Decorative Arts II, as the title suggests, is a record of new aesthetics: specifically those
produced by a relatively recently emerged technological ability to make sound from a deluge of
data. Even more specifically, this is a record of sounds produced using– and more often
misusing– a number of AI tools (ranging from local models to large language models). These
are aesthetics of noise, ghostly aural fragment, artifact, distortion, metabolized .wav and
non-intentional psychoacoustics.
So, while I take the music to be moving, to be a space in which one's attention might open up,
this is also, without reservation, a concept album. As with the last Decorative Arts record, this is
an exploration of genre. Ambient music sits in an odd space between the experimental and the
utile, the technical and the emotional. It can function as background sound to work,
consumption, and lifestyle, or it can offer something to the spirit. It does so by spare, often
liminal gestures and, in some cases, by testing what timbres and intervals can be introduced
into a sonic space, to elevate without breaking its transporting spell. This second assay at
ambient continues the exploration of the first– walking the line between the aesthetic and the
aesthetically challenging. Through the use of AI, it also introduces questions about authorship,
intention, the aleatoric, and the deepening collaboration between musicians and the technology
with which they work.
the endcore
The political brand of AI can only be said to be declining on the left– if it has not already totally
crashed. The rise of AI tech oligarchy, techno feudal data enclosure, and the environmental
impact of datacentres– the energy, precious metals, water, and land they demand– make strong
grounds for political arguments against this tech. To say nothing of the endgame of corporations
like Spotify who have already done so much to devastate music as a remunerative, and
therefore plausible form of work. What begins as infinity-core– access to all music, all the time
for one small fee– ends as Spotify-core: music made for algorithms, vibes, tagged and
quantified life moments.
Spotify is not only extractive of rent, but too, of user data. They are in the marketing business,
not the music business. As they serve a profit motive, the end-core will mean not having to split
royalties– neither for the vibes nor the soundtrack, let alone the everyday magic of music.
Businesses like Spotify are interested in AI because a well-made Large Language Model (LLM)
might be in the position of replacing musicians altogether. Which means no artists to pay. No
producers, no engineers, no labels. And who better to provide an absolutely complete data set
for training such an LLM, but a corporation like Spotify, to whom artists obligingly submit their
work in order just to have a crack at the game of quantitative aesthetics central to the erstwhile
platform orthodoxy.
all your days a dataset
The obvious question, then, in light of these scarcely defensible impacts, is: why fuck with these
models? I am not a purist. Nice gig for those who can hack it, but I am not such a one. And for
my part, I believe purity politics have played a role in allowing us to slip so far under in the
current political undertow. I believe the bell has been rung. We were always already cyborgs,
and we were always already training the models. AI is with us, and it is not going anywhere.
The lighter argument is this: AI is a tool, like a drum machine, like a Max-MSP patch, like a
sequencer, or a sampler. These technologies produce results that cannot be said to be the sole
provenance of the artist who wields them. We know these arguments: the sample is theft of
intellectual property and the drum machine has no soul. All of my favourite music is premised on
these instruments. AI is just another tool for making meaning and sharing affect, if deployed with
that intention.
The more fraught argument, and the one that motivates the conceptual element of this record, is
that this technology, its story– both in terms of public reception and the datasets on which it is
premised– cannot be left to the tech oligarchs. Nor to the simplistic reactionary narratives
liberalism is so apt to produce whilst being utterly powerless to enforce. We need to devise our
own stories about our own informed and unmitigated relationship to this tech. We can only
accomplish this by working, learning, and playing alongside and with it. God bless the Luddites,
but smashing looms did not save the workers. An organized vision of what work could be was
what changed things.
attention: commodity / relief
I’ve been reading broadly about and working with machine learning over the last two years, as I
integrated these tools into my new media projects. This album is one culmination of that work
and play. And I have learned a lot about what this tool is, and is not. What it can do, and what it
can’t. Autotune does not make us all singers. Just as an MPC does not make us all Dilla. This
record was made with AI, but it was also made with friends, my voice, tape, amps, synths,
pirated software, my ever burgeoning library of ad hoc recordings, and years of experience in
playing in the mud of experimental audio.
If you seek out experimental music you know what it is to listen with your body, with your
intuition, and with the aesthetic preferences you’ve evolved over the course of your every
encounter with intentional sonic material. If you know this, you also know that in the end, the
origin of a sample, the conditions of a recording, the artist’s intention, the boutique instruments
and who played them– none of this matters. What matters is what hits. What hits the tape, what
hits the a/d converter, what hits the speaker, what hits your ear, and what hits your spirit when
you give a piece of music your attention. And this is all the record asks.

Thanks to Dexter Outhit for his collaboration. To Cale Weir for photography and digital album art. Thanks to April Martin for designing and manufacturing the physical cassette artwork. Thanks to Daniel Field for supporting the record and for putting faith in this project.