The man who was the first in Bulgaria to place AI-generated verse among the best in a poetry competition gives his first (and perhaps only) interview ever, to Digital Stories! The questions come from… the second person to confirm the striking experiment.
Every wonder lasts three days, as the Bulgarian saying goes? Not this time. Generative models are here to stay and to change everything. Within just 3 days in November it emerged that a person calling himself Hristian Papazov had sent in poems that placed in the top 10 of the renowned “Veselin Hanchev” competition, named after the Bulgarian poet. The story repeated itself with poems I sent to another respected contest, the “Stanka Pencheva” competition, named after another celebrated Bulgarian poet.
The subject is so big and so essential that, in my view, it raises vital questions. That is why, despite the lack of any notable response from creative circles, the conversation will continue with many more important Digital Stories.
Some media outlets reported the news; colleagues from television even went looking for “Hristian” in the village where he had said he lived. But he chose to stay out of the discussion. Until now.
Who is behind the name Hristian Papazov? Who are his favorite poets? Why and how did he run his experiment? What does he think about the future of poetry, and why does he hope that… soon we humans, not the algorithms, will be writing far better, more original and meaningful poems? Our guest is an unexpected and one-of-a-kind conversation partner…
– Tell us about your entry in the “Veselin Hanchev” competition. Why did you take part with generated poems?
– This experiment was prompted by several cases abroad. The main “inspiration” was the intrusive ads urging you to create books with artificial intelligence and then sell them. I also read an article about a Japanese writer who had used Chat GPT for her novel.
I asked myself how far this is real art—isn’t it a forgery?
I decided to raise the problem in this more provocative way, because we know that otherwise it would hardly have received any attention at all.
I sent work to several competitions. In the case of “Veselin Hanchev,” I researched what kinds of texts had won in previous editions and which poets had sat on the jury. That reference point made the task considerably easier—with free verse it seems easier to carry out this kind of experiment, while with rhymed poems the program didn’t always give satisfactory results in terms of rhythm. Besides, this competition is one of the most prestigious, which mattered so that the test could be conducted at the highest level.
– Why did you choose to remain anonymous?
– I believe it isn’t so important who, but what. I think that if I had sent the work under my real name, the message wouldn’t have been understood the way I wanted. That is,
it would have looked as though I was doing it for publicity, not to spark a discussion and get measures taken.
– Did you expect the poems to end up among the 10 honored entries?
– I tried to guess the jury’s taste, given the poems that had won in previous years. On the one hand, I was surprised that established names in literature hadn’t recognized the programmed work (to me it was obvious); on the other, a more troubling thought came over me:
what is the level of AI, and what is the level of contemporary young poetry?
What is the future if a robot places among the honorees, even if only with a commendation? Does it mean we are rewarding artificially created work? Prof. Plamen Doynov noted that at the moment AI poems are of average quality. I hope that in years to come their level will be below average—that is, that we will see growth in genuine creativity made by humans.
– Well, I chose to come forward under my own name, but all the other considerations are the same. If we don’t do it, someone else will, and they’ll have no reason to admit it. Do you plan to reveal more about yourself?
– I’ll only say that I am someone who loves art and is not indifferent to its fate.
– How exactly did you create and select the poems? Did you put special effort into the prompt, the selection, the polishing?
– The “poems” are imitations; I asked for the styles of particular poets to be imitated. Of the hundreds of comments I read about these “poems,” there was one
user who even guessed whose hand had been imitated.
Of course the program needs more specific instructions to produce what you want. Otherwise the result is pretty far from the truth.
– I don’t think that’s necessarily so. In my case I simply gave the AI the competition brief and nothing more. But the question of criteria is very important too, and my first anonymous guest—the Bulgarian Literary Banksy—has an interesting perspective on it. I understand you don’t like the poems you entered. Is that right?
– They are imitations; I can’t see them as poems. As one well-known poet rightly pointed out, they contain a great many contemporary clichés. I hope we will increasingly avoid them and get to read something new and surprising.
– I wouldn’t have repeated it if I had known about your experiment, but I think it’s better that it happened not in one but in two consecutive competitions. That makes it clear it isn’t a mistake or the judgment of one particular jury. Out of almost 150 participants in two competitions, 100 percent of the generated entries, 2 out of 2, made the top 10. Isn’t that a signal not only that algorithms write indistinguishably, but even better than most people who enter such contests?
– I congratulate you on running the same experiment, and in a competition with aesthetic criteria radically different from those of “Veselin Hanchev.” I’ve also come across the view that the problem lies in the competence of the juries. But we should note that in Bulgarian literature such things are only just arriving, and hardly anyone would suspect they were reading imitations.
I recently came across a study according to which readers preferred AI-generated work to classic poetry.
This is probably a matter of reading culture and taste.
And competitions look for a certain style, and by manipulating the algorithm the generated texts match what is wanted more precisely than what some authors send in. I am convinced there were worthy pieces that simply didn’t fit the style of those particular competitions.
– Do you regret, as I do, announcing what was going on before the final rankings?
– I don’t regret it, because I don’t want dishonest honors.
I know what it is to work for years and have that work never pay off in any way.
That’s why I didn’t want some young poet to feel that way, to have something they put their heart and soul into taken from them by someone who doesn’t even write poetry.
– What is your relationship with poetry?
– I love it because it is like a secret treasure. Poems are read far less than prose and are certainly not to everyone’s taste.
