What’s the film about?
My father’s health is in decline and this made me want to spend time thinking about him and his amazing landscape watercolour paintings. I found myself writing a song called My Father’s Dreams.
I was looking for a good subject on which I could base an AI film and this song seemed perfect. It would help me to keep digging into the AI world, and to allow me to think more about my dad’s life and work.
I used AI to visualise the subjects of my dad’s paintings that are described in each line of the song. I experimented with AI backing vocals to accompany my own voice on the recording of the song.
The film ends with some stills of my dad’s paintings.


Why make the film?
I am excited by the creative possibilities of AI, but much of the work that I see is not that interesting to me: often the focus seems to be on doing old things more cheaply and quickly rather than on doing new things, and also on making work which emulates other people’s style.
I am interested to see if I can make things with AI which I think are really beautiful, and which capture my own taste and style rather than someone else’s.
What was it like to make the film?
My AI experiments have often given me the sense that the AI platforms are aimed at non-professionals. Suno talks about ‘making any song you can imagine’. Does a serious musician want this galactic level of possibility? Or would they rather that AI helps to propel them further along their chosen path, helping them to make things with a clear through line from their previous work?
To begin with, I thought Midjourney was also aimed at non-professionals. Youtubers told me about ‘style refs’ – a system which instantly facilitates existing visual styles. I played with these and got some nice results, but they weren’t my results. I didn’t feel ownership, and it still felt like I was playing with a clever toy, rather than a new kind of paint brush.
I decided to see if I could ‘break’ Midjourney in order to achieve more authentic results. When I began to use my own photographic artworks as style references, I started to get weird, new versions of my own style rather than pastiches of other people’s. I began to feel some control and ownership of the AI outputs.


Is more control always good for creativity?
As I learn more about the AI tools, and as they get better, I will no doubt have greater ability to control the outcomes. I will be able to perfect shots which I struggled with in this film, for example the one that accompanies the line: ‘the sea breathing in and out on the beach’, where I couldn’t quite get the tide to come in and then go out again.
But there is something very exciting about working with AI, precisely because of the lack of control. Making this film reminded of designing record covers at the beginning of my career. I was using experimental darkroom techniques. I got to a level of expertise where I knew just about what results to expect, but only within a few degrees of latitude, and the surprise aspects of the outcomes were exciting and a little bit shocking.


How do I feel about the creative use of AI now?
It seems to me that when you are using AI as a creative, you have to learn how to avoid what Brian Eno calls the ‘chasms of mediocrity’ that AI wants to deliver. You can do this by working as hard as ever to come up with good ideas, and by carefully considering what data you are getting the AI to work with.
I am acutely aware of the difficult questions around AI and it’s impact on working creatives. And I am invigorated by this crazy technology of the future. I am already seeing brilliant people do great things with it, individually and in collaboration with others across geography, disciplines and languages. I want to continue using it and see if it will help me to make things which are meaningful, emotional and authentic.