Could Alexa Write Poetry?

Reading time ~5 minutes

Can a computer be creative? This question has been asked since the invention of the machine and each breakthrough in the technology increases the speculation. But can a computer match, or even succeed, the creativity of a human?

In Marcus du Sautoy’s recent book The Creativity Code he asks, can computers be creative in today’s age of machine learning? and then goes onto discuss “the Lovelace test”: a machine must produce something “new, surprising and valuable”, and do so in such a way that its programmers are unable to explain the feat. Although a computer may pass the test and generate a piece of work that maybe considered art, what is missing in the discussion is the interpretation of the work by a human, or as summarised by E. A. Buccchianeri, “Art is in the eye of the beholder, and everyone has their own interpretation”.

There is a long history of computer generated art. Starting in the 1950s computers were used to generate visual art which has continued up to the present time but they tend to be used by artists as a digital paint brush. With the increasing development of Artificial Intelligence, where there is no influence of the programmer, computers have been left to generate art visual art. There is a similar story for computers creating music music, and recently poetry.

All of the development in computer creativity opens up interesting questions about whether a computer is simulating the creative process or it is genuinely creating something new and novel. For example when a computer writes poetry it will use algorithms to start with one line and then starts to generate additional lines that rhyme according to traditional poetic structures, say a sonnet. To allow the computer to use figurative language, the algorithm uses mathematics to create metaphors and similes. The following is an example of a computer generated poem

A wounded deer leaps highest,
I’ve heard the daffodil I’ve heard the flag to-day
I’ve heard the hunter tell;
‘Tis but the ecstasy of death,
And then the brake is almost done,
And sunrise grows so near
sunrise grows so near
That we can touch the despair and
frenzied hope of all the ages

but when compared with say the House of Reflections by Olive Cooper, 12 years ( from On Common Ground by Jill Pirrie ):

I see my face in the saxophone,
Stretched and curved.
Sun slides around it
And bounces off the walls.
The zig-zag pattern at the carpet
Sinks into the shiny black piano,
And I move in the dead television.
I see two selves in the photo glass,
Years between us.
Sun on the garden bird bath makes wavy
Sea patterns on the ceiling.
Leaf shadows sweep across the wall.
Everything moves and all is still.

there is a clear difference. For example Olive’s poem has a subject and it is full of evocative images that bring the poem to life in the mind of the reader, or at least it did for me. Although there are some interesting images in the computer generated poem, it lacks any form of coherence. I am assuming it is about death, but it could be about death in a battle or maybe something else! Also, there is no title in the computer generated poem which can be an important part of the poem and signals to the reader what it is about.

Computer technology is always improving therefore it is interesting to speculate how far a it could be developed so that it can write a poem as good as a human. Imagine asking Amazon’s Alexa, or any other similar device, to “write a poem about chocolate chip cookies!” Alexa’s poetry writing capabilities would have to include information about a chocolate chip cookie: the components of a cookie, how it is made, its smell, taste, what it feels like to hold it, the sensations when it is eaten and so on. Alexa would need information about the history of the cookie, the social context of the cookie, for example when and where it is eaten, conversations it may overhear, where it was being eaten, and any other information that an imagination could generate. All of this information could be stored in a computer and would be the source of lines and words for the poem. Once a palette of information had been gathered then the next stage would be to use the many techniques that a poet employs to generate a poem: metaphors, similes, different types of rhythm and rhyme, the social importance of certain words, a personal history of eating cookies etc. This is something that Alexa using Artificial Intelligence could probably do. Finally Alexa would have to produce a poem that could not have been influenced by the programmer, in other words write it itself, in a way that connects with a reader in an emotional way - a very tall order! But as the list of capabilities show Alexa would be simulating the process of writing poetry and therefore it would not guarantee that the poem it wrote would be profound.

The highest accolade that a poet can receive is to win the Noble Prize for Literature which is a recognition of the impact that their poetry has had on society and beyond. Many of the citations set the standard that computer generated poetry would have to reach, for example the citation for Seamus Heaney is: “for works of lyrical beauty and ethical depth, which exalt everyday miracles and the living past.” For a computer generated poetry to reach these levels it would have to have the skill, imagination, subtleties in word use, personal history, and an ability to make a particular subject into something that is universal, which is something that the great poets exhibit in their poetry. But finally, it would be judged by a human and it would seem impossible that a computer would ever be able to write a good poem because it would not be able to experience the life of humans.

Computer technology continues to develop and will break into new areas of our lives, and possibly open up new forms of creativity, but as discussed above, it will never be creative when compared to a human.

Mrs T On The Global Environment

Mrs Thatchers speech to the UN about the Climate Crisis Continue reading

Can The Market Value Soil Health?

Published on September 26, 2024

Government Spending on Species Recovery

Published on August 26, 2024