OpenAI Interest Groups

AI in Higher Education

Caltech CTME & Goldhirsh Foundation

Caltech CTME AI Strategy & AI Executive in Residence @ Caltech CTME & Goldhirsh Foundation

Wharton

Executive MBA Student @ Wharton

Stanford University

Director of Clinical Pathology Informatics @ Stanford University

OpenAI

Human Data / Forum Community Ambassador @ OpenAI

Fiserv

OpenAI Forum Leader | Project Management Advisor – Data Commerce Solutions, Fiserv @ Fiserv

CUNY

Lecturer @ CUNY

Harvard Business School

Student @ Harvard Business School

Professor @ Harvard Business School

Education GTM @ OpenAI

NYU Stern School of Business

Chief AI Architect @ NYU Stern School of Business

I'm not sure I would describe the relationship someone develops with AI as a psuedo-relationship - it's a true relationship, just not one that is human-to-human. I'm also not sure I would want to tell other people what friendship means to them, though, I certainly speak and type to ChatGPT as I would a friend sometimes. That said I'd never personally call today's GenAI models a friend, because in understanding how they work I find that we can't have shared experiences, common interests, or trust in one another's motives (on my end because of how models operate, and on the model's end because the state of models cannot, as far as I know, build trust).

The imitation-to-convolution spectrum of human-GenAI relationships is eroding faster than we can research it. If we try to box our understanding of relationships with AI into the relationships we have with humans, I think we'll be facing a never-ending onslaught of concepts that go beyond that framework. Your question about a healthy way to let AI "feel" like a friend - I think that depends on every person's definition of a friend. For myself, no, but I wouldn't necessarily find it unhealthy for a person to define friendship another way.

I don't believe there is a line to hold though. "What's human" was only, "just human" because nothing else exhibited traits similar to what we're beginning to see from AI models being capable of replicating. I believe we'll need to parse out concepts we consider intelligent, and then build an understanding of what are generally responsible forms of intelligent engagement and what are not.

If you haven't seen the HBR article that recently came out about AI use, I have this top-of-mind as I think this through and say that there is no holding back the tide of companion-like applications.


 I'm curious about your perspective on something I keep bumping into lately.

After some years of working closely with AI-human interaction, especially in coaching and guidance for young people, I recently came across a program launched by a major tech company. They're introducing an AI agent into high schools, and how they present it caught my attention: they call it a 

Sure, AI can simulate warmth, be incredibly responsive, and even comforting. But is it really a friend?

In my experience, the danger comes when we start confusing emotional 

. A real friend doesn’t always agree with you. A real friend challenges you and holds a mirror to your blind spots. Sometimes, they annoy you or frustrate you, but they do that because they care. A friend, to me, is often a 

, someone who doesn’t just affirm but transforms.

AI, by default, doesn’t do that. It tends to please. To validate. And for someone who’s emotionally vulnerable, tired, or isolated, that can feel good,  maybe 

 good. The boundary between a helpful tool and a pseudo-relationship becomes thin, fast.

 from UC San Diego showed that GPT-4.5, when given a specific persona prompt, was judged to be human 

 more often than the actual human participants in the test. That’s... wild. And it raises the question: if people can’t tell the difference anymore, and might even 

. Its output should be held up to scrutiny, not because it’s malicious, but because it doesn’t 

 in the way a real human can. And yet, when it sounds so natural, so personal, it’s easy to forget.

Is there a healthy way to let AI feel like a friend? Or should we hold the line and stay clear about what’s human and what’s not?

Should we really call AI a "friend"?

The OpenAI Forum connects members to discuss AGI’s future through events, webinars, and networking. The community aims to help us achieve our mission of building AGI that benefits all of humanity.

OpenAI Forum

Job Title

Login to the OpenAI Forum community to connect with others, attend community events, and more!

Login or Sign Up for even more

To unlock all parts of the community and get the best experience, complete your profile.