Representational Frames

What I see in AI music now are the same old tropes of the merging of man and machine. But in context with that, there are lots of other conversations that are going on about the pitfalls of becoming too entranced with machines. Not that it's new--we've been doing it for over a century now since the Futurist movement in the early 20th century, although back then it was more about flight and aviation as the new technology.

Younger generations are first discovering futurism through technologies that are just emerging in the mainstream and using it as a source of emulation. The example I always cite is the influence of piano rolls in the 1930s on composers, namely, Duke Ellington, who, at least anecdotally, used to watch piano rolls and was inspired to write music by watching them. In the 1960s musicians played along with broadcast radio and records. Now we're "playing along" with AI, but we're not actually making anything personally unique. The equivalent in 1978 would be me making records based on the ones I was playing along with and saying they were my creation--even though they might have slightly different music on them--and none that I created.

It's interesting to compare generative AI with generative music 25 years ago, which was more esoteric and arty, which used things like Markov Chains (Sony's Flow Machines), or other randomness generators like putting multiple sound sources out-of-phase (like my track Io (1999), which would come back into phase in 10,000 bars), or Atmosphere Generation, which was a random-shuffle playlist on 3 separate stereo systems. Nobody seems to be interested in these aspects anymore, but they are the roots of generative music--just as blues, jazz, folk, musical theater, and classical were the roots of rock'n'roll.

Now it's just "snapshots" of data.

The snapshot metaphor is useful because its origin is from hunting, and was later used as a metaphor in photography. If we see life as film frames (and it is to some degree, separated by eye-blinks) human vision has a frame rate as well, but rather is drop-frame, or "posterized" time, a term borrowed from video compositing where a 24FPS video can become 12 or 6FPS, or just 1 frame, which would be a snapshot essentially.

In Google's explanation of Search Generative Experience (SGE), it uses this metaphor:

"When appropriate, SGE will show an AI-powered snapshot...snapshots serve as a jumping-off point from which people can explore a wide range of content and perspectives on the web...This allows people to dig deeper and discover a diverse range of content, from publishers, creators, retailers, businesses, and more, and use the information they find to advance their tasks"

So when we're searching, we're actually out hunting for easy prey, which doesn't necessarily mean it's the best prey--and it won't be ready-to-eat, just as a snapshot won't suffice if it's a film you want to make--although can serve as "representational frames"--something film editor Walter Murch used to define the mood of the scene he was working on.