Geoffrey, can you provide a brief overview of your role as the Vice President of Data Analytics at Alight and your expertise in AI and data analytics within the HR domain?

I look after Alight’s AI, personalization, and analytic capabilities from a product and data science perspective. This includes Alight’s chatbots, search engines, personalized nudges, and recommendations capabilities we provide to our clients and their employees as well as some of the AI-enabled automations we’re putting in-place to deliver high-quality ongoing service.  

We’re continuously enhancing our capabilities. For example, We recently unveiled Alight LumenAI, our next-generation AI engine to power Alight Worklife ®, Alight’s employee experience platform. 

We observed three consistent imperatives for creating leap-frog HR AI capabilities: 

1. Adoption of Generative AI (Gen-AI) 

2. Tying together AI capabilities with one unified view of an employee

3. The use of ever-growing internal and external datasets to improve model performance.  

That’s why we launched Alight LumenAI – to ensure we could continue bringing market-leading AI capabilities to our HR clients.

I’ve been passionate about AI and data-enabled SaaS products for a long time, regardless of sector, and that’s reflected in my prior roles building AI-powered experiences in cybersecurity at SecurityScorecard, finance at Bloomberg, or consumer goods at Arena AI.

I joined Alight because applying AI and data science to the Human Resources (HR) space is a chance to deploy AI “for good” – ensuring people are enrolling in the right benefits, preparing appropriately for retirement, having a seamless employee experience, and generally maximizing the wellbeing opportunities offered by their employers.

Right now is an especially exciting time to be at Alight: our clients are being pushed by their CEOs to demonstrate transformational AI strategies within HR and the AI capabilities Alight offers fit very well into these strategies and can deliver  transformational wins.

AI and employee experience:

How do you envision generative AI and AI-powered platforms shaping the future of the employee experience in the workplace?

There’s a baseline level of transformation happening everywhere, where most of the tools we use to do our work are getting generative AI upgrades. 

Taking a step back, we broadly see AI fitting into 5 categories within the HR domain – and these mirror the capabilities that we bring to market for our clients:

• AI Personalization: capabilities that drive a greater than 10% increase in targeted client HR outcomes through personalized, “next best action” content
• AI Assistance: capabilities with natural language/intent models to maximize digital engagement, supporting a 90% self-service rate
• AI Recommendation: capabilities providing automated decision support and choice optimization for benefits and care, saving employees on average $500 in premium expenses annually
• AI Insights: capabilities with data trend analysis for high-precision employer analytics to identify hotspots in the employee experience and HR processes
• AI Automation: capabilities that streamline repetitive workflows, such as document processing or at-scale call monitoring

Whereas in the past, an HR team might adopt a few of the above capabilities, we’re seeing that teams succeeding with AI transformation are adopting tech across all 5 of those categories.

Can you give examples of how AI-driven innovations have already improved the employee experience in organisations?

 We’ve been working with clients to deploy AI for years, even before the surge in generative AI interest that has taken place over the last year.  A great example we’ve seen is through our Interactive Virtual Assistant (IVA), or chatbot, that has helped employees answer their benefits questions in a personalized and self-service way and helps drive a 90%+ digital interaction rate (so folks aren’t needing to get in touch with a call center).  

When we launched IVA about 5 years ago, its initial performance was “ok” – but in the intervening years we’ve spent millions of dollars on teams tuning the algorithm based on the results of performance across 30M interactions with employees – so that now our IVA offers market-leading performance. 

It’s an important lesson to remember: AI systems often require ongoing maintenance and investment by professionals to achieve differentiated performance.  Having a “human in the loop” is incredibly important.

Our AI-powered IVA continues to see increased engagement from employees and was recently enhanced to also execute transactions – for example, by allowing employees to re-enroll in their health coverage plans during annual enrollment.  

We’re also excited to be piloting a GenAI-enhanced version of our IVA, powered by Alight LumenAI, that provides more holistic and helpful answers to questions where information was locked-up in complicated policy or benefits documents.  The results have been pretty spectacular – one of our clients when they used it for the first time said, “this is amazing, can we just roll this out now!”

Efficiency and productivity:

In what specific ways can AI enhance efficiency and productivity for both employees and employers in today’s evolving work environment?

In the HR vertical, efficiency is often about trying to reduce call, ticket and email volume for HR teams so that work shifts away from repetitive administrative employee needs and towards more consultative high-value activities.  

