by Larry Magid
This post first appeared in the Mercury News
A recent article in Communications of the ACM, a magazine of the Association for Computing Machinery, warned that although AI tools can help smooth over difficult communication between friends and coworkers, reducing misunderstandings, hurt feelings, and social discomfort, they also have a downside: As author Samuel Greengard cautions, “Left unchecked, they may erode human intuition and sabotage empathy.”
It got me thinking about my own use of AI as well as that of children and teens who, compared with most adults, have fewer years of experience when it comes to in-person communications, merely because of their age.
Human interactions, including engaging in difficult conversations, is one of many things people are outsourcing to AI. And I have to admit, I was guilty of that only hours before I sat down to write this article, when I used ChatGPT to help a friend write an email to fellow board members of a non-profit organization regarding a policy disagreement they were having. The person asked my opinion before sending the email, but rather than helping them re-write it on my own, I pasted it into ChatGPT, which did a fine job coming up with just the right wording to be persuasive, yet non-confrontational.
If they had asked for my help several years ago, we probably would have talked at length about the nature of the conflict and ways to be assertive without being too aggressive, and we would have drawn on our collective experience to craft a thoughtful message. But it was so much easier to outsource that work to a chatbot.
I don’t feel too bad about my use of AI in this instance, mainly because I do have experience with such matters and was able to create an adequate first draft on my own, using AI mostly to fine-tune what I had already written. I also have enough experience to know whether the AI got it right.
But just because I feel OK using such a tool, I worry about those who don’t have that experience, especially young people.
In this case, the email was about a policy dispute, but people are also using AI to craft emotional responses to situations ranging from personal disputes and even breakups to expressing empathy or sympathy for other people’s feelings and experiences.
In that same Communications of the ACM article, danah boyd, professor of communication at Cornell University and founder of the research institute Data & Society, put it this way: “Empathy is something that we learn, not something that we just suddenly wake up and understand. Different tools allow us to strengthen or weaken our ability to be empathetic.”
Research supports boyd’s observation. A meta-analysis published in the Journal of Counseling Psychology examined 18 randomized controlled trials involving more than 1,000 participants and found that empathy training can measurably strengthen people’s ability to understand and respond to others. The findings, according to the study, indicate that empathy training tends to be effective.”
Whether empathy comes from formal training or simply from life experience, it’s something we develop over time and need to practice. Not in the way we practice a musical instrument or a sport, but simply by interacting with people over the course of years or decades and learning what works, what doesn’t, and how our words and actions affect others.
Neither humans nor AI always get it right
Of course, we don’t always get it right. As the saying goes, “To err is human.” And when it comes to listening, apologizing, showing empathy, expressing compassion and kindness, or resolving conflicts, I’d rather we make the occasional mistake than check out completely and rely on machines to express our humanity.
It’s not just empathy I worry about. It’s the growing tendency to turn to AI for all kinds of decisions, ranging from which smartphone to buy to far more serious decisions such as whether to quit a job, end a marriage or undergo a potentially risky medical procedure. I’m not saying that AI isn’t useful for some of these issues. I use it myself to help decide which products to buy or what movies to watch. But there’s a difference between using AI as a research tool and letting it make decisions for you.
When it comes to life-changing decisions, especially those that affect relationships, it’s important to do your own thinking. And if you need help, consult a therapist, a trusted friend, or someone else who can help you sort through your thoughts, feelings, values, and options before you do something you might regret or fail to do something that could improve your life or benefit others.
And if it’s health-related, AI can be a useful research tool, but always check its sources and never make medical decisions or start or stop treatment without consulting a qualified healthcare professional. Remember that AI isn’t definitive. It can be wrong, incomplete or lacking important context, and it can cause unnecessary anxiety or provide a false sense of reassurance, as I’ve learned from my own medical research.
Talking better than texting
And, with or without AI, be cautious about using texts, emails or any form of online messaging to address emotionally charged issues with friends and loved ones. Research, along with my own personal experience, suggests that conversations offering more social and emotional cues can reduce misunderstandings, so an in-person conversation is usually the best choice when possible, followed by a video or phone call. A 2021 study published in New Media & Society found that face-to-face contact was associated with greater well-being than digital communication, suggesting that online interactions may not fully replace the emotional benefits of being together in person.
As AI becomes more capable, the challenge isn’t keeping up with the technology. It’s making sure we don’t lose the human skills that technology can never truly replace.
[related_articles location=”left” show_article_date=”false” article_type=”automatic-primary-section”]
Larry Magid is a tech journalist and internet safety activist. Contact him at larry@larrymagid.com.