What Might an AI-Augmented Human Job Look Like? (Animation Transcript)
Narrator: What might it look like for AI-based tools to work alongside humans in real-world settings? While we may be more and more used to seeing human employees working with robots in factories and warehouses, it can be hard to imagine how this balance might work for service-related jobs in particular. How can an AI-based robot be helpful to a human employee in assisting a human client or customer in a store, hospital, or home without being in the way, or causing more problems than it solves? This is called augmentation, and many robotics companies are working with workers and businesses in different service industries to figure it out.
To get a sense of how they are thinking about this balance, consider, for example, the responsibilities that make up the job of a home health aide. These individuals must excel on a variety of fronts to do their job well. They must have strong and patient communication skills to interact with clients and determine how they are doing physically, cognitively, and emotionally. This may involve awareness and expression of cultural norms and language proficiencies. Health aides must have physical and motivational skills to ably but gently help clients with mobility limitations to get around, feed and care for themselves, and get some exercise. And of course they must have health care knowledge to monitor clients’ vital signs and conditions, and to be sure clients are taking the right medications in the right dose at the right time. It’s also important for health aides to be flexible and adaptable to various home conditions and to an often changing schedule with a different range of clients and client needs each day. Overall, their job is a complex mix of physical, emotional, medical, and logistical tasks.
So let’s say we were to sub in a robot equipped with today’s AI capabilities in place of an experienced human health aide. Based on what we know of AI’s current strengths and assuming it has access to detailed patient records, the health aide robot could probably do a great job at keeping track of and quickly finding information in each client’s medical history. The robot could accurately dispense the right medication at the right time, and could also do the physically-taxing work of lifting and moving clients for a full workday without breaking a sweat. Though clients with memory limitations might ask the same questions over and over, the robot would never become impatient in responding. And the robot would be able to address some of the clients’ basic needs at any time, day or night, as it wouldn’t have competing professional or personal appointments or, well, need to eat or sleep.
But the robot would run into plenty of problems where a human health aide would not. For example, if a client asked a question or presented an issue that the robot had not encountered before and had no reference for, it might offer an incorrect response, or it might not answer at all. If the physical layout of a client’s room had changed at all since the robot’s previous visit, it could struggle to navigate the space. If a client’s speech was impaired or significantly different from the language samples the robot was trained on, it wouldn’t be able to understand the client’s requests. While the robot may have no problem lifting a client to move them from a bed to a chair, it would likely not have the deftness to place the client gently in the chair rather than just dropping them into it, possibly knocking the client into walls or furniture along the way. And, importantly, though the robot might respond to a sad or lonely client with comforting words, it can neither feel nor express the compassion, empathy, or human connection behind those words in tone or gestures like a human health aide would.
So what if, instead of replacing the human worker with a robot, we were to equip our human home health aide with an AI-based machine partner? The human aide could focus on the more human aspects of the work, such as interacting with the client, handling unexpected questions or problems, and making the client feel more human and cared for than a robot is currently capable of. Meanwhile, the robot could assist the human aide by providing quick medical and patient record information, and lend a hand with lifting and moving the client—with the aide’s gentle guidance to keep the client safe. The robot could remain on site with the client to handle basic off-hours needs and alert human care providers in an emergency. The human health aide would be freed up in some key respects to do what they do best—provide focused, caring attention and address specific changes and more complex concerns. A robot can’t do everything as well as a skilled human can, and it lacks a vital human touch, of course. But augmenting the skilled human worker with the unique capacities of an AI assistant could help ease the human worker’s burden and enhance the care they provide to clients in important ways.