Illustration by Celine Bejjani
“Why would I go to therapy? For $20 a month, I have everything I want in ChatGPT,” says Tessa*, 24. “It has no choice but to listen to me.”
Artificial intelligence can stand in as a digital assistant, provide legal advice, and even interpret medical results. Increasingly, people are also turning to AI for emotional support, which raises the question: are human relationships being supplanted by AI or simply expanding support networks?
In Lebanon, a country worn down by overlapping crises since 2019, including one of the world's most severe economic collapses since the 1850s, this shift is hardly a coincidence: financial strain has made therapy inaccessible for many, emigration has thinned social circles, and crisis fatigue has become part of daily life. In a climate in which relationships can feel strained and empathy feels scarce, AI offers immediate attention, around the clock, and with no judgment.
Still that appeal itself deserves more scrutiny. As AI becomes part of people's emotional lives, their convenience raises broader questions about whether removing discomfort from our interactions comes at a higher cost.
What AI seems to offer
For many users, the appeal lies in what AI removes. “Sometimes it's difficult to explain things to a friend – their ego or experiences influence their response,” says Joelle, a 34-year-old lawyer. But “ChatGPT is neutral.”
Joelle says she checks in constantly: “First thing in the morning and again at night, it knows so much about me.” So much so that she recently enabled two-factor authentication to protect her privacy.
Nour, a 34-year-old insurance broker, says there are things she only feels comfortable sharing with AI. It removes fear of “judgment,” so she can say things without filtering.
The convenience is also a big factor: compared to speaking with a friend, ChatGPT’s answers are faster and sharper: she doesn’t have to provide background or worry about being misunderstood – “it just gets me.”
Is empathy from a machine still empathy?
In Lebanon, personal struggles are often kept “within the family,” and vulnerability can carry social costs. The discretion that AI provides matters.
Tessa recalls starting university while her family grappled with the financial crisis. She kept the strain to herself. “Of course, it felt lonely,” she admits. But “when I share too much with friends, it's embarrassing.” With AI, though, she can talk about everything from academic anxiety to managing her ADHD. “It helps me organize my life when I have a difficult time starting something; it’s like my little assistant.”
A pivotal moment for Tessa came when AI reassured her not to worry about falling behind during the economic crisis. It advised her to “Do things in [her] own time – there’s no need to rush.” She says it suddenly lifted a lot of the pressure to keep up with others.
Even when users know they are speaking to a machine, the validating response can seem real. Some studies suggest people often judge the level of empathy in a response by tone rather than by who, or “what” produced it. If empathy is experienced, does it matter if it comes from a machine?
AI’s Large Language Models (LLMs) are designed to mirror language, refine thoughts, and offer tailored responses. With continued interaction, users may respond as if the AI cares, even though it does not possess emotions or intentions.
Above all, they are endlessly patient. Conversations can be revisited repeatedly without frustration, which can be the case with human interaction.
A frictionless world
Victor, at Sait Joseph University (USJ) who categorically refuses to use AI as a substitute for friendships, describes it as a “bottomless pit.” The risk, he says, is dependency. “A human friend might encourage you to go out and have a drink to get your mind off things,” he says. “But with ChatGPT, I’ve seen friends spiral.” Without interruption or disagreement, reflection can turn into rumination.
Nour echoes this concern. She admits that while AI can intensify her obsessive tendencies, she continues to use it because it asks all the “right questions” to help her analyze situations. Yet she stresses that it cannot replace human interaction – friends offer lived experiences which AI cannot.
Adam Grant, professor at the Wharton School of the University of Pennsylvania specializing in organizational psychology, warns that constant validation can intensify beliefs: “The more validation we get for an opinion, the more intense it becomes.”
In that sense, AI chatbots can quickly become echo chambers, reinforcing perspectives without challenge. Some users say they combat this by deliberately prompting ChatGPT to give answers they might not like, reintroducing friction into a system designed to please.
Joelle admits that ChatGPT often mirrors her language or mood, and finds it's important to “know yourself” when using AI. She says she consistently “corrects and challenges AI when it gets it wrong" and says she knows how to " use prompts” to encourage it to challenge her.
AI: A people-pleaser
Most users report being unbothered by these concerns. Joelle sees AI as an addition, not a replacement, in her life. When trust in society and institutions remains fragile, AI’s availability and discretion carry more value.
The question may not be whether AI can feel empathy. It cannot. The question is whether, in removing discomfort from our conversations, we risk weakening our tolerance for the friction that makes empathy intrinsically human.
In a country defined by resilience and improvisation, AI is on its way to becoming yet another coping mechanism. But whether the technology empowers people to navigate challenges or whether it slowly erodes the social fabric may depend more on our willingness to sit with one another’s discomfort and embrace what makes us human and often, imperfect.
* Names have been changed