Can Computers Write Poetry?
Are computers capable of producing original expressive work such as poetry? Researchers, poets, and computer scientists have long sought to answer this question. The jury is still out, but a closer look at the available evidence suggests that while computers are capable of producing poetry, they are still far from producing `human’ poetry.
As early as 1950, mathematician John von Neumann developed a Monte Carlo algorithm to generate random reprints of T.S. Eliot’s poem “The Waste Land.” His algorithm was able to produce some new verses, but it was far from producing true poetry. In 1968, computer scientist Alan Turing developed a program for the Turing test, a loose measure of Artificial Intelligence (AI). He hoped that the program, called ELIZA, would be able to pass the Turing test by writing a poem.
Since Turing’s time, there have been several attempts to create computer-generated poems. Researchers at Carnegie Mellon University created a program called POETRY that was able to generate poems in the style of noted poets. Others have used neural networks and Markov models to generate poems using large databases of text. However, the results have not been particularly impressive.
One common technique used to generate poetry is to take a sample of text, such as a short story or a book, and create a poem by rearranging the words. This technique has yielded some moderately successful results, but it is still far from producing true poetic creativity. Another approach is to use natural language processing (NLP) algorithms that try to understand human language and generate “natural” poetry. The results of this technique are limited, as the algorithms are not sophisticated enough to truly understand the meaning of words.
So can computers write poetry? The answer is yes and no. Computers can generate some poems, but they are far from producing the kind of poetic creativity that humans are capable of creating. It may be possible for machines to eventually produce original poetry, but this will require a level of Artificial Intelligence (AI) far beyond what exists today.
Computer-generated Poetry Competitions
Computer-generated poetry competitions are becoming more popular in the AI world, with entries ranging from poems composed with natural language processing algorithms to algorithms that generate words and phrases at random. Although the results of these competitions can be quite impressive, the computer-generated poems still lack the nuance and meaning that human poets can create.
In 2018, a competition called “Bot or Not” was held by the New York based non-profit Poets Online. The winners were two AI-generated poems written by Elsa, a neural language model developed by the MIT Media Lab. The first poem was titled “Robot Rising” and the second was called “Lamentations of a Human Heart.” While the poems were impressive in terms of their structure and creativity, they still lacked the depth and subtlety of human expression.
Computer-generated poetry competitions are a great way for computer scientists and AI developers to test the limits of their algorithms and take a first step towards creating truly expressive and creative computer-generated poetry. However, these competitions have yet to come close to producing the kind of poetry that humans are capable of writing.
Conclusion
In conclusion, computers are currently capable of generating some poetry, though the results are limited. Some researchers have used natural language processing algorithms and neural networks to generate poetry, while others rely on algorithms to rearrange text to create a new poem. Despite the advances being made in the field, computer-generated poetry is still far from being able to produce the kind of poetry that a skilled human poet can.
Applications of AI in Poetry
Artificial Intelligence is being used in a variety of different ways to enhance the craft of poetry. AI can be used to automate the process of editing and revising a poem, as well as to generate ideas for poems. It can also be used for pattern recognition, which may help to uncover connections between different lines of a poem and give the poem more depth. AI can also be used to create new forms of poetry, such as interactive or digital poetry.
AI can also be used to analyze the language used in a poem and detect elements of style such as similes, metaphors, and allusions. This can be especially helpful for poets who want to improve the quality of their work, as they can use AI to point out areas where their work can be improved. Similarly, AI can also be used to detect subtle changes in language or meaning that may affect the overall tone or atmosphere of a poem.
One of the most intriguing applications of AI in poetry is its potential to generate entire poems from scratch. While current applications are limited to generating fragments of poems, AI could eventually learn to generate entire poems in the style of a given author. For example, AI could be used to generate a poem in Emily Dickinson’s style, or it could be used to create an entire anthology of original poetry.
Limitations of AI in Poetry
Although Artificial Intelligence is being used more and more in poetry, there are still some limitations. AI algorithms are limited by their ability to detect and understand the human meanings behind words and phrases. This means that AI crafted poems may lack the subtlety of emotion and expression that is necessary to create true poetry. Similarly, AI has difficulty understanding the rules of grammar, rhyme, and meter that are necessary for successful poetry.
Another limitation of AI in poetry is its reliance on existing databases of words and phrases to generate new work. This means that AI may be limited in its ability to generate truly original work, as it will always be working within the confines of a predefined set of words and phrases. Finally, AI can’t yet replace a human poet’s ability to tap into the deepest corners of the psyche and create works that are both meaningful and moving.
The Future of AI in Poetry
In the future, AI may become an important tool for poets, used to edit and revise existing work, generate new ideas, and even craft entire poems from scratch. However, the ultimate success of these efforts will depend on the progress of AI technology. As AI algorithms become increasingly sophisticated and able to detect and understand the complexities of human language, they may eventually be capable of creating truly creative and original works of poetry.
At the same time, it is important to remember that AI can only ever be an adjunct to the creative process. Even if AI algorithms become sophisticated enough to generate meaningful and expressive poetry, they will still lack that spark of originality and unique insight that is the hallmark of great poetry. Therefore, while AI may be able to help enhance the creative process, only human poets can unlock the creative potential of language.
