‘A poem is not a puzzle with a correct answer’: Anne Brewster in Conversation with Hazel Smith

By and | 15 February 2023

AB: At one point you write: ‘I am a machine learning to be human’. What does this mean? You talk about the interface of the human and the machine with humour and quirkiness invoking the trope of collision as well as that of connection.

HS: I was thinking of artificial intelligence and machine learning. I was reflecting on the way in which computers and human beings are increasingly merging. In the future we really will have machines that are learning to be human, but that process has already begun in that robots are often built to mimic human behaviour. There are a lot of debates about how far this process can go, how much computers will be able to mimic or even transcend human cognition, and also the degree to which robots will be able to simulate human emotion.

Maybe in this poem ‘Personhood: A Free Preliminaries’, there is some ambiguity about whether the persona is a person or a machine or both. We tend to assume it is a person but it could be a machine. And perhaps in saying ‘I am a machine learning to be human’ I was also thinking of our need to learn to be not only human but also more humane.

AB: You are an award-winning innovator of electronic writing in Australia. Can you say more about your computer-generated poetry – about that particular poesis?

HS: One of the reasons I want to employ computer generation of text is to be liberated from the restrictions of my own mind and my habitual modes of thinking. Although our own minds are wonderful resources, they can also be limiting. Obviously, the computer behaves very differently from me and throws up different results. These poems may also require a shift on the part of the reader. Experiments have been done (notably in Australia by Benjamin Laird and Oscar Schwartz) that show that people often find it hard to guess whether a poem was written by a computer or a person. This seems to refute the argument that only humans can write meaningful and moving poems, though it is important to bear in mind that a human is always directing the computer. It is even more complicated when, as is the case in ecliptical, the poems are computer-generated but with subsequent human intervention and are as variable in style as the human-written ones.

The text generation was done by the process know as machine learning and it was carried out by my collaborator, Roger Dean. The machine, in this case the computer, learns the structure of a body or corpus of text and uses it to generate new text.

In ecliptical there are some poems that are left close to how they arose from the text generation process, and other poems that I shaped a lot further, imposing my own systems upon the raw material. So I used the text-generated material to various different ends. But I do feel that the computer-generated texts, though themselves variable in style, form a good contrast with other texts in the book and that such contrasts are an important part of my aesthetic.

Text-generation apart, I am interested in the way computerisation is leading to the development of different kinds of language. This is already happening, both in the everyday world and within computer science, from text messages – which can involve linguistic play – to different forms of computer languages such as Python. Poets can draw on these changes to write poetry. In electronic poetry poets, such as mez, have drawn on computer languages to make poems that are part computer code and part natural language: this genre is known as code poetry.

AB: At one stage you were a professional violinist. There are frequent references to the violin in your poetry. For example in ‘Personhood: a few preliminaries’ you employ a rather violent image: ‘I stab the violin and the violin stabs’ me’. Does playing the violin inform your poetry?

HS: My experience as a musician is hugely important to my poetry and it has influenced my interest in sound and structure more than I can say. Being a violinist also meant that I was a performer, and I often write works for performance. But the fundamental qualities of music have enormously influenced the way I think about writing. Music is very different from language because it is much more abstract: although most of us find music immensely powerful, it is usually very difficult to say what a piece of music means. Music has power because it addresses aspects of experience that are non-verbal or at least not easily verbalisable through the everyday ways we use language. This raises the question of how you can recreate these kind of conditions in poetry, which is primarily verbal. It is a challenge, but I often strive for the realm of the non-verbalisable in my writing practice. It involves thinking of words as musical units rather than semantic ones and stretching linguistic meaning beyond its normal limits. However, you can never escape semantics in poetry, so there is a tension between working simultaneously with semantics, and also with their disruption, which it is thrilling to explore. Music is so important to me that I find it difficult to imagine what my writing would have been like if I hadn’t been a musician. Maybe I would not have been a poet at all.

Of course being a professional musician is very different from enjoying and valuing music. The image “I stab the volin and the violin stabs me’ is violent because it relates to my ambivalent relationship to being a musician. Although I love music, and loved playing the violin, I did not find myself particularly well suited to being a professional musician, and sometimes disliked that role, hence the stabbing metaphor. Playing a musical instrument is actually very sensous and pleasurable so this imagery noticeably overturns that.

AB: Thank you, Hazel, for your insights into your writing process and its contexts.

HS: Thank you, Anne, for your perceptive and challenging questions, it has been a pleasure to respond to them.

This entry was posted in INTERVIEWS and tagged , . Bookmark the permalink.

Related work:

Comments are closed.