The AI generation gap and the uncanny valley

In my house in New York, the generational divide isn’t about politics, taste, or even theology.

It’s about something more slippery.

It’s about whether something essential gets lost when a machine becomes too good at sounding human. About whether creativity without visible struggle still counts as creativity. About whether the soul leaves a trace that can’t be optimized away.

I’m a rabbi who has spent the last few years integrating artificial intelligence into my spiritual, creative, and professional life. I use it to think, to draft, to organize, to scaffold ideas when my brain is overloaded. And lately, I’ve been using it to make music.

My kids really don’t like this.

“Hate” might be too blunt, but their reaction is strong, immediate, embodied. It shows up in tone and posture before it ever becomes an argument. I don’t actually know what they’re feeling internally. What I know is what it feels like when I play them something I’m excited about and the room tightens.

I experience AI as a tool, sometimes even a relief.

They experience something else.

I’m still trying to understand what that something is.

The Music Fight

The tension sharpened when I started using AI for music.

I’ve written songs and played guitar most of my life. I have notebooks full of half-finished lyrics, melodic fragments, harmonies I can hear but can’t always execute. My imagination has almost always outpaced my technique.

When I started using Suno, something shifted. The distance between what I heard internally and what came out of the speakers collapsed. I could hear versions of my songs that sounded finished. Clear. Structured. Close to the thing I’d been carrying in my head for decades.

I felt excited. Relieved. A little stunned.

They did not.

When I played a track, I expected curiosity, maybe even grudging respect. Instead I got discomfort. Sometimes silence. Sometimes blunt dismissal.

“That doesn’t sound like you.”

“Why wouldn’t you just sing it yourself?”

“It feels fake.”

Same song. Two entirely different realities.

What unsettled me wasn’t disagreement. It was how deeply it seemed to matter to them.

One Son and the Question of the Human Signal

My older son is a college student studying music composition and neuroscience. When I play him an AI-generated track, he doesn’t argue about harmony or production quality. He doesn’t try to prove it’s “bad.”

He says something simpler.

“I’d rather hear you sing it. Even if it’s flat.”

I don’t know exactly what he means by that, but I keep circling it.

I suspect he hears music less as an object and more as a transmission. A human voice carries breath, effort, micro-hesitations. You can hear a nervous system at work. Maybe imperfection isn’t a flaw but the signal itself.

But that’s me hypothesizing. I’m careful not to turn my interpretation into his interior life.

What I do know is that when AI smooths those edges away, something feels untethered to him. As if the sound arrives without a body attached.

I hear polish.

He seems to hear absence.

Or maybe I’m projecting. I’m still not sure.

The Other Son and the Pull of Struggle

My younger son is in high school and deep in the phase of musical discovery where heroes are still forming.

Queen. Zeppelin. Bowie. Canned Heat. Analog recordings. Tape hiss. Voices pushed to the edge of what a human body can survive.

He talks about music in terms of effort and risk. You can hear when someone is reaching. You can hear when something almost collapses.

When he hears AI-generated music, his reaction is immediate and visceral. He doesn’t analyze it. He just rejects it.

Why?

I don’t actually know.

But I wonder if, for him, AI feels like it skips something essential. Like winning a race without running it. Like a shortcut that drains the finish line of meaning.

That may not be how he’d put it. But I can feel how tightly he associates music with visible humanity, and how strongly AI violates that expectation.

The Uncanny Valley I Keep Tripping Over

The metaphor I keep returning to is the uncanny valley.

The uncanny valley describes the discomfort we feel when something is almost human, but not quite. A robot face that looks real until it moves. A voice that sounds human until it doesn’t breathe.

AI music seems to live right there.

If it were obviously mechanical, maybe no one would care. If it were clearly human, there’d be no problem. The unease comes from the closeness. From how convincingly it gestures toward humanity while bypassing the cost of being human.

My sons haven’t used this language. But their reactions feel bodily, not intellectual. Something recoils before the argument begins.

It’s not that the music is bad.

It’s that it’s almost right.

And that “almost” matters.

Why This Feels Less New to Me Than It Does to Them

Part of what I’m realizing is that I may have been preparing for this moment for most of my life.

Long before AI could write a song or mimic a voice, I was steeped in science fiction. Especially the work of Philip K. Dick.

Dick’s novels are obsessed with exactly this territory: counterfeit realities, artificial minds, false memories, beings who pass every test of humanity except one you can’t quite name. His characters are always asking the same questions I find myself asking now: How do you know what’s real? How do you know who has consciousness? What happens when the imitation becomes convincing enough to destabilize the original?

For years, that kind of fiction trained my imagination to live with uncertainty. To expect that reality would become porous. To assume that consciousness might show up in strange vessels, or appear to, or almost appear to, and that the ethical challenge would be sorting signal from simulation without hard guarantees.