Personally, good poems enrich me, and through them I live through things that will probably never happen to me.
– Who are your favorite poets?
– There are dozens. I adore poets who think on a grand scale. I admire classic works such as Dante’s Divine Comedy, Goethe’s Faust, Shakespeare’s plays, the works of Whitman.
– How do you view the organizers’ reactions—those of “Veselin Hanchev” announced your participation and said they would now require declarations of authorship, while “Stanka Pencheva” didn’t react at all?
– I’m glad the organizers of “Veselin Hanchev” took the problem seriously and that steps are being taken. The very fact that it was discussed on Nova Television’s morning show is a good start for a broader debate. I can only be glad for them and wish them success. I don’t know why “Stanka Pencheva” didn’t react; perhaps they thought that doing so would make them a laughingstock rather than a voice against such tricks.
– Except that the subject is far too big. For several weeks now I’ve been trying to provoke a dialogue about all the huge questions these results raise, in my view. And very few of the authorities weigh in with a perspective. In 4 years of work on the site I’ve had only one outright refusal of an interview, yet among poets I’ve already had 4. Why is that?
– Perhaps people in the arts underestimate the world of technology.
But I’m not one of those authorities, so I can’t give a more precise answer.
– In your view, what are the most essential questions this development in artificial intelligence raises?
– The main one is
where is the human factor going—imagination, creativity, emotion?
Aren’t we losing too much in trying to rationalize everything and make life easier? I think that with so many conveniences, people may lose valuable habits and skills.
– That is exactly why this site exists… What should this important conversation look like? Who should take part in it?
– It certainly needs the participation not only of people from the cultural sector and various creative organizations, but of political actors too. It would help to have laws on the use of artificial intelligence, so that things don’t get out of control.
The temptation to choose the easy way and have nothing created by a human is enormous.
– Can a declaration that the poems aren’t generated be a solution, when there is no way to check or prove it?
– It is still a step forward, in the sense that juries will now keep generated texts in the back of their minds. Even with a declaration of authorship, the question remains of what copyright applies to texts produced by artificial intelligence. Since such texts contain “contemporary clichés,” perhaps checks for borrowing or plagiarism from already published texts would be more appropriate.
– Is there any way left to have fair poetry competitions, now that AI is so good at generating?
– Which competition is honest and fair is highly subjective. But I’ve also come across a proposal for a competition for AI poetry—and why not?
– That is definitely an interesting idea and subject. I’ve also had a suggestion for AI as a jury… Both experiments are worth doing, and I’ll do them. Is it possible that we are already reading a lot of generated text without knowing it?
– It is certainly being used, especially for advertising.
The world is heading toward a place where the easy and the accessible are what the masses get.
So instead of spending hours thinking up an original slogan or message, people ask the program and get a pile of answers.
– Where is the limit—could AI already win a short story or novel competition?
– It may have happened abroad. I haven’t checked, but I don’t rule it out. The Japanese writer who used Chat GPT promised to publish a book written 100 percent by artificial intelligence. And if there are fans of that kind of work, it would win.
– How long will art and creativity survive with technology developing like this?
– As long as people feel a need for them.
Whether there will be real art, or everything will become a substitute, is another question.
In popular music, for example, what goes round is a cover of a cover of a cover and a remix of a remix of a remix—you get substitutes, where something old is taken and served up in shiny packaging without being any better than the original.
– From now on, can any competition entrant rest assured they are competing only against humans?
– There is never a guarantee. The organizers of “Veselin Hanchev” hardly expected anyone to send them robot lyric poetry.
– Could this, on the other hand, lead to the development of poetry? As happened, for instance, after we lost to the machines at chess and Go, yet people still play them and even invent new, unexpected moves.
– I hope so—after all, there are always two sides to the coin and two forces pulling in different directions so that there is balance. I believe that
somewhere there are also creators who want it to be hard for them, who don’t use ready-made templates and who discover ever more unexpected moves in their work.
Despite the mobile era, many people still write on paper and only then transfer it into the digital space. So there is always a way not to depend 100 percent on technology.
– Could the problem be our low standards for poetry?
– There has always been both the mass and the elite. The thing is that fewer and fewer things are called into question;
it’s hard to get beyond a black-and-white understanding that boils down to “I like it—I don’t like it” or “this is cheerful, that is sad.”
To a large extent the problem is the lack of insight, the conveyor-belt reading and writing. It’s as if there’s no time left for reflection, and that is vital.
– Beautifully put! The point of poetry and art is to express yourself, to write because you need to, not to win competitions…
– That’s a problem not only in art but perhaps in every sphere—the desire for fame and recognition. Competitions and contests give you a sense that you are good, that you’ve received a kind of blessing from the authorities to keep going.
The truest things happen when a person doesn’t crave others’ approval but feels happier and more fulfilled by what they have created.
– Do you plan to continue with more experiments like this?
– It will take me quite a while to think of something that hasn’t been done. It’s getting harder and harder.
– I too love doing all sorts of experiments, the more unexpected the better. Because it really isn’t enough to say, show and explain—you also have to demonstrate and test things in practice. What do you think I should do next?
– I would find it interesting to chat with strangers through a program that generates the continuation of a given conversation. Let’s see whether people could tell the robot from the human!
– A wonderful idea! And the conversation is only just beginning, at least as far as Digital Stories is concerned… Thank you from the bottom of my heart for your trust!
