Found poetry is something we got to talking about a little bit last semester in e-lit when I got to talk a bit about the Blackout Poetry Tool. Keith Holyoak also talks some about it in his article which explores the question Can AI Write Authentic Poetry?

Holyoak dove too deep into that question of AI truly being able to create real poetry, the explanation for what found poetry is serves as an important reminder, yes, but also gives a good preface into his stance and somewhat even goes into the why.

I, for one, already have an idea of where I stand on this issue for the reasons expressed in one of my previous blogs. Artificial intelligence can only use what it’s been given, as is the case with humans also. But here’s the catch: even though humanity is given such boundless resources and knowledge, we still have limited knowledge of and access to much of it. I mean why do you think things like AI and other technologies are still developing and discovering things? Because there’s still more out there.

So that said, AI is still limited to human knowledge, language and opinions, with a “brain” modeled much after the mind of man (I mean, man created it after all, and its “brain” is the internet which includes record of the endless spewing of human thought, knowledge, stupidity, wisdom, and interaction). It may not be able to experience or develop the way we do, but it can calculate and model what our experience and development is like. On top of this, its “brain” has the ability to work much faster than ours do, in part because we have so heavily relied on it’s computational powers as a replacement for our own (rather than using it more like a crutch when some part of us needs rest).

And because this “brain” has become so vast with the developments of AI technology and its codes modeled to function similarly to humans that the way it functions is now somewhat unpredictable at times.

Computer programs can now learn from enormous sets of data using methods called deep learning. What the programs learn, and how they will behave after learning, is very difficult (perhaps impossible) to predict in advance. The question has arisen (semiseriously) whether computer programs ought to be listed as coauthors of scientific papers reporting discoveries to which they contributed. There is no doubt that some forms of creativity are within the reach, and indeed the grasp, of computer programs.

Can AI Write Authentic Poetry? Keith Holyoak

When we did AI-generated poems in class a few weeks ago and shared them, most did not include much wordplay or unique uses of clichés (rather, the clichés were used in the most cliché ways where it was not necessary to a strong poem). So given how fast AI can learn and develop, I decided to try it again with a poem I wrote last year and recently re-edited.

I think it goes without saying, but just to clarify, the first document attachment is my own work (with italics being words pulled from the English Standard Version of Psalm 22 John 19:30), while the second was AI generated. When it comes to creative writing, I’m sure that if we’ve made it this far into a degree in English Writing (regardless of our status as graduate or undergraduate) that we’ve had the phrase “show, don’t tell” beaten into our minds. AI language models like ChatGPT may see these words a lot, considering its vast “brain” that is the internet, but it hasn’t yet reached a point where it can understand the word. It associates definitions to words because there are online dictionaries, sure, and that leaves open the possibility; in the meantime, while it can connect the dots, it has not yet been able to elicit or create its own meaning from that in the same way we have because it has no ability to personally experience these things its prompted to write about.

(Sorry this post ends pretty abruptly, Holy Week is next level busy this year).