Anything AI can do to reduce the volume of administrative calls and tickets is immensely helpful.  AI can help HR teams diagnose, at scale, what is driving the high call and ticket volumes to shorten what are often very long process-improvement cycles…and it can also help create more effective interception-points to help individuals self-service their needs.  

For example, imagine an employee with an HR need logging into their internal HR portal, and then using an IVA chatbot to try to answer their question, and then using a voice-based Interactive Virtual Response (IVR) call-routing system when they call the call center. That’s three interception points where AI has an opportunity to help an employee self-service for a better, faster experience before they get to an agent.

Intelligent Document Processing is another great example of how we partner with clients to improve experience and reduce cost.  Many HR processes still depend on employees submitting documents (deposit checks, birth certifications, etc) and so when we deploy intelligent document processing we reduce the time it takes to process documents and provide feedback to users from 10+ minutes to near-instant.  Not only is this fast feedback loop a better experience for employees, it also tends to reduce calls to the call centers from employees asking about document status.

Personalization:

How can AI enable more personalised experiences for employees, and why is personalization important for overall employee satisfaction and engagement?

Personalization is a pretty broad term and can encompass many things. It can start as basic as knowing what benefits someone is eligible for and only showing them information about those and scale all the way to using AI to nudge or prompt employees according to a next-best action framework.  

Without a baseline of personalization in place, employees can quickly become disengaged by an experience that feels irrelevant to them. Once that baseline is there, you can start to play with personalization that drives outcomes. We partner often with clients on personalized communication campaigns that drive outcomes such as increased 401k contribution, HSA contribution, or increased utilization of specific programs like healthcare navigation.

For example, in March 2024, a pharmaceutical client selected Alight to help improve the financial wellbeing of their workforce through personalized messaging that encouraged the adoption of changed saving behaviors.  With only 75% of employees participating in a Health Savings Account (HSA) and a majority saving below the maximum allowed amount, the company aimed to encourage greater participation in retirement and health savings plans and ensure that employees were taking advantage of the company match to both the 401(k) and HSA.

With a focus on employees who had not yet maximized the value of tax-advantaged accounts, the company partnered with Alight to leverage personalized email and web messaging that would influence saving behaviors. This personalized messaging was made possible with adaptive, “Always On” AI technology that dynamically adjusted engagement strategies to drive up retirement and health savings contributions over time. Upon partnering with the client, Alight took strategic steps to ensure seamless integration and successful implementation of the AI-driven program. 

Key initiatives included:

• Assessment: The Alight team conducted a comprehensive needs assessment to understand the specific challenges and goals of the client in-depth.
• Data analysis: Extensive analysis of existing data, including employee participation rates, savings patterns and financial behaviors, were undertaken to inform the AI-driven personalization strategy.
• Integration planning: Alight collaborated closely with the client to develop an integration plan, identifying areas for personalized content implementation within existing communication channels.
• Customization framework: A tailored framework for content personalization was established that considered the unique characteristics of the client’s workforce and desired outcomes.
• Pilot programs: Small-scale pilot programs were initiated to test the effectiveness of the AI-driven approach, allowing for adjustments and refinements before full-scale implementation.
• Continuous monitoring: The Alight team implemented continuous monitoring and feedback mechanisms to track the success of the AI-driven program and ensure ongoing adaptability.

Post-implementation, Alight conducted thorough assessments of the system’s impact on both 401(k) and HSA participation, and success was substantiated by the substantial increase in employee contributions to both. Additionally, tax savings projections were delivered to show the true value of these funds.  Planning, testing and effective execution of the new AI-driven messaging system took less than six months.

As a result, the pharmaceutical company realized a substantial 17% increase in employees starting or increasing their 401(k) savings. Achieved a commendable 6% increase in employees starting or increasing contributions to the HSA. Notably improved the average 401(k) contribution rate, showcasing an impressive increase of 5.4%, and demonstrated tangible financial impact with an average increase of $1,750 in employee HSA contributions.

Measuring value:

What strategies can companies employ to effectively measure the value derived from their investments in employee experience and well-being initiatives, using data-driven insights?

Most importantly, companies need to know the outcome they’re trying to achieve upfront, and they need to be measuring that on an ongoing basis.  Once that’s in place, there are varying levels of sophistication clients can deploy to measure and attribute changes in the planned outcome to the interactions they are executing.  

