by Wenjuan Zheng
“there is no one else in the world.
there is no one else in sight.
they were the only ones who mattered.
they were the only ones left.
he had to be with me. she had to be with him.
I had to do this. I wanted to kill him.
I started to cry.
I turned to him.”
Did this poem touch you? As Adam O’Riordan, a famous British poet, said, “poetry expresses what is to be human. It is therapy for the soul.” But, what if I told you that such an emotional and humanistic piece is not even written by a human being. Would you be shocked?
The author of the romantic poetry is an accidental byproduct of Google’s artificial intelligence (AI) project, which is a recurrent neural network language model (RNNLM). RNNLM was initially developed by Google with Stanford University and the University of Massachusetts to enhance Google’s translation and image captioning tasks. While the algorithm is very complicated, AIs like this work by using statistics about sentences from prior works to construct new sentences. The result are machine-written sentences that mimic some of the linguistic patterns that exist in human writing.
AI might transform our life, including writing, in an unprecedented way. We have long accepted that computers can do a better job on calculations than humans. However, writing is a whole other kind of animal which requires emotional input and streams of creativity. As one of the writers put it, “I had thought my job was safe from automation—a computer couldn’t possibly replicate the complex creativity of human language in writing or piece together a coherent story. I may have been wrong.” The writer’s concern is not out of nowhere. A year ago, a Japanese AI wrote a novel, The Day a Computer Writes a Novel, which nearly won a literary award. A few months ago, a fan of Game of Thrones, a famous HBO show, used an AI to write a version of the sixth book since fans have grown so impatient with George R.R. Martin’s slow writing pace.
This advance of technology inspired “a combination of amazement and handwringing” (Brogan 2016). There are reasons to be concerned about the implications of AI, but a closer look at the story, in fact, can show us the limitations of AI in replacing human creativity. Far from replacing the human, AI’s writing usually requires a lot of human input. For example, the Japanese AI that wrote the novel which nearly won an award was actually aided by researchers who first wrote 80% of the story. Also, while an AI can reproduce Martin’s writing style in the sixth book of Game of Thrones, it cannot produce a globally cohesive storyline and sometimes reintroduce characters who were already written out. Last, the Google AI can’t always succeed in generating a poetic sentence, and sometimes the poetry is a pure chance of luck. In other words, AI has a great potential to complement the creative process of human, rather than to replace human’s task.
As students or educators in the higher education, we simply cannot ignore these new trends. We should not think of AI as the enemy, but as a useful tool for educators and students. Writing is a challenging task. It is a mental exercise that probably works the best when you are entirely free from all the grammatical and structural constraints. The emergence of AI-powered writing assistant tools probably is excellent news for some ESL students who often have language and mental barriers to write. “Write down everything in your mind” is the slogan for many writing instruction books for students; however, we can’t turn in those messy drafts to professors, and no one is willing to read or even decode the complicated ideas hidden in sentences. We need effective proofreading. AI-powered writing assistants such as Grammarly and Atomic Reach, which provide sophisticated grammar checks for writing or access the readability of the written concent, become handy. It is not to say that we should completely rely on those tools. We cannot because the technology is still in its infancy at this point with some defects. For example, Grammarly sometimes give out wrong suggestion for word. However, utilizing those tools can help users tackle some mistakes and improve our writing. These tools can be useful as long as we are aware of their limitations and students understand they need to discern when a correction is legitimate or not. Instead of focusing on the technical part of how to write, we should probably focus on why we do things and the idea that the piece of writing is trying to convey. Therefore, AI can complement human writing but never replace its creativity.