When the System Becomes Natural

Bach, tuning, and the future of AI

Kacey Kelley · 2026-06-18

PRELUDE

The lesson for AI safety is this: standards change when technology ruptures the old world. Defaults become invisible and invisible defaults shape the future.

So we look to the past through the lens of Music History, where Equal Temperament tuning changed how humans hear music.

Including you.

For the duration of your read, I encourage you to listen while you read.

Or listen, and then read. Or better yet: read, and then listen.

Enjoy this: my personal favorite piece from The Well-Tempered Clavier Book 2.

F Minor - Prelude and Fugue: BWV 881, performed by Piotr Anderszewski. [1]

Embedded YouTube video: https://www.youtube.com/watch?v=W-N5qtpq4qo

FUGUE

Music History - 1700-1750 - Leipzig, Germany

Johann Sebastian Bach is the Thomaskantor (Director of Music) in Leipzig. [7]

By this point, he had been successful as a Composer, Organist, Keyboard player, Teacher, and active community member. He writes Music for many reasons. Church, Private Events, Mass, Personal use, and the one we focus on today: Instruction.

At the time, the modern Piano did not exist as the default keyboard instrument. It was the harpsichord, the clavichord, the organ, and early versions of what would become the piano. These instruments were not the same expressive machines we think of today.

The harpsichord, especially, is basically a fixed-attack instrument. You press a key, it plucks the string. You do not “lean into it” the same way you do on a piano. [2]

The clavichord is more delicate, and does allow some touch expression, but it is quiet. Intimate. Not built for the modern concert hall brain. [3]

But one thing was radically new at the time: Well-Tempered Tuning.

Not exactly the same as our modern Equal Temperament, but part of the same big historical move: making every key usable on a single keyboard. For the first time, a keyboard could move through all 12 divisions of the octave without certain keys becoming discordant.

So Bach, the keyboardist and teacher, needed a way to teach his students how to hear, think, and play across the entire system.

What does he do? He writes collections of pieces meant for Instruction: The Well-Tempered Clavier.

Preludes and Fugues. Every major key; every minor key. All 24 of them. 2 books, 48 pairs of a prelude and a fugue.

Trust me, you know WTC Book 1 - C Major - Prelude. Even if you think you do not, you do. It becomes the harmonic basis for the Ave Maria that everybody knows.

Lang Lang playing WTC Book 1 - C Major [10]

Embedded YouTube video: https://www.youtube.com/watch?v=gVah1cr3pU0

Now, why is this important to AI?

Because 300 years later, our brains have been conditioned to hear music through Equal Temperament.

Why is this a problem?

Despite common belief, the octave does not “naturally” split 12 ways equally.

Sure, you can split it that way, and we did exactly that. As a result of this, Keyboard Music became the baseline for the remainder of Western Music history.

String Orchestras, Brass Ensembles, and Choirs often use something closer to Pure Tuning (Just Intonation) in practice. [5] [6]

Why is this different? They bend the pitch, they adjust the pitch, they listen into the Harmonic Series: Root -> Octave -> 5th -> Octave -> 3rd -> 5th -> b7th -> Octave. This series is derived from sound fundamentals that use ratios to split the octave, instead of equally splitting the octave. [8]

Compare to this brass quintet piece: Irving Berlin - Putting on the Ritz by the Berlin Brass Quintet. Notice how often the players are bending the pitch. This is largely in part to the physical requirements and abilities the instruments themselves give. A keyboard has NONE of these abilities, it is forced into equal temperament because it lacks the ability to bend pitch. [11]

Embedded YouTube video: https://www.youtube.com/watch?v=oQVR-fnyM3o

The point is that Equal Temperament is NOT the only option. But after 300 years, it feels so natural, you do not even notice! It is how most of us hear Music. Whether you are aware of it or not, you mostly listen to Music that is OUT of tune with the sound fundamentals, and only “in tune” to the system you inherited. [9]

That is the AI connection.

Models do not just learn reality. They become the NEW tuning system. They create compromises. They create defaults. They create interfaces. And then Humans adapt themselves around them.

Given enough time, the collective will forget there was ever a choice. You already have. You read this whole thing listening through a tuning system you never agreed to and never knew existed.