Aging with AI: How technology can support caregivers without replacing them 

Beth Mynatt has dedicated her career to tech that helps older adults age in place. Now, with the major AI-CARING project winding down, she and her colleagues reflect on what they're discovered and how it can be of service.

by Madelaine Millar

Khoury Dean Beth Mynatt 
Khoury Dean Beth Mynatt 

This story is the third of a three-part series on Khoury research into tech and aging through the AI-CARING Institute. The first story covers the importance of responsible, accurate AI, while the second examines the value of AI that yields flexibly to human autonomy. 

Twenty-eight years ago, a computer scientist named Beth Mynatt sat down with a cognitive aging researcher named Wendy Rogers and asked a simple question: Did mom have a normal day today? 

“I joke that it’s the tagline for my entire research career,” Mynatt said. “It’s very easy to say and very, very hard to answer.”  

It’s also very, very important. Only care partners who know what their loved one’s “normal” looks can spot the changes that warn of progressive health issues like Alzheimer’s disease, or help a suffering older adult to return to their version of normal. Knowing whether today was normal is what lets caregivers support their loved ones when things are tough and celebrate with them when things go well. But it’s difficult to tell if a loved one had a normal day if — like most American caregivers — you don’t live with that person. 

So Mynatt has chased the question of mom’s normal day, from the hand-built touch screens of the earliest proof-of-concept studies through hundreds of scientific papers and tens of thousands of citations on human–centered computing and personal health informatics. She’s followed it through research center directorships and increasingly prestigious fellowships, gleaning insight into the design values that underpin responsible, respectful, and effective technology. She’s become a globally recognized expert in ubiquitous computing and assistive technology, the co-PI of the National Science Foundation’s $20 million dollar AI-CARING Institute, and the dean of Khoury College of Computer Sciences. And now, with the AI-CARING Institute’s five-year mandate drawing to a close and participating Khoury researchers reflecting on what happens next, Mynatt has never felt closer to a satisfying answer.  

“There are these longer questions, like where does AI technology fit into family norms and interactions, the ways that we take care of each other, and how we want to be represented? You only get those insights if you’re living with the technologies long enough, and AI-CARING has allowed for that in a way I never was able to do before,” Mynatt said. “We’re starting to see not only how these technologies fit together, but how they fit into the fabric of everyday life. When we can get there, then we’re really close to something for which I could imagine widespread adoption.” 

Building on Mynatt’s decades of work, Khoury researchers across three study tracks have surfaced key insights into how to design AI tools that meaningfully support human aging, including this one: AI tools’ role is to support caregivers, not replace them. 

Take PhD student Jiachen Li and Assistant Professor Varun Mishra’s work on a tool that sends a short, AI-generated summary of an older adult’s day to their remote care partner. In the status quo, care partners learn whether their loved one took a walk today by calling or texting to ask. 

Varun Mishra 

“We realized that no one likes that phone call; people don’t like being asked if they took their medications or not, if they did their PT,” Mishra said. But it’s important that caregivers have that information, so the summaries supplement that mundane information-sharing part, so they can engage in more meaningful conversations.” 

 “We did not observe anyone deciding to not talk to each other or changing their communication frequency when they have this new information,” Li added. “People get worried that AI-mediated information sharing will replace existing communication … we should design it from the beginning to be an enrichment of existing communication, instead of replacing anything.”  

Jiachen Li

The idea that offloading the irritating can contribute to better relationships with care partners is corroborated by Mynatt’s prior work. She recalls speaking with care partners while working on Emory University’s Cognitive Empowerment Program, where she created tech to support aging people with mild cognitive impairments. 

“What we heard from the caregivers was, ‘Since the diagnosis, I feel like all I get to do is nag, answer questions, or get frustrated,’” Mynatt said. “They told us, ‘If the system could offload some of that, I get to be a human being. I get to be a spouse, as opposed to being consumed with caregiving.’” 

The same design principle — to support people, not replace them — also guides both AURA, the intelligent reminder and alert system, and Robin, the HIPAA-compliant conversational agent and smart speaker. Instead of trying to take care of an older adult’s every need, researchers Zhi Tan, Reina Chan, and Tamara Zubatiy looked for the mundane tasks that annoyed and overwhelmed caregivers: checking that the oven was off, remembering what’s-his-name from that one movie, signing up for a crochet workshop, making to-do lists, keeping track of which pills to take on which days. 