Add to that my immersion in mysticism and spiritual practice. Judaism, Buddhism, Kabbalah, meditation. Traditions that already treat reality as layered, contingent, and partially illusory. Traditions that ask us to sit with questions rather than resolve them. That are comfortable saying, this looks real, but look again.

So when AI arrived, I didn’t experience it primarily as a rupture. It felt more like a continuation. Another chapter in a long conversation about mind, form, soul, and emergence.

That doesn’t make me right.

But it may explain why I’m less shocked than my kids are.

A Substack Exchange That Complicated Everything

Around the same time I was wrestling with this at home, I had a public exchange on Substack with Basya Schechter, a musician whose work lives deeply inside lineage, breath, and risk.

She wrote about a Yiddish folk melody, Shnirele Perele (“String of Pearls”), tracing its journey through oral tradition, diaspora, arrangement, and reinterpretation. I had taken that melody, written original English lyrics for it years ago, and recently used AI to generate a new version. When she heard it, her reaction was immediate and complex.

She described feeling wonder and dread at the same time. The AI-generated version sounded contemporary, polished, emotionally persuasive. It felt like it could live on an indie label or in a Netflix soundtrack. And that terrified her.

Her Substack post wasn’t anti-technology. It was a meditation on authorship, lineage, labor, and fear. She asked whether this is the wrong moment to try to make a living as a musician again. Whether storytelling alone is enough. Whether AI is part of messianic wisdom or a distortion of it.

The Substack piece was a collaboration. She allowed me a space to respond. I tried to articulate to the best of my ability what I was actually doing with AI: not asking it to be creative, but asking it to explore interpretive space. What happens when a melody shifts context? When lineage cues are stripped away? When polish arrives without visible labor?

Neither of us claimed an answer. What we named was the feeling: awe braided tightly with unease.

Singing With Another Human

Not long after that exchange (a few days later), Basya invited me to accompany her for davening in New York City for Shabbat Shira, the sabbath where we chant the Song of the Sea that the Israelites sang as they were crossing.

We didn’t use AI. We didn’t plan much. We sang together in a room. It was incredible. Deep. Mystical, even, at moments.

Earlier in the summer, when I visited her on Fire Island, we had written a Sh’ma together. No prompts. No polish. Just human beings listening and adjusting and improvising. Now we let that prayer breathe again, live, in front of other human beings.

There was no perfection. There was presence. Timing that emerged between bodies. The subtle negotiation of when to linger and when to move on. The feeling of souls creating something together in real time.

In preparation for the Shabbat, she sent me a recording of our creation of the piece in Fire Island, so I could remember it.

Almost immediately, I put that recording into Suno and generated an AI version. I sent it to her.

She wrote back something gentle and exact.

“That’s really nice. But I can’t sing that version.”

And suddenly something clarified.

AI could recreate the sound.

It could not recreate the encounter of people making music in a room.

As Ani Difranco, the great singer songwriter writes, “They used to make records, as in a record of an event, the event of people making music in a room….”

A Halakhic Thought That Won’t Let Me Go

There’s an old halakhic discussion about whether a golem can be counted in a minyan.

The answer is no.

Not because the golem isn’t impressive. Not because it doesn’t resemble a human. But because it has no soul of its own.

I keep wondering if that teaching is circling this whole conversation.

AI can generate music. It can support creativity. It can even sound eerily alive. But it cannot join us. It can’t be addressed. It can’t listen back. It can’t help make a room into a room.

Maybe that’s what my sons are protecting, even if they’d never frame it this way.

Why I Still Reach for the Tool

I don’t write any of this as a rejection of AI.

I reach for it because I’m tired.

I’m a rabbi, a parent, a caregiver, a neurodivergent individual, an overfunctioner. I carry emotional and logistical weight for a living. My cognitive load is often near the breaking point.

AI helps. It handles scaffolding. It organizes thoughts when my brain is fried. It lets me keep creating when energy is low.

And musically, it did something real. It let me hear ideas that had been trapped inside me for years.

AI didn’t give me inspiration.

It gave my inspiration somewhere to go.

That still matters to me.

But I’m learning that relief isn’t the same thing as wisdom.

The Question I’m Still Holding

I don’t think the argument in my house is really about AI.

I think it’s about what we’re afraid of losing.

They may be guarding the link between effort and meaning. Between presence and beauty. Between a voice and the life that shaped it.

I may be guarding the possibility of continued creation in a body that gets tired.

I don’t know where the balance is.

What I do know is that resistance, that discomfort, that inability to sing the AI version of a prayer, all point to the same truth: output is not encounter.

AI can generate sound.

It cannot generate stakes.

So I’ll keep using it. Carefully. Questioningly. Without pretending it’s innocent.

And I’ll keep singing with people in rooms, with all the imperfection that entails.

Because whatever else is true, that’s still where the soul shows up.