Director: Ting Song
Writer: Ting Song
Running time: 2mins
The concept of Artificial Intelligence (AI) in filmmaking puts me on edge. It raises all kinds of ethical dilemmas that nobody has a convincing answer to at this point – especially when AI are deployed in the world of work. In my day-job, I have routinely had to interview wide-eyed executives, sporting fixed crocodilian grins as they wax lyrical about the potential of the new technology. Whatever bells and whistles they chose to put on that ‘potential’, what it always boils down to in the end is an opportunity to undermine the rights of labour in their workplace.
Sure, it might not be that all the workers will be replaced by machines. They accept that there will always be a ‘human element’ to production. But if there is an AI capable of critical and creative thought on site, the value of that human in-put is, in their eyes, diminished, and more expendable than any time in human history. In the creative sector, we are already nearing the point where me saying this is speculation. There are a growing number of articles in which corporate types rejoice in the knowledge that soon designers, writers and creators who have honed their skills over a life-time of hand-to-mouth graft will have to charge less.
At the same time, the AI undercutting those creatives may very well be drawing from their work, without consent or compensation. The firms behind many of the leading AI content generators have very publicly been caught simply feeding art they found online into their software – which ‘learns’ from the content before reproducing its style over whatever imagery the user asks for. So closely does the AI ‘learn’ from the input, that very often a ghostly rendition of the original artist’s signature even appears in the corner of the generated image. There is a word for that kind of thing when human artists do it… and even though those behind AI generators swear that an AI cannot be found guilty of plagiarism, that has not stopped artists bringing lawsuits against said tech-bros, for creating a machine to allegedly do the plagiarism for them.
But like any technology, I do also want to make it clear that I don’t regard AI as inherently bad. Like a hammer, in the hands of a serial killer it could be disastrous, but in the hands of an artist of limited means who holds themselves to rigorous ethical standards, it could be a marvellous tool to foster creativity. For example, if an artist were to feed their own imagery into an AI generator, it could then help them to animate it in record time. At the same time, if you had the ideas for a melody, you could try to feed it to an AI, and it could produce a full orchestral rendition of it. For independent filmmakers operating on a shoestring, it presents the opportunity to explore their ideas in a way that their schedule and budget would previously have never allowed.
All this brings me to In a Peony’s Dream. You may have noticed that I have said a whole lot about AI, but not much about the film I am reviewing – and that is because I cannot work out where exactly it falls between these two potential uses of AI. Director Ting Song has credited three ‘Animation Designers’ in the film’s credits – Jingdan Yan, Linxuan Zhu, and Siyu Xiong – but also seems to have credited the visuals as the product of ‘AI Image Generation’ by programmes named Stable Diffusion and Midjourney.
It is therefore difficult to know where credit is actually due for much of what we see in the film. If the cartoons – at times elegant and at others, monstrous – were originated by the Animation Designers, then the production seems to be using the technology to try and bring their own ideas to life in ways that time and money might have otherwise prevented. If that is the case, this does hint at some things AI may be very good for in independent art, in the years to come. While all the nightmarish hallmarks of AI art are there – three pupils in every eye, hair warping and waving into a matted limb, and the hands… oh God, the hands! – it is a picturesque and coherent set of moving images, which lend themselves pleasingly to a non-linear, experimental animation at least.
However, if it is the other way round, and the filmmakers have choreographed a set of movements around images that were presented to them after an AI keyword search, there is a real possibility the images being brought to life were the partial creation of an uncredited, unaware partner. However easy on the eye those are or aren’t, I do not think it is appropriate to risk crediting that with a conventional score.
I have contacted Ting Song for clarification on this front, and not received any comment. So, at this stage the only thing I can do is hand out an ‘unrated’ classification. In the past I have reserved that for the absolute lowest of the low, and I’d like to make it clear that I do not regard In a Peony’s Dream as anything like those reprehensible works that have previously garnered an ‘unrated’ sticker. However, exciting as this moment of ‘technological revolutions’ may be for some artists, it is also a time for caution: of taking time to state explicitly what should or should not stand in the use of AI in art relating to ownership and the sale of labour, of not rushing headlong into a series of ‘innovations’ that may make life for artists objectively worse in a decade and leaving society objectively poorer for it in the process.