It all begs the philosophical question: Can computers think?
These are both questions first posed by Alan Turing, a British computer scientist and creator of the Turing Test. The Turing Test is a measure of computer capabilities in comparison with human thought processes. For instance, writing poetry.
But, whether a computer can actually write poetry is up for debate. In many ways, a computer can gather words, organize them into comprehensible order, and present them as a poem.
For instance, take a look at this poem:
I heard a fly buzz when I died
The stillness in the room
Was like the stillness in the air
Between the heaves of storm
The eyes around had hung them dry
And breaths were gathering firm
For the last onset when the king
Be witnessed in the room
Now, read this second poem:
Slowly I dream of flying. I observe turnpikes and streets
studded with bushes. Coldly my soaring widens my awareness.
To guide myself I determinedly start to kill my pleasure during
the time that hours and milliseconds pass away. Aid me in this
and soaring is formidable, do not and wining is unhinged.
Can you tell which poem was written by a computer, and which was written by a human? The first poem was written by famed poetess, Emily Dickinson. The second, an algorithm (and author), Racter.
Looking to take an actual Turing Test? Click here.
Computers have been “writing” poetry since the 1950s, when the concept of artificial intelligence was gaining quite a bit of traction. The Turing Test manifested as a method of evaluation to determine the thinking capacity of computers. According to Alan Turing, computers can be said to have intelligence only if they are able to sustain text-based dialogue with a human who cannot identify whether they are talking with a human or a computer. But, once is not enough. For Turing, a thoughtful computer can trick a human into thinking the computer is a human at least 30% of the time.
Racter is perhaps one of the best examples of a “thinking” computer. Having published an entire anthology of poems, it’s clear that computers can indeed write poetry. But, Racter is just one example.
In the 1980s, Ray Kurzweil developed an advanced poetry-writing algorithm. The algorithm, RKCP, is fed a source text, analyzes the text for language patterns, and then regenerates language that imitates the structure of the first text.
Not so much! To figure out how well computers can write poetry, Oscar Schwartz and Benjamin Laird developed a Turing Test for poetry. In 2013, the duo published bot or not, a game that tests poetry-writing algorithms (like Racter and RKCP0, using human users as evaluators.
The research completed by Schwartz and Laird presents some interesting (albeit uncanny) results. With data gathered through bot or not, Schwartz and Laird have found that there are instances when humans are fooled by computers. In other words, human users consistently identified a poem as human-written, when in fact it was computer-written. Within the bot or not database, there are poems written by computers that at least 65% of humans believe was written by a human.
Want to contribute to their research? Visit their website.
Being said, computers can certainly write poetry, but at what point do computers understand what they’re writing? Let’s ask Siri. Go on….Try it!
Mika Beaudrie is a STEM Writing Instructor. She holds a master’s degree in English and writes extensively about Black rhetoric in the sciences. Currently, she is busy traveling the world while investigating language epistemology.