What is “responsible” AI?  

AI adoption in HR is accelerating, fast: 43% of enterprise HR teams are using some form of AI, according to SHRM. But many of these teams are still in pilot mode or keeping adoption concentrated to select use cases instead of a full-scale enterprise roll-out. 

Meanwhile, pressure from the top is building. Nearly 60% of HR teams are working AI into their strategies, but say that readiness, data, and governance are their top barriers.  

And that has us asking: What does the “right” AI look like?  

The data is pointing to responsible and well-governed models that are proven to improve productivity and results. 

What is responsible AI? 

Responsible AI looks like and means a few things to us and our clients:  

• Application: Efficiency, connectedness, and foresight drive how we bring AI into our world.
• Intelligence: Accuracy, consistency, and usefulness are prioritized over novelties.
• Governance: Clear guardrails, continuous monitoring, and defined oversight. 
• Privacy: Disciplined and transparent data practices, limited to use only what’s necessary. 
• Accountability: Above all, humans maintain oversight and responsibility for outcomes. 

Our chief AI officer talks more about this in his latest company update.

What to ask at RFP about AI 

A quick poll of our RFP team found that AI questions make up about 5% of total questions in RFP and RFI, just for general health and wellness administration.  

Our RFP manager broke down some of the best AI questions she’s seen in RFP lately. Here’s her list of recommendations, based on her 15+ years of experience in our industry: 

• Is my data being used to train the model? 
• How do you mitigate and monitor for bias and hallucinations? 
• What security controls and guardrails are in place? 
• How are AI decisions documented, audited, and verified? 

What’s most important for HR and broker teams to think about in the HR technology buying process is “what do we need for our business and will this keep us in compliance?” 

Keep reading: 10 questions to ask at RFP about AI 

How are you measuring responsible AI? 

Proof and measurement is arguably where we see the biggest divergence between custom-labeled AI and true intelligence infrastructures. But more than half of today’s HR teams say they aren’t measuring the success of their AI investments.  

We’re helping benefits administration teams measure the impact of AI and bring that back to their leadership team to have more strategic conversations about what they’re investing in and why it matters. 

Here’s an example of the data we bring to our business review conversations:  

Satisfaction and service quality:  

AI helps us measure quality service and resolutions in the call center by taking notes, analyzing call sentiment, and helping live agents pull up the member’s plan information.  

• 62% of calls saw sentiment improvement in member calls  
• 91% of all calls are positive or very positive sentiment
• 87% of all AI chats stay resolved after 7 days 

Our annual Benefits Insights report goes into more detail on how to measure AI success in benefits 

Responsibility in AI looks at how it’s applied, how it’s governed, and how confidently HR leaders can stand behind it. As AI becomes more embedded in benefits and HR, the real differentiator won’t be who adopted it first. It’ll be who adopted it thoughtfully.