AI and the Future of Crisis Care Pt 1

Why Human Connection Must Remain at the Center

John Draper, Ph.D.

Every generation encounters a technology that promises to change everything.

I've heard respected leaders compare this moment in artificial intelligence to the discovery of electricity. During a recent Crisis Jam Fireside Chat, my longtime friend Tom Gruber—the creator of Siri—suggested something even more revolutionary. He compared AI to the discovery of fire.

Fire transformed civilization. It illuminated the darkness, cooked our food, powered industry, and ultimately changed nearly every aspect of human life. But civilization also learned something important. Fire had to be controlled.

Artificial intelligence feels remarkably similar.

Over the past year, I've watched AI moving from intriguing novelty to inevitable reality in varying corners of behavioral health. State commissioners are asking about it. Crisis providers are experimenting with it. Technology companies are integrating it into their platforms. And last October, OpenAI estimated that millions of worldwide ChatGPT conversations each week involved suicidal thoughts or planning. In response, Dr. Ursula Whiteside facilitated an open letter to OpenAI, co-authored and signed by national suicide prevention leaders, outlining recommendations for AI safety in conversations involving suicide.

In many ways, our field is trying to answer the same question civilization once faced with fire:

How do we harness extraordinary power without allowing it to burn what matters most?

As AI becomes woven into nearly every aspect of our personal and professional lives, it will inevitably assume a larger role in crisis systems. Its potential to improve access, efficiency, quality, and scalability makes that not only likely—but desirable.

The challenge is making sure those advances strengthen the human connection at the heart of effective crisis care rather than quietly replacing it.

The Debate We Should Be Having

At the Crisis Now Summit IV in Washington, D.C., I had the privilege of participating in a debate on the proposition: “AI is the future of crisis care.” I argued the negative. Not because I oppose AI. Quite the opposite. I believe AI has the potential to become one of the most transformative advances our field has ever seen.

The question is not whether AI belongs in crisis systems. The question is where it belongs—and just as importantly, where it doesn't. That distinction matters because our goal should never be to ask whether AI can replace people. It should be to ask whether AI helps people care for one another better.

That distinction may ultimately determine whether AI becomes one of the greatest advances in crisis care and suicide prevention—or whether it inadvertently becomes what social media increasingly appears to have become: a technology celebrated for connecting us while quietly contributing to our disconnection from one another.

People experiencing emotional crises often tell us they feel profoundly alone. If we confuse increased access to LLM-generated “conversations” with increased human connection, we risk deepening the very problem crisis care exists to address.

This first blog of two is not about fearing technology, nor is it about protecting jobs. It asks a much simpler—and I believe much more important—question:

What is the active ingredient in effective crisis care? What do people in crisis need most?

Human Connection Is Foundational to Effective Crisis Care

Last year I was listening to a podcast conversation between Rick Rubin and Jack Clark, co-founder of Anthropic.

Clark made an observation that has stayed with me.

“We've never had something to compare humans against that's as human-like as AI... I think this is going to yield truly amazing breakthroughs in our understanding of what makes humans distinct.”

I agree. And perhaps nowhere is that distinction more important than emotional crisis. Emotional crisis is not simply a problem to solve. It is a profoundly human experience.

From the moment of our first cries, it has been human caring that soothed our distress. Someone heard us, wrapped us in their warm embrace, reassured us with their voice, their words. Long before we understood attachment theory or interpersonal neurobiology, we experienced one of the most fundamental truths of being human: We are wired to associate compassionate human connection with safety and coping through overwhelming emotions.

When we are overwhelmed by grief, terrified of tomorrow...convinced we are a burden...or wondering whether life is worth living...what we often need most is not another source of information.

We need another person.

