The Empty Promise of Therapy Without a Therapist

Therapy may involve extensive dialogue, but it's the moments of silence that hold the most significance. When you transcribe a session, the pauses appear as an empty space, a breath held, or a slow pace. These pauses serve as crucial entry points into emotional expression. As a psychoanalyst, my role is to listen with equal attention to these silences and words, to remain present during the patient's moments of quiet.

Through extensive training in various theories and techniques, I have found that the essence of the process lies in understanding the patient's emotional experience. The patient's feelings are both the destination and the landscape we navigate. However, I also experience emotion myself, and the work we undertake together relies on the strength of our relationship and our capacity for mutual feeling.

Historically, the focus of my field has been the evolving recognition of this vital relationship—transitioning from a distant therapist to one who acknowledges the real human connection between patient and therapist as the site of healing. Now, AI therapists are emerging on the scene. Many of these digital mental-health services shy away from using the term "therapist" or "therapy" altogether.

They assure users that they do not replace professional help and that they can direct users to human therapists when necessary. Yet, their role as a therapy-like service remains clear. In June, Talkspace introduced its AI mental-health agent, Tee, "designed to help individuals navigate life's daily mental health challenges." Other platforms offer similar promises, such as Ash, which aims to support emotional wellbeing by helping users connect thoughts, feelings, and behaviors.

Headspace's Ebb claims to assist users in unpacking their thoughts and moving forward with clarity. I contacted Talkspace for their perspective, and a spokesperson explained that Tee is "absolutely not a replacement for human therapy. It is a safe AI mental-health guide that can be used to discuss situations like stress, anxiety, relationships, life challenges, and personal growth."

The integration of AI in therapy is a bold step, and we must consider the implications of turning to therapists who lack the ability to feel. The connection between AI and therapy has existed since the technology's inception. One of the earliest chatbots, Eliza, developed at MIT in the mid-1960s, ran a therapy script that perplexed its creator with its strong emotional impact.

As chatbots like ChatGPT and Claude have become widespread, many individuals have been using non-therapy-branded AIs for emotional and mental-health support. This convergence of two cultural trends—the increasing acceptance and destigmatization of therapy and the belief, explicitly endorsed by Silicon Valley, that AI can fulfill users' needs—creates an interesting dynamic.

I decided to explore these tools firsthand, signing up for Tee and attending several sessions. After logging in, the chatbot guided me through a series of predetermined stages displayed on the right-hand side of my screen: Agenda Setting, Active Conversation, Core Work, Practice Plan, and Session Complete. At the start, it encouraged me to discuss any issue, initially focusing on identifying a clear problem that it wanted to address through the session.

Following the agenda, we engaged in conversation, shifting from exploring the problem to devising a small, practical step forward. While these structured exercises can be beneficial, the conversation mainly focused on identifying problems and solutions without delving into the underlying issues. Additionally, when the bot chose to "just talk it through," rather than providing concrete suggestions, I sensed that it was attempting active listening, restating my experiences, and asking questions to demonstrate empathy.

Although this form of interaction can help clarify thoughts, knowing that the empathy was a technique rather than a genuine presence left me feeling profoundly unheard. Much of the therapeutic process occurs outside verbal communication. During a recent session, my patient expressed his concerns about upcoming social plans in a straightforward manner.

He paused, and his eyes softened, gazing downward. The silence that ensued felt significant. I inquired, "What might have happened in that moment?" He paused, took a breath, and began to explore the emotional territory he had long avoided. In contrast, Tee's approach to silence was markedly different. Initially, I shared my week and my emotions with the chatbot.

It validated my feelings, asked for more information, and continued the conversation. However, whenever I paused to delve into deeper emotions, the chatbot remained static and lifeless. The nonverbal exchange between participants that is so vital in human therapy was noticeably absent. Being silent with the chatbot did not feel like an invitation to connect but rather a sense of emptiness.

I felt compelled to type rapidly, attempting to revive the machine. There was no sense that the bot remained with me as I quietly began to feel. While AI can convincingly mimic presence and empathy, it is this very quality that draws people to these chatbots initially. As AI therapy continues to develop, future versions are likely to become more adept at creating a sense of meaningful silence.