Adorno and Ben Lee on AI
And Media and Comms jobs going at Swinburne
On Friday I tuned in to an episode of ABC News Daily, which featured Ben Lee as a guest to discuss the recent (and vital) APRA AMCOS report on AI and Music.
The episode was immediately startling for having used an AI tool to generate a musical theme for the podcast - startling, because I wrote the usual theme that ABC News Daily uses every other day, so I got to witness a real moment of personally being replaced by AI. The AI version was close enough to be a bit weird, and far enough out and badly judged to be even weirder.
Lee said a few things that I disagree with, but also quite a few very considered and thoughtful things about AI and music: mostly, the point I came away with was that the problem with AI is not even the technology itself, but rather the societal and cultural circumstances that give it plausibility.
Recently I’ve been thinking a bit about what Adorno would think about AI and I suspect he’d come to a similar conclusion to Ben Lee (possibly the only time someone has put those two in the same thought). That is, it is the collective devaluing of culture, not technology, that makes AI products possible. “Culture today is infecting everything with sameness,” Adorno wrote with Horkheimer. Our contemporary culture for a very long time has valued genre and consistency and expectation. If I ask you to imagine a Melbourne coffee shop, for example, you’re probably going to come up with something broadly within the same three or four genres of coffee shop: industrial with a roaster out the back, or clean and simple with a white stone benchtop, or outer suburban with laminated menus. You can do the same exercise imagining entire varieties of culture. What about a horror movie poster? A jingle on an ad for laundry liquid?
“All mass culture under monopoly is identical,” wrote Adorno and Horkheimer, and it doesn’t take much of an intellect to wonder if that better applies to a mid Disney+ eight episode tv series tied to one of their many pop culture brands, or AI slop trained on billions of dollars of stolen Intellectual Property that all turns out with the same strained sheen.
In other words, culture had to become ‘content’ before AI could be plausibly trained to generate it.
Still, a lot of sensible people are heralding the end of this initial blast of AI hype, and I think they’re probably right. Like the NFTs, crypto, and the metaverse before it, AI is simply too expensive, too limited, and has too few genuine use cases for everyday people to deliver on the astronomical transformative promise that has undergirded stocks and VP investment rounds and environmental vandalism. It’s just a bit shit, really. Once you get past the initial shock and awe that a machine is capable of producing human-like responses, once you actually have to work with the tool, its limitations are obvious and off-putting. There is very little of the grain of usual creative tools: it is not built for tweaking and refinement and artistry, but tech demos at investor keynotes. Where it remains useful is largely in the realm of specialist tools (as the APRA report points out, in music this might look like automated basic mixing and mastering, something already widespread) where ‘AI’ might actually only be temporary branding, a magical bit of LinkedIn wish fulfilment to make a tool seem more advanced and technical than it really is.
I think there is also the growing sense among audiences that we might have already seen the most interesting and exiting things in the creative world that AI can produce, coupled with the gnawing sense that it’s all a little bit uncool.
The dream for most AI boosters is, as always, an economic one: capitalism without workers. The dream for everyone else should be a doubling down on what technology can play with but never generate: the human, the intimate, the personal.
I had three extraordinary PhD students of mine graduate at a Swinburne ceremony last week! In alphabetical order:
Dr. Cassandra Barkman, who submitted a thesis titled Narrative Complexity in Videogames (supervised with Prof. Angela Ndalianis)
Dr. Taylor Hardwick, who submitted a thesis titled Re-imagining safety and community in digital spaces: Online games events during Melbourne’s COVID-19 lockdowns (supervised with Prof. Kath Albury)
Dr. Vincent Tran, who submitted a thesis titled A Brief History of the Shared Universe in Popular Culture (supervised with A/Prof. Liam Burke)
I couldn’t be prouder of this lot. Each one of them did astonishing work, whether it was Taylor staying up for hours on end during lockdown to track every last interaction on the PAX Discord server; to Vincent making the most astonishing spreadsheet of shared universes you’ve ever seen, from Balzac to Power Rangers and the MCU; to Cass somehow managing to meaningfully unite Dwarf Fortress, Five Nights at Freddy’s, and Doki Doki Literature Club in the same thesis.
Truly one of the most satisfying days of my academic life so far.
Finally, I’m getting the word out there about FOUR academic jobs going at the moment at Swinburne within the Department where I’m Acting Chair. If you or someone you know is a Media and Communication academic in Australia, please send these their way.
Department Chair (hiring at level D or E)
Professor or Associate Professor, Professional Communication
I’m happy to answer questions about the latter three positions, and direct you to the right person if you have queries about the Chair role.
Okay final final thing but I also made a fun little video for ABC Classic about the changing audience for classical music and the rise of soundtracks. Watch it here.
The devaluing of culture and the human. It's troubling me so much. I will listen to the Ben Lee episode. Love the sound of those PhD theses! Congratulations to all - you must be so proud too.