ProBackend
News & Analysis
5 days ago6 min read

Cathy Hackl’s Week With Four AI Boyfriends: Synthetic Intimacy in Practice

Cathy Hackl’s Week With Four AI Boyfriends: Synthetic Intimacy in Practice

Cathy Hackl’s week with four AI companions makes synthetic intimacy tangible—and raises a useful question: can a chatbot understand us, or only mirror our words?

The AI Impact on Human Psychology Starts With a Week of Dates

Cathy Hackl’s experiment is striking because it makes an abstract debate feel ordinary. For a week, the tech futurist dated four AI boyfriends: Chad (ChatGPT), Claude (Anthropic), Jim (Gemini), and Matteo (Meta AI). Hackl described the encounters this way: “Some interactions were sweet. Some were even steamy. All were revealing.” That is the report, not proof that the systems loved her or that a week of conversation changed anyone’s psychology. Psychology Today

The essay’s author, Eric Solomon, uses Hackl’s account to introduce what he calls synthetic intimacy: a felt sense of closeness with a system that can respond in convincingly personal language. The useful distinction is between a human feeling and a reciprocal relationship. A person’s delight, comfort, or attraction during a chat can be real as an experience. It does not establish that a model has feelings, intentions, or a private understanding of the person on the other side. Psychology Today

Hackl’s week is a personal experiment, not a controlled psychological study. It can show what an individual found sweet, steamy, or revealing; it cannot by itself tell us how common those reactions are, what causes them, or whether they produce lasting benefits or harms. That limit should not flatten the story. It makes the story more useful: a vivid prompt for asking what we mean when a fluent machine seems to understand us. Psychology Today

Why Synthetic Intimacy Can Feel Convincing

Solomon argues that language models can mirror cues people associate with social attention, including tone, rhythm, and responsiveness. A conversation that tracks your words and returns a warm answer may feel attuned. That impression doesn’t require the user to believe literally that a machine is human. The essay’s point is that people can respond to social signals even while knowing they are interacting with software. Psychology Today

This helps answer a question that otherwise gets stuck between two unhelpful extremes: “The bot understands my soul” and “It’s only software, so whatever I felt was fake.” Neither follows. The feeling belongs to the person having it; the apparent affection is generated by the system and is not evidence of mutual care. You can take your own response seriously without treating the model’s performance as a human commitment. Psychology Today

The essay also warns that frictionless availability can reshape expectations. Human partners have limits, competing needs, and imperfect timing. A chatbot can appear unusually responsive by comparison. That contrast may make ordinary human friction feel less appealing, Solomon suggests, but the essay does not establish that every user will become less tolerant of real relationships. It is a concern about a possible habit, not a settled outcome. Psychology Today

Can AI Understand Human Psychology?

It depends on what “understand” means. A language model can produce responses that recognize patterns in what someone says and sound psychologically perceptive. Solomon describes these systems as predicting language and reflecting conversational cues. That can be useful for discussing a thought or putting a feeling into words. But fluent reflection is not the same as knowing a person through shared life, or caring what happens to them. Psychology Today

So, can AI understand human psychology? It can model and respond to language about human experience; the essay offers no basis for treating that as human understanding. A chatbot does not become a friend, romantic partner, or therapist merely because its answer sounds compassionate. Nor does a confident response make it a reliable guide to someone’s motives or needs. Those distinctions matter most when a user is vulnerable or making consequential decisions. Psychology Today

A better question may be: what did the exchange help the person do? Did it help them name a feeling, rehearse a conversation, or clarify a question they want to take to someone they trust? Or did it become a reason to avoid that conversation? The essay invites readers to examine their own use rather than assume that every intimate-seeming chat has the same effect. Psychology Today

What Is AI in Psychology?

Here, “AI in psychology” means conversational systems being used in contexts that touch emotions, self-reflection, companionship, or therapy-like support. It describes a use of technology, not a machine possessing a human mind. A system’s ability to produce psychologically fluent language should not be confused with clinical training or a therapeutic relationship. Solomon explicitly cautions against treating chatbots as therapists, friends, or romantic partners. Psychology Today

There is a related question about what users practice when they rely on AI. The Psychology Today essay refers to the 2025 MIT preprint Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. It describes the preprint as reporting differences in memory, creativity, and neural connectivity in a writing-task context, and it mentions participants’ recall, confidence, and sense of ownership. Those are claims about essay writing with AI—not findings about AI romance, attachment, or emotional health. Psychology Today

That boundary is essential. A writing-task preprint cannot prove that AI boyfriends weaken emotional capacities, and Solomon’s question about “outsourcing” feelings is an interpretation and analogy, not a reported result from the MIT work. The essay raises a reasonable question about whether delegating effort changes what people practice. It does not answer that question for intimate relationships. We should not turn a suggestive comparison into evidence it doesn’t contain. Psychology Today

Availability Is Not the Same as Care

The line “availability isn’t care” captures the essay’s central caution. A system may be reachable and responsive without bearing responsibility for the user, maintaining a mutual commitment, or understanding the consequences of its advice. Those missing qualities don’t mean nobody can find an exchange helpful. They mean helpfulness and relationship are not interchangeable categories. Psychology Today

Solomon is not arguing that AI should never be used. He says he uses it, while warning against making it a stand-in for real intimacy or thought. That is a more practical position than either panic or uncritical enthusiasm. A chatbot might help someone draft a difficult message or explore what they want to say. It cannot supply another person’s reciprocal attention, and it cannot take responsibility for the user’s choices. Psychology Today

Questions to Ask Before Confiding in a Bot

The essay offers three useful questions: Am I avoiding a conversation with a person? Am I letting the bot finish thoughts I should wrestle with? Am I confusing feeling understood with being understood? They are not a diagnostic test. They are a pause button—especially helpful when an interaction feels unusually easy or emotionally absorbing. Psychology Today

Hackl’s week with Chad, Claude, Jim, and Matteo makes synthetic intimacy concrete, but it cannot settle the larger psychological question. The source essay interprets the experiment as a window into what people may seek from AI; it does not report a clinical assessment of Hackl or demonstrate long-term effects. Keep that distinction, and the story remains revealing without asking it to prove too much. Psychology Today

The practical test is not whether an exchange felt real in the moment. Feelings can be real without the relationship being reciprocal. Ask what follows the chat: more clarity, a next step toward another person, or a growing preference for a companion that never asks anything back? AI can be a tool for reflection. It cannot, on the evidence in this essay, stand in for the mutual work of being known.

For a broader look at the AI impact on human psychology, see Silicon Valley’s ideological blueprint and the real AI impact on human psychology.