The gold-standard for these is treatment vs. control groups, but even basic attribution can give a basic measure of success.  In many cases, if there is a specific action an employer is trying to drive, they can track who took that action after experiencing a personalized nudge, and attribute these to the personalized nudge. Examples of impact we’ve seen using this basic measurement methodology include:

• Nudges delivered over 6 months to direct employees to financial coaches resulted in a 7% increase in enrollment in HDHP health plans
• Nudges delivered over 6 months to encourage employees to contribute more to their HSA campaigns resulted in a 33% conversion rate from messaging to action, and the increase in HSA contributions yielded ~$1M in FICA tax savings for the employer 

Data utilisation:

Could you elaborate on how organisations can responsibly utilise employee data to enhance the employee experience while maintaining data privacy and security?

Sure – organizations need to think both about overall data security as well as ensuring appropriate use of data specific to each experience use-case.  In general, the less places you send and store your employee data, the better and the less opportunities there are for data breach or inappropriate use.  When it comes to appropriate use of data, using it to enhance the employee experience should be governed by standard data risk management and security review processes.

Alight’s clients include government entities and defense contractors, so we’ve already been operating in a very robust data and cybersecurity framework.  Last year we formalized our approach to AI risk and now assess all use-cases of AI technology against an 8-part risk framework that looks at things like data risk, bad output risk, bias risk, etc.

Challenges of implementing AI:

What are the common challenges that organisations face when implementing AI-powered solutions for employee experience, and how can they mitigate these challenges?

We like to use an “AI Intrapreneur” framework that lays out five important considerations for any new AI use case and recommend careful consideration –if you’re thoughtful about these five factors you will successfully launch an AI use-case:

• Pick the right areas – Focus on problems AI can solve now, not speculative future capabilities. Validate with small, low-risk pilots.
• Resource wisely – Build in-house for differentiated capabilities, use vendors for commoditized capabilities.
• Avoid high-risk AI uses – AI will make mistakes: don’t use AI where those mistakes have severe consequences.
• Keep humans in the loop – Humans must oversee AI systems. Design AI use cases for human oversight.
• Measure extensively – Rigorously measure performance, error rates, biases and business impact. Establish feedback loops.

We took the above approach in our current Gen-AI IVA pilot – testing with a small number of users at a small set of clients, building some of the technology ourselves so that we could be differentiated in the market, and being very thoughtful about how we keep humans in the loop to ensure accurate answers to employee’s HR-related questions.

Ethical considerations:

Are there ethical considerations organisations should be aware of when integrating AI into employee experience initiatives, and how can they ensure ethical AI practices?

The most important ethical consideration – which we touched on in the above – is understanding what the consequence is of a bad model output and its consequence on a person.

Leadership and management changes:

With the integration of AI, how do you foresee the role of leadership and management evolving in HR and employee experience, and what challenges might this transformation present?

The biggest shift is likely to be that whereas before managers might be managing the quality of output of their team, they will now spend an increasing amount of time managing the quality of an algorithm’s output.  No AI system is perfect, and they all require some amount of human oversight.

Final thoughts:

As AI technologies evolve rapidly, what advice would you offer HR and business leaders to stay informed and leverage the latest AI innovations effectively for employee experiences?

Read and absorb as much as possible and stay curious!  Don’t expect to stay fully up to date – even AI researchers are getting surprised these days by sudden developments in the field.

More generally, be aware of your organization’s overall risk appetite and be comfortable with it – some organizations want to be on the leading edge, others may want to take a more conservative approach – both are OK.

Geoffrey Peterson

Vice President of Data & Analytics at Alight

Geoffrey Peterson is the Vice President of Data and Analytics at Alight Solutions, a role he’s held since 2023. Before joining Alight, he was Global Head of Product Management and Data Governance at Bloomberg and a Senior Product Manager at Security Scorecard. Earlier in his career, he was a Business Analyst and Associate at McKinsey & Company before moving into management roles at South African Breweries Limited. Peterson earned a BA in Computer Science and Economics from Harvard University and an ME in Computer Science from Cornell University.

The post AI-Tech Interview with Geoffrey Peterson, Vice President of Data & Analytics at Alight first appeared on AI-Tech Park.