How well can large language models interpret employee feedback, and what does this mean for HR and people analytics leaders?
In this episode of 600 Seconds in HRTech, Stacia Garr sits down with Joseph Freed, Chief Product Officer at Perceptyx, to discuss PYX Labs, a public benchmark designed to evaluate how well large language models interpret employee experience data.
The conversation explores how organizations are using tools such as ChatGPT, Claude, and Microsoft Copilot to analyze employee listening data, and how effectively these models can identify themes, interpret employee sentiment, and recommend actions for leaders. Stacia and Joe also take a closer look at what this means for the roles of people analytics professionals in the future.
QUESTIONS ANSWERED
How well do large language models analyze employee feedback?
LLMs can perform well when retrieving information and identifying common employee experience themes such as well-being and performance management.
The discussion also examines how models handle themes that are specific to an individual organization, including areas such as change management and future vision. Organizational context can play an important role in interpreting this type of employee feedback.
Can AI turn employee listening insights into action?
The episode explores what Joseph describes as the “insight to action gap.”
PYX Labs tested whether AI could take employee experience findings and recommend what leaders should do next. The discussion highlights the importance of behavioral science, organizational context, and expert judgment when translating employee feedback into recommendations.
How can AI change the role of people analytics professionals?
AI can give people analytics professionals new ways to explore employee data, ask questions, and investigate patterns across different datasets.
This creates opportunities for HR and people analytics experts to spend more time interpreting complex findings, connecting information across employee listening programs, and helping organizations determine appropriate actions.
TIMESTAMPS
00:00 – Introduction to Joseph Freed and PYX Labs
01:22 – Why Perceptyx created PYX Labs
03:36 – The criteria used to evaluate LLM performance
04:58 – Where AI performs well: common employee experience themes 06:00 – Where AI struggles: organization-specific themes
07:22 – How the PYX Labs benchmark works
10:54 – Testing AI's ability to move from insight to action
12:00 – Why AI struggles to connect the dots across employee datasets 14:50 – Advice for HR leaders using LLMs with employee data
16:28 – Stacia’s Analysis
RESOURCES & LINKS
Learn more about Perceptyx
Perceptyx Official Website: https://www.perceptyx.com/
Joseph Freed on LinkedIn: https://www.linkedin.com/in/joefreed/
Connect with RedThread Research
RedThread Research Official Website: https://redthreadresearch.com/
RedThread Research on LinkedIn: https://www.linkedin.com/company/redthread-research/
Stacia Garr on LinkedIn: https://www.linkedin.com/in/staciashermangarr/
続きを読む
一部表示
13 分
20 分