Tamara Zubatiy 

Zubatiy recalled one family with three adult children who helped their parents grocery shop. While Robin was capable of setting up a grocery delivery, they instead used it to coordinate the family’s calendars and track whose turn it was to take the parents to the store.  

“They had this great system, but all of the calendars were disparate,” she said. Alongside unifying the calendar, “Robin made it possible for them to pop in on speakerphone and deliver a message about the schedule to the parents. The family was so stoked about it, because then they could keep up better with their parents and minimize their driving time.”  

“AI care is not an automation or a substitution, but it can really shift the needle around making human caregivers more effective and supported,” Mynatt agreed, comparing it to the difference between replacing a cashier with an automated till versus giving them a barcode scanner.  

AI-CARING began while Mynatt was still working as a Regents’ and Distinguished Professor at Georgia Tech, and she brought the project to Northeastern when she became the dean of Khoury College in 2022. She felt the college’s balance of technical savvy and HCI expertise would make it a great fit for AI-CARING, and she was impressed by the longtime health informatics expertise of Khoury researchers including Stephen Intille, Misha Pavel, and Holly Jimison. Mynatt was grateful and flattered that Li, Tan, and Zubatiy followed her up to Northeastern, but for Zubatiy, the decision was a no-brainer.  

“Beth Mynatt is a freaking vision,” Zubatiy said. “I consider her such a deep inspiration, both in terms of a model for how to pursue multiple causes you care about while exceeding across the board, but also in terms of thinking about things from a systems perspective and not an atomized perspective.” 

That perspective comes into focus when Mynatt defines success. The structure of research funding traditionally breaks the human experience down into problems to be solved: a faucet left running, a forgotten appointment, a missed pill.  

“But that’s not the fabric of day-to-day life,” Mynatt said. “Care technology needs to support the things the aging person wants to do, that are important to them … then there are some health-oriented things, some caregiver-oriented things, some alerts and emergency things too. But if you only design for emergencies, the odds of people bothering to have it are so low.” 

Instead of approaching old age like a string of emergencies to resolve, Mynatt believes successful AI care technology creates a generation of older people living normally and aging happily in place, with care partners adequately supported to have lives outside of their responsibilities.  

Even as that goal appears on the horizon, Khoury researchers continue to ask new questions. What does it mean to consent to use a piece of technology when its capabilities change after you bring it home? How do new sources of information change our communication obligations and our expectations for how others communicate with us? And with the AI-CARING Institute wrapping up at the end of 2026, are there ways for the public to access high-quality, research-backed aging support technology?  

Some of the researchers — like Li and Chan, who are both still completing their PhDs — want to continue contemplating those questions in a traditional academic setting. Others, like Zubatiy — who is also the co-founder and CEO of the AI media vetting company Barometer — are curious to explore a commercial approach.  

Mynatt believes in balancing both. AI-CARING has allowed the team to conduct much longer studies than is usually possible, finding out what impacts AI technologies have over time and responding to unexpected consequences. But the trade-off has been smaller sample sizes; they can spend months observing 10 people’s experiences, but not 10,000. A commercial approach would force the team to focus on the most effective approaches they’ve uncovered and to prioritize deploying them in a cost-effective way. Combining it with a research approach would let them critically examine the results of a larger sample and confirm whether their insights hold true at scale.  

Standing at the end of AI-CARING and the beginning of what comes next, Dean Mynatt finds herself recalling those first fragile touch screens, hand-built almost 30 years ago by a fresh-faced assistant professor who believed that elderly people and their caregivers deserved dignity and support. 

“The technologies have caught up so much. I’m very excited that we’re at this sweet point where it’s feasible and we know enough to have large-scale impact,” Mynatt said. “And I’m very excited for Khoury College’s leadership to continue in this space. This team is amazing, and we’re at the cusp of really pushing this to the next level.” 

The Khoury Network: Be in the know

Subscribe now to our monthly newsletter for the latest stories and achievements of our students and faculty

This field is for validation purposes and should be left unchanged.