According to the renowned poet, William Wordsworth, “Poetry is a spontaneous overflow of powerful feelings; it takes its origin from emotion, recollected in tranquility.” Feelings and emotions, two realms dominated by humans, completely out of the reach of machines. Or that is what we believed until recently before artificial intelligence exhibited a potential to touch this exclusively human realm.
At this point, the possibility of AI writing poetry is still draped in questions, speculations and doubts. However, we cannot deny the fact that it is these questions and speculations that later materialize into solid facts in the future. Every invention and discovery began as a speck of thought after all.
However, in a situation where we teach artificial intelligence, human language, it won’t be long before we have AI poets. Add to this blend, the evolving quest of technology to curate artificial intelligence that is ‘learning to learn’.
One convincing example that backs this possibility is, GPT-3, OpenAI’s language generator that has been the reason of success of writing tools like Copy.ai. The brilliant essays generated by the tool attests to the accuracy and potential of AI that can be used in the field of writing.
Another interesting example is Verse by Verse, AI tool designed by Google that enables the user to compose poems with the help of ‘suggestions’. These suggestions are generated by machine learning algorithms that identify the language patterns. This information is then applied to the text generated.
The users have 22 American poets to select from in order to attain the suggestions. The type of poem is selected after deciding on the poets. To this the programs gives the added advantage of poetic forms like free verse and quatrain. The tool also enables the users to decide the number of syllables per line. After the first line of the poem is typed in by the user, the entire poem is generated by the tool. What sets the tool apart from the rest is the fact that it offers line by line suggestions, thereby making it more interactive.
Although the tools seems impressive, it carries a lot of implications. Plagiarism through AI, fake poets, or AI wearing the mask of poets, to mention a few. And a more functional dilemma is whether AI will be able to develop enough intent and purpose to break through the same rules of language that it is being taught. In other words, would AI be developed enough to work with a poetic license? The questions go up to the level of questioning the basic grounds of criticism when it comes to poetry and the place of a poet. Because ultimately, enjoying a poem has more to do with the human touch that is blended with the poetic forms and structures. Whether artificial intelligence will be able to scour that mountain is still a question in wait for a convincing answer.