What Over 70 Years of Research and Practice Have Taught Us

Long before the phrase evidence-based practice entered our vocabulary, Samaritans founder Chad Varah understood something remarkably modern in 1953. He didn't establish the world's first suicide hotline to assess people. He established it to befriend them. That is, managing crises through establishing an authentic human connection. His Befriender model trained counselors to convey genuine concern and caring to people in crisis, while bearing witness to their suffering. In so doing, they reinforce a sense of belonging when the individual is feeling profoundly alone. They provide acceptance and understanding, when a person is feeling judged and confused.

They communicate, “I am here with you,” because their presence is felt, within and beyond the words they use.

Beginning with his published article in 2007, internationally renowned suicidologist Brian Mishara's landmark observations of Lifeline counselors found that the strongest predictor of positive outcomes wasn't sophisticated assessment or problem solving. It was good contact—the counselor's ability to establish authentic rapport, empathy, and genuine human connection that allowed callers to feel safe enough to continue the conversation and work collaboratively toward recovery.

In his 2023 unpublished dissertation, Lifeline International's Alan Woodward reached the same conclusion as Dr. Mishara—but from the caller's perspective. After interviewing 53 Lifeline Australia callers over twelve months, he concluded that what participants sought most was “relief through authentic human connection:”

“Participants sought someone who authentically cared for them and their wellbeing… They wanted telephone counsellors that could challenge them at times and offer different perspectives. They needed telephone counsellors who had sufficient life experience themselves to be able to do so meaningfully.”

In her published findings from 2017, Dr. Madelyn Gould's evaluations of Lifeline follow-up calls revealed something equally profound. After learning that over 90% of the suicidal Lifeline callers who received follow-up calls stated that these calls “kept them safe,” and approximately half said the calls “stopped them from killing themselves.” When asked why these calls were so helpful, the recipients' primary feedback were messages like “they cared enough to call me” and that “someone who doesn't know me had an interest in me, cared about me…”

In all of these studies, people in crisis weren't simply looking for answers. They were looking for someone who cared.

What if We Can Make AI Sound More Human?

Chatbots are getting very good at simulating human empathy. They have read the research and practice accounts from the past 70+ years of research and practice. Our chatbots are learning more about what we need every day. Can they potentially reduce our loneliness in ways that might prevent us from falling into despair?

A new randomized study published in the Journal of Experimental Social Psychology offers an important perspective. First-year university students were randomly assigned to text daily for two weeks with either a highly supportive “empathy infused” AI chatbot, a randomly assigned human peer, or simply journal about their day. While the chatbot reduced negative emotions in the moment, it did not reduce loneliness over time any more than journaling. Only those who texted with another human became significantly less lonely. The authors concluded that while AI can simulate empathy and provide immediate emotional relief, it does not appear to satisfy the deeper relational needs that reduce loneliness over time. They suggest that AI's greatest value may ultimately lie not in replacing human relationships, but in helping people sustain and strengthen them.

How chatbots and AI can be used to strengthen human relationships and enhance crisis care will be discussed further in Part 2 of this series.

Looking Ahead

In the next blog, I'll explore the many ways AI can strengthen crisis systems—from improving operational efficiencies and expanding access to enhancing quality, coordination, and scale. Those opportunities are real, and we should pursue them aggressively.

But I don't want to undersell my concern that AI could eventually become the fast food of emotional support—available anywhere, anytime, convenient, inexpensive, and seemingly “better than nothing” when someone has nowhere else to turn.

To whatever degree those benefits are true, I don't believe any society would intentionally replace healthy meals with fast food simply because it scales more easily.

Our goal should not be simply to make emotional support more convenient. Our goal should be to make authentic human connection more available. Crisis services themselves do not always achieve that standard today. Sometimes we have become so focused on assessment, documentation, referral, and throughput that we have neglected the very thing people tell us they value most: feeling genuinely cared for by another human being. AI should not tempt us to become less human. It should challenge us to become more human.

As crisis services workforce shortages continue and AI becomes woven into every aspect of our culture, perhaps the question is no longer whether we can scale technology.

The real question is this: How do we utilize this moment and our technologies to better scale authentic human connection?