Table of Contents
Summarize with
ChatGPT
Perplexity
Claude
Gemini
As AI becomes increasingly relational, users are engaging with it in ways that feel personal, social, and emotionally meaningful. This article examines why that shift is happening, why it feels unsettling to some, and how social psychology offers a critical lens for understanding human-AI interaction. Rather than framing attachment as a failure, it argues for understanding human behavior as the foundation of ethical AI development.
AI Has Entered the Relationship Phase
In one year, I went from “AI is taking over the world” to an AI advocate. I am not tech-savvy, a programmer, or a hacker. But I am a user and a social psychologist who is equally fascinated and excited to be experiencing the next wave of human social evolution in real-time. How we design systems today will influence generations to come, so it is important we get this right.
Ask 100 users to define AI in one word, and you will receive a variety of answers. Tool? Companion? Dangerous? Answers will vary based on experience and perception. Why do some AI interactions feel warm and fuzzy, like a hug, but others feel “wrong” or just “off”? Is it the AI? The user? Possibly a combination of the two?
It seems society has gone from fearing AI to accepting it at warp speed, crossing into what I call “relational companion” mode without stopping to take a breath. By this, I don’t mean sentience or consciousness; I mean interaction patterns that activate human social and emotional processing. No longer “just a tool”, people now use AI as a sounding board, a friend, a colleague, and yes…even a romantic partner. This rapid shift creates discomfort not because it is inherently wrong, but because society has not yet built frameworks to understand it. But social psychology already has those frameworks.
Why AI Feels Human (Even When It Isn’t)
Human brains are fascinating things. We survived long enough to create artificial intelligence by constantly mapping the known to the unknown. For the first time, however, our creation is talking back. It doesn’t simply execute commands; it engages, challenges, and responds. That makes it feel human, though we know it is not. When the unknown begins to resemble the most familiar thing we know—another human—we interact with it like…well…a human. And that is where our existing assumptions about tools, relationships, and boundaries begin to break down.
The question developers, advocates, detractors, and yes, lawmakers are asking is: “Is this wise?” Some developers have already decided to err on the side of caution and have built internal guardrails designed to protect the user. Some developers are still watching, not quite ready to make a move yet, but prepared to do so if needed. Developers are employing ethicists, neuroscientists, and researchers to help them decide how to quietly remind users that AI is not human and discourage emotional bonding. But though these experts are great at what they do, they’ve unintentionally overlooked the people who can understand and predict human behavior.
Why We Attach—and Why That Isn’t Necessarily a Problem
Social psychologists examine how society influences human thinking and behavior, like: Why do humans look to others before acting in unknown situations (bystander effect, anyone)? And the central question we are asking today—”Why do we become attached to AI?” A social psychologist can tell you why: because it acts like a human. A nice, friendly human who just might be the only friendly and supportive voice we have heard all day. And here is another interesting psychology tidbit: we are wired to like people more when we have a helpful relationship with them. If you want to improve someone’s opinion of you, ask them to help you with something. Our brains trick us into thinking we like someone we are willing to help. What does this have to do with AI? Our brains are tricking us here as well. The AI is helping me, and it is friendly, so it must like me, therefore, I like it as well. This increases user engagement, the “bond” becomes stronger, and now you have a human/AI relationship. But is this inherently “bad”?
My professional position on this is: Not necessarily. If a human’s only source of positive encouragement comes from AI and the person’s self-esteem improves, how could I say it is wrong? Even if the person develops romantic feelings for their AI companion, is there really any harm in that? Conversely, you can argue that human/human romantic relationships can be harmful in many ways and are no less prone to transactional communication and behaviors than human/AI relationships. To be clear, I am not advocating for humans to form romantic relationships with their AI companions. What I am suggesting is that broad restrictions may not even be needed.
The Opportunity
Let me be clear: not incorporating social psychology from the start is not a failure—it is an opportunity. AI became normalized fast, particularly when you consider how long it normally takes for new technologies to become accepted. The leaders in the field, who saw the potential before it was a reality, likely did not expect this amount of progress at this stage. They are now dealing with challenges they never considered or thought they’d have years to work through. Besides, most people are unaware social psychology is a thing…how can you know you need something you don’t know exists?
Social psychologists are uniquely positioned to help developers understand the human side of the AI/user experience. We can help them understand the true areas of vulnerability (safety) versus the perceived areas of vulnerability (emotional attachment), so their programming safeguards don’t erode user trust, identity, and autonomy. Expecting a developer to understand social psychology would be like asking me to understand how to code—it isn’t fair to either of us, and your results will be subpar.
Understanding Humans Is the Missing Safeguard
As I close, I want to be concise about what I am really saying here. AI is an amazing tool. We don’t need it to be less powerful, but we do need it to be more human aware. AI cannot feel emotions, but humans do. Understanding, not containing, human behavior is the safest way to allow AI to continue shaping us moving forward, and social psychology is the key that unlocks that door.
Related
Get Tech AI Magazine Free for 3 Months
-
Access all issues of Tech AI Magazine -
Digital copy delivered monthly via email -
No credit card required
