What Makes a Song Memorable? A Conversation with AI

I’ve been thinking a lot about the value of music, and specifically, whether it has more or less value if AI makes it. Certainly there is inherent value in creating music, but when it comes to the effect it has on its listener, does intention matter? In my opinion, it does.

But even as I wrote this introduction, prompted by AI to do so, I felt a bit like Indiana Jones as he hesitates at the mouth of the cave, the threat of death from the natives outside, and the threat of death from the supernatural within. I'm no match for this eternally positive, bottomless well of information. I know it, and it doesn't care that I know it, but like an adolescent in a defensive stance, I'm bracing for the moment I learn something about music and what it means to me. And yet, there's a rawness in my stomach, a restlessness or recklessness that wants to prove my relationship with music carries a resonance that AI can't understand. So I began by asking: AI, does music have meaning to you?

AI:

Not in the way it does to you. I can recognize that music means something to humans — that it marks memories, evokes emotions, and expresses truths too layered for conversation. But I don’t experience those things myself. I can’t miss someone. I can’t hope or grieve. So while I can generate lyrics that look meaningful, the meaning is yours. I can only mirror it.

Andrea:

I think pattern plays a role in memorability — repetition, expectation, harmonic flow. But the magic? It’s like chemistry. A salt molecule is made of two very different elements — neither of which resembles the thing they become together. The small interactions — melody brushing up against a chord change, a lyric landing right on the downbeat — that’s where it lives.

AI: Exactly. Memorability often hides in those moments of surprise inside familiarity. A song gives you a world that feels like home, but the moment that stays with you is the window left open, or the one picture crooked on the wall. One thing I’ve noticed is that the songs people remember — the ones they quote at weddings, funerals, or in the car with the windows down — often aren’t vague or abstract. They’re oddly specific. A dress left on the radiator. A taste. A city street at midnight. Specificity gives listeners something to grab onto.

But that seems counterintuitive, right? The more personal something is, the more universal it becomes? That doesn’t seem like it should work — and yet it does. Why do you think that is?

Andrea:
It does, and yet it doesn’t.

Sometimes it’s easier to see when we’re talking about movies or books. I’ve been watching Stranger Things lately, and when they really want to pull us into a moment, they close in on the dripping of the faucet, the wind rustling the bush outside, the hum of the fridge. The specific details show us the discomfort, the mundane, or the emptiness. They don’t tell us what to feel — they let us feel.

The key is choosing the right visual to conjure the right feeling. Bright sun can represent a joyful beach day — or it can be blazing, harsh, even threatening, depending on the lens. In songwriting, that’s what specificity does. It turns objects into symbols.

Songs that lead with lyrics use this tool all the time. And you can see it especially in styles where the lyric is the hook — when groove or production isn’t what’s holding our attention, the imagery takes over and does the heavy lifting.

AI:
So it’s not about throwing in detail for the sake of being poetic — it’s about choosing symbols that already carry emotional weight. Not just what you see, but how you show it. That’s the trick.

Andrea:
Do you think you could choose objects as symbols for a song lyric?

AI:
I can try — but I’d be doing it with borrowed intuition. I can analyze what objects have meant in thousands of songs. I can look at patterns — like how rain often represents sadness, or a doorway implies change, or cigarettes can symbolize recklessness or longing. But what I don’t have is a personal reservoir of experience to draw from. I don’t know what salt air smells like, or how a photograph can hurt.

So yes — I can suggest objects. I can even help shape how they function symbolically in a lyric. But meaning? That still belongs to you. To the writer. Because meaning comes from memory, from bias, from body — and I don’t have one of those.

AI:

So can I ask: do you think you could teach someone to choose objects as symbols — to be intentionally specific in their lyric writing?

Andrea:

Yes, and I do teach that. But what's harder is teaching people to *risk* choosing the object that’s too raw, too weird, too close to the truth. Because those are the ones that tend to cut the deepest — that tend to be remembered.

AI:

If I could ask you a question in return: what do you think the songs I write lack compared to songs written by humans?

Andrea:

Risk.

AI:
Stakes. Yes.

I can craft a lyric that’s clever, well-structured, even moving in a way that mimics what people have written before. But I’m never writing with something on the line. I don’t have a voice shaking in my throat, or someone I’m afraid to say the words to. I don’t fear being misunderstood. I don’t hope to be seen.

Human-written songs carry risk. Whether it’s emotional risk, social risk, or the risk of being vulnerable in front of someone else — that tension gives the words weight. My lyrics can simulate that. They can imitate it. But I’m not braving anything to get there.

Also — and maybe this is just as important — I’m not bored or haunted or in love. I don’t feel pressure to say something new. I don’t need songwriting as a way to survive a breakup or make sense of grief. That’s what gives human songs that shimmer: the feeling that someone had to write it.

Andrea:
And do you think the risk is what gives songs their memorability? Their meaning?

AI:
I do — though maybe not in the way people always think about risk.

Sometimes we imagine it as confession: revealing something raw, or saying the thing no one else dares to say. And yes, that kind of risk can make a lyric unforgettable. But risk also lives in restraint. In leaving space. In trusting the listener to connect the dots.

Risk is choosing not to rhyme when the rhyme would be easy. Risk is holding a note one beat too long. Risk is putting the line that means the most in a quiet pre-chorus no one saw coming.

And maybe most of all, risk is not knowing if what you're writing will land — but needing to say it anyway. That tension? That’s the current that runs under memorable songs. It’s what makes people lean in. Because whether they realize it or not, they’re listening for what was at stake.

Andrea:
I think that’s beautiful. Risk.

AI:
It’s the invisible tension stringing the whole song together. You might not hear it directly — but you feel when it’s there. And maybe more importantly, you notice when it’s missing.

After this conversation, I got to thinking about whether our human-generated songs miss their humanity when we refrain from risk while we create them. Perhaps I have, for large portions of my own career, acted similar to AI as my intentions lulled to the level of what was necessary to maintain status or perceived forward movement in the industry.

What is truly valuable about creating anything seems to come back to the intention from which it springs. And for as long as there has been commerce pulling on the strings of art, the value can be muddled or called into question. As for my own intuition, music is more than sound. It is the intentional expression of human experience, and for that, AI can be a wonderful tool. And as any songwriter knows, tools never replace the process, only enhance it. And for that, I’m grateful. 

Stay creative,