Approx 800 words; 3 minutes read time
Empathy is the ability to imagine another's feelings and react accordingly. We have the capacity for it hardwired into us from birth - it's part of our nature. But empathy is mostly learned - nurtured into us over time. In this way, empathy is like language - we are capable of both, but we must learn and practice to master either. And because empathy is learnable, we can teach machines to do it too.
They might even be better at it than us one day soon.
***
We hear a lot about artificial intelligence (AI) these days. AI is programing that assesses situations and takes action much like natural or human intelligence. AI - is not a future technology waiting to happen. Instead, it is here now, in TVs and smartphones and many other devices. Speech recognition and self-driving cars are two prominent examples. But AI is also used in tracking consumerism and even in "contact tracing" for COVID-19. It's everywhere, and it's growing. And learning.
AI, as we know and use it, is not yet like human intelligence. Most AI remains mechanical and logic-based, whereas human intelligence uses logic and emotions. These interact in complex ways achieving generalized intelligence.
Logic alone is cold and binary. Zero and one. Yes and no. This path or the other. We've all confronted an automated phone system that bewilders with unnecessary questions, taking forever to get anywhere. Now, contrast that with a real person who listens and connects right away. That's the distinction in generalized intelligence. Logic gets you from A to B without considering the ride. But generalized intelligence - with both logic and empathy - gets you there and cares how the trip was.
Empathy in generalized intelligence allows social groups to function and thrive. As we empathize, we see different viewpoints and care about what others think and feel. Effective social networks depend on this empathy to self-govern. In turn, communities form around these networks of trust and understanding. And the community supports the individual while the individual buys into the community. For intelligent machines to enhance society, they must care like we do (or rather, like we should).
***
Advancing AI requires making artificial generalized intelligence (AGI) more feeling to make it smarter. And at the core of enhancing AGI this way is artificial empathy. As discussed, empathy is learnable. And we can distill it into component thoughts and actions to program machines with it.
Programming empathy is possible even now and is an active area of AI research. Artificial empathy involves distinguishing social cues and programming actions around these to make emotional sense. But even if we can design machines to care, is it real? After all, in programming a computer to act with empathy, we aren't making it feel like a human. We are only teaching it the nuances of empathic action. That's not the same as caring. Or is it?
When we empathize, we imagine what it would be like to be another person. But we don't know for sure. Instead, we relate to their feelings and believe it as if it was our own. We then act to comfort and support that person as best we can. We can only approximate and care, knowing another is feeling. If a machine can do the same thing, without having to feel, the difference is moot. After all, we need only the result - empathetic action - in times of need. If it walks and talks as genuine, maybe then it is.
***
Machines might not only be good at empathy soon, but they could also be superior at it. As humans, we are susceptible to all kinds of bias that prevents us from connecting with others. Prejudice clouds our judgment and keeps us from acting as we should. With machines, these biases would not be present.
Unencumbered, machines might be even better at empathetic tasks than humans. Emotional support robots would always be available and in tune with their human, for example. And emergency call bots could handle even the most nuanced situations. Even mundane tasks like tech support would live up to expectations.
***
Yes, artificial empathy holds promise. But all this begs the question: if we are remarkably close to programming AI to empathize well, why do we fall so short ourselves? Is all that bias we own too high a burden to overcome? Or is it that humans need this next step - outside ourselves and into the machine - to become fully human?
As we trend towards empathetic machines, maybe they will soon take the lead. Hand in hand, man and machine, both trending towards being more human together than either ever could be alone.
Until next time. Science. Fiction. Create.
JRC