Last week several of our team members attended the 2025 Society for Industrial and Organizational Psychology (SIOP) annual conference in Denver, Colorado. This marquee event in our field brings together industry leaders, practitioners, and academic professionals to discuss, learn, and share on the most important topics. This year we heard a lot about building AI with “humans in the loop,” agents disrupting employee experience workflows such as hiring and skills-based learning, and the future of employee listening.
The Microsoft People Science led several sessions and participated in panels on topics related to agents at work, the future of nudge interventions, and manager action taking. Our sessions were packed with attendees, and we also hosted a customer networking event focused on how to measure AI’s impact on the employee experience.
We had an energizing and productive time at the conference, and we look forward to applying many of the ideas into our work. Our team is buzzing about all that they learned:
Building AI with a human in the loop
AI will reinvent the workplace, and we should welcome it rather than run from it so that we can develop it with proper guardrails and ensure it enhances engagement and productivity, fosters innovation, and maintains ethical standards. By embracing AI and establishing ourselves as the “human in the loop”, we can leverage AI’s potential to streamline processes, improve decision-making, and create a more dynamic and efficient work environment. Being the "human in the loop" means taking an active role in leveraging AI to enhance the workplace. It involves thoughtfully considering how AI can assist in hiring the right candidates, fostering career development, and helping individuals find meaningful and fulfilling work. Additionally, it includes identifying areas where AI can be applied to tasks that are rule-based, routine, and repeatable, while pushing the boundaries for what the future may entail, ensuring that these processes are optimized to reduce bias and improve reliability. This plays to our strengths as People Scientists.
Jaime Gonzales, Principal People Science Manager
AI coaching will help scale support at work, but empathy remains uniquely human
Using AI for employee coaching is likely to significantly scale (and in many cases enhance) coaching experiences globally, particularly for employees who do not normally have access to great coaching on-the-job (e.g., front-line employees, deployed soldiers). Aspects of coaching like career plan development, performance appraisal, feedback, and collaborative ideation are all likely to benefit from the development of AI solutions in the coaching space. Further, AI can help managers identify skill gaps on their teams, suggest relevant training programs, and offer tailored career advice, better fostering employee growth and development and actually enriching the leadership of a manager.
One aspect of coaching that the consensus suggested would remain uniquely human is the support that managers provide employees in moments of great need, vulnerability, and struggle. In these moments, the perception is that AI is likely to lack the nuanced understanding required in situations that demand empathy and direct human connection (those moments when an employee needs to know that their manager is standing beside them). The thinking goes that even if AI developed an improved capacity to understand and act with empathy, that the current workforce would not feel as supported and satisfied if the desired support were not coming from another person. Thus, while AI will likely augment several managerial tasks, it cannot (yet) substitute the profound importance of human empathy in some high-need and emotional work situations.
Eric Knudsen, Principal Manager, People Science Analytics
The data sources are expanding, but action continues to be the ‘why’
While there were more sessions than ever about different types of data collection methods and the use of passive data + AI analysis, the problem we are trying to solve has largely stayed the same: driving action. The growing interest in passive data collection methods and non-sentiment data are driven by better access than in years past and a desire from leaders to lighten the load on employees needing to provide their input via traditional surveys. These expanded approaches offer valuable insights, but it is crucial to remember that they are most valuable when they help leaders and managers make better, more informed decisions about how to best serve their people. While passive data may be held closer to HR and other groups who may own the data, those most insightful data points need to be shared back to the people within the organization (leaders, managers, and individual contributors!), ensuring that the insights lead to tangible improvements in their experiences and performance.
Carolyn Kalafut, Principal People Scientist
In uncertain times, we lean on psychological safety to help us adapt and be resilient
When everything around us is changing rapidly, psychological safety becomes increasingly critical. Among all the varied session topics at SIOP (e.g., AI, skilling, listening), psychological safety was present throughout. It truly underscored that speaking up, asking questions, and being authentic are key to navigating the spectrum of workplace changes we are currently navigating. When our leaders and cultures create safe environments, it demonstrates a commitment to people-centric change and acts as a lighthouse to guide us through. Presenters discussed creating spaces where training can thrive through vulnerability, where employees are comfortable with providing honest feedback, and where experimenting with AI and agents is celebrated. By leading with psychological safety, we ensure that change is not just about adapting but also about thriving, resilience, and meaning.
Megan Benzing, Senior People Scientist
At a Turning Point: AI, Work, and the Role We Must Play
At this year’s SIOP conference, the conversation around AI took a meaningful step forward, and the momentum is only growing. Rather than chasing trends, discussions focused on how AI is reshaping work and the role we must play in that transformation. The message was clear: our responsibility is not just to analyze this transformation, but to help build it. Three key themes emerged for me across the sessions I attended.
First, grounding efforts in familiar frameworks and well-established paradigms such as the Technology Acceptance Model or principles of social learning can help bring clarity and structure to adoption strategies. It doesn't have to feel like you are starting from scratch.
Second, start small, and start now. Use AI in everyday workflows, identify a couple of use cases, test, and iterate. We heard great examples from job analysis to item creation and automated dashboards, by starting with breadth rather than depth you can help identify where deeper investments will have the most impact.
Third, we need to evolve our skillsets. That means understanding how AI systems work, collaborating more closely with IT, product and design, and developing solutions that scale. We have a critical role to play in shaping AI at work, so it is ethical, effective, and truly human-centered.
Caribay Garcia, Principal People Science Manager
