform or scale.
    • Create. AI creates things, and humans develop news skills and take on responsibilities to design, test, deploy, monitor, and maintain AI solutions.
  • Several examples of use cases where Roberts is really seeing value already included:
    • Sentiment analysis at scale by bringing together all sorts of employee experience and listening data and creating an interface where you can ask questions like "What are employees saying about the new leadership development program?" or "What do employes think of the current compensation package?"
    • Using their own product, NowAssist, for more efficient HR Service delivery by summarizing an employee support case in situations where it is moving from one support agent to another.
    • Using generative AI to search large internal knowledge bases to answer policy questions better than current search solutions. It is important here to test this thoroughly to minimize inaccurate results. The solution should also link back to the source of its answers.
    • Beyond Generative AI, Roberts also shared where they are using traditional AI and machine learning for use cases such as headcount prediction, likelihood of a requisition closing, end of quarter headcount totals, expected terminations, flight risk, and process mining for possible operational efficiency gains.
  • Roberts suggested that data dashboards, including HR dashboards, are ripe for disruption by AI. Many people are not happy with the dashboards they have available, as there is far too much data--way more than they need and it is confusing. AI can help find the key insights and the most important pieces of information, allowing the full data dashboard to take a backseat and still be available for ad hoc or more rare data needs.
  • At ServiceNow, they use a simple value and effort matrix to help prioritize the AI use case ideas. Low effort and high value should be prioritized; high effort and low value can be ignored; those that require more consideration are those that are either low effort but low value, or that are high value but also high effort. Layered on top of this are considerations of ethics, bias, and privacy.

Links to resources shared on the call: