By John Kendall Hawkins
Lloyd Schwartz is the Frederick S. Troy Professor of English Emeritus at the University of Massachusetts Boston, a longtime commentator on classical music and the arts for National Public Radio's Fresh Air, and a noted editor of Elizabeth Bishop's poetry and prose. He has been awarded the Pulitzer Prize for Criticism, Guggenheim and National Endowment for the Arts fellowships in poetry, and the Poet Laureateship of the city of Somerville, Massachusetts. His poems have appeared in the New Yorker, New Republic, and Atlantic, and have been selected for the Pushcart Prize, The Best American Poetry, and The Best of the Best American Poetry. Among his poetry books are Little Kisses, Cairo Traffic, and Goodnight, Gracie, all published by the University of Chicago Press. - U of Chicago Press
John Kendall Hawkins is a PhD candidate in philosophy at the University of New England (Australia), where his dissertation examines the future of human consciousness in the age of AI. He is a former poetry student of Lloyd Schwartz at the University of Massachusetts, Boston, during the year he won the Academy of American Poets Prize.
The following conversation about the times we're in, poetry, and the advent of AI creative production in music and poetry, and its initial quality, took place by Zoom on November 3, 2023.
#####
Hawkins:
You were the Poet Laureate of Somerville. That interests me. I once asked Charles Simic why he resigned as the US Poet Laureate, and he wrote back that he received way too many emails from folks wanting to get tips on how to write poems, and it kind of flipped him out after a while -- it was so distracting. Funny anecdote in a way. How was your laureateship? What was involved? What did you find most interesting?
Schwartz:
I'm still the poet laureate of Somerville, Massachusetts. I was appointed at the end of 2019 and I've loved doing this more than I ever imagined. Somerville's a scruffy town-- as you know, very diverse and one of the most densely populated cities in the US. And a lot of poets live here-- many of whom I would have guessed really lived in Cambridge, right across the city line. I'm Somerville's third PL. The city doesn't insist on anything-- no poems for the opening of supermarkets or other public occasions--so the projects I've come up with have all been my own ideas.
My first event was a reading by about 30 Somerville residents who were not poets. They were local officials, including the mayor and the former congressman and state reps, a restaurant owner, a shrink, an artist, a few musicians, Miss Black Massachusetts of 2018 who runs a shelter for battered women, and so on. They each had to pick a poem that had stuck with them, read the poem to the audience, and say why they had cared about it. It was very moving. More than 100 people showed up and nobody left before it was over. Then I had planned to do an event with Somerville poets choosing to read a poem of their own and something they loved by another poet, but Covid interfered with that event, and instead we put posters with their poems on them all over the city.
Another big project, which thanks to the Somerville librarian is still going on, is something that I call "Let's Talk about a Poem." Once a month I choose a poem that is important or interesting to me, and whoever wants to join the discussion is welcome. At first this was in person at the library. A handful of people would show up. But when the library was closed for the pandemic, the librarian suggested we do it on Zoom. I have over 100 people on my mailing list and between 25 and 50 people tune in every month, from all over the world. Some of them are poets, but all of them love poetry and are really smart. I learn so much from them about poems I think I know all about. I retired from teaching at the end of 2019, and this is a little like teaching, but it's more informal than that. I think I'll keep doing this even after I'm no longer poet laureate. I never want to stop.
In some way my most exciting project was putting together an anthology of poems by Somerville high school students. With the help of one of their English teachers, I invited them submit poems they wrote about Somerville-- about living or going to school or eating or hanging out with friends in Somerville. Some of the students already loved writing poems and were pretty talented. Some of them had never written a poem before, and a few of them had never written anything in English. This is the generation I want most to care about poetry, to fall in love with it. I was very happy with that project.
I still haven't commented on Simic's comment. Since I was appointed, no one ever asked me about writing or publishing poems. I think that's the provenance of a national figure rather than a local PL. I can understand getting tired of those questions, but that hasn't been a problem for me. So, I don't ever want to resign. I love being identified as the poet laureate of Somerville!
Hawkins:
Creativity and Artificial Intelligence (AI). There's a debate going on between detractors of AI, who insist that AI will never reach human potential, and proponents, who see AI as a great enhancer of human creativity - that it can spark ideas and generate new approaches. When I think of creativity, I think of passion and I think of spontaneity. As a poet would you say that the computational creativity of AIs will ever match the creative spontaneity of humans?
Schwartz:
I doubt it" and hope not. Your question reminds me of one of my favorite lines of Elizabeth Bishop. And it's not in a poem. It's in an essay that she tried to write about writing poetry, and that she never finished. But we have the drafts and [the unfinished essay has] been published. She said that the qualities in poetry she admires most are accuracy, spontaneity and mystery. ( I think she was partly talking about her own work.) But I think those are three profound qualities in the great works of literature and music and art, and I wouldn't disagree with her. Those are certainly the qualities that I either consciously or unconsciously look for in what I read. I don't find it universally in evidence, but I do find it in the work I love. And I think that would be very hard for AI to reproduce.
Hawkins:
You mentioned accuracy. That's a tricky word when it comes to poetry, because of the use of metaphorical language. What does Bishop mean by 'accuracy'?
Schwartz:
I think she means a kind of precision of language and a precision of feeling. And that has to do with maybe the technical sides of poetry also. If you're writing in meter, you can be syncopated. But the syncopation has to convey some kind of overall rhythmic continuity. But the more important element is emotional accuracy, which could also mean emotional honesty-- find the precise words to embody complex feelings. Those were the qualities I found completely missing in the AI version of her poem [that you sent to me].
Hawkins:
Let's get to that in a second.
I was thinking about how this ChatGPT phenomenon is really taking over. There are the devil's advocates who argue how it can help students write better by modelling organization and forms. That kind of thing. I believe it'll just make everybody lazier. It feels like we are leaving the deep, enriching age of humanity that's risen up since the dawn of the Enlightenment and are now entering into the age of pseudo consciousness? Of letting machines think for us.
Schwartz:
I agree with you. Yeah, I think that's true. It sounds like AI could help provide a good role model for a Freshman composition, where guidelines are important. But not with any serious creative thinking, in either poetry or prose. But you know, my biggest fear is in the political sphere, and that what I understand it can do there is really terrifying.
Hawkins:
How do you see AI affecting politics specifically?
Schwartz:
Well, from what I understand, with deepfakes you could have a very realistic image of Joe Biden saying something extremely damaging about himself, when he isn't in fact doing so. Something just completely manufactured. That's what worries me. I'm afraid that that the electorate is very gullible. I mean, they seem to have given Trump enough votes to make him president. And he is ruthless and many of the people who support him are ruthless. And if they can harness artificial intelligence, they can do a lot of real damage.
Hawkins:
Let's get on to Elizabeth Bishop. She is renowned for her precise language, attention to detail, and ability to find the beauty and wonder in everyday things. You share some of her gifts in your own writings, such as the playfulness you exhibited in your last volume of poetry, Who's on First? Which I'm hoping you'll read something from at the end of this.
Schwartz:
Thank you!
Hawkins:
Anyway, I was thinking about how a large language model or ChatGPT could never do what Bishop or you do with poetry. That flash of spontaneity is not there. The humor would be lost. To test the supposition, I opened up Bard, which is Google's version of ChatGPT. And I typed in the following instruction: "Write me a poem in the style of Elizabeth Bishop." That's all. That's all I did. And it spouted out a poem in very quick time - "The Algorithm." There's immediately a sense of playfulness. Like it's picked up on something about Bishop in terms of her use of language. Here it is:
The Algorithm
The algorithm is a curious thing,
A maze of code and logic's ring.
It learns our habits, tracks our ways,
And shapes our world in myriad ways.
It tells us what to buy and watch,
Who to like and who to block.
It feeds us news that's tailored tight,
And shows us ads with uncanny light.
The algorithm is a powerful tool,
But it can also be a cruel.
It can echo our biases back,
And trap us in a narrow track.
So let us be aware of its sway,
And use it with wisdom every day.
For the algorithm is a mirror too,
Reflecting back our thoughts and dreams anew.
You think it's laughable, right?
Schwartz:
Well. Yeah. What surprised me was how much it didn't sound like Bishop. Even technically. In terms of meter, Bishop was not someone who used couplets a lot, and her rhythmic language, even when she's being syncopated, there's always some kind of control so that her phrases have some kind of rhythmic connection. And this little verse doesn't have that. The rhythm gets very clunky. And if you had sent it to me and said this is supposed to be an AI version of a famous poet, I never would have guessed Bishop.
Hawkins:
Right. So, the question is how did Bard it screw up?
Schwartz:
That's a very good question because I'm curious to know what "Bishop" poems it fed into the system that made the computer think it was sounding like Elizabeth Bishop?
Hawkins:
Still, there is a kind of cheekiness to it -- in the title and the ironic "self"- referential lines. Is it Bishop playful though.
Schwartz:
I don't think it is. I was just remembering a verse that writer Mary McCarthy remembers Elizabeth Bishop reciting when they were Vassar students together. Elizabeth Bishop's dorm room was right next to the bathroom. And in an interview for a Bishop documentary Mary McCarthy quoted this verse by Bishop.
Ladies and gents, ladies and gents,
Flushing away your excrements
I sit and hear behind the wall
The sad continual waterfall
That sanitary pipes can give
To still our actions primitive.
She was an undergraduate when she recited that. We have no documentary evidence, as she didn't write it down, except Mary McCarthy remembered it. And it's kind of brilliant. I don't see that level of wit or play in the Bard poem.
Hawkins:
But I can see happening in the future how these things will be introduced to creative writing classes or writing classes in general. And I wonder how we will deal with them in the classroom. And is there any way that such knockoffs can be helpful, in their deficits, for better understanding an established human poet? Is there a way of using this in a classroom that would enhance the experience of writing at all? Or is it just a waste of time to introduce it at all.
Schwartz:
I don't know. I mean, I could use it in a classroom to demonstrate examples of bad writing. And yeah, point out what's closer to real Bishop. I mean, the opening line is okay, "the algorithm is a curious thing," although thing is a kind of weak word. It sounds like it's about to be a limerick. "A maze of code." I think that's pretty good. But "logic's ring"? I'm not sure what that is. A logical argument coming round in a ring to meet the other side? Or is it a kind of ring of logic or the ring of truth? I don't know, it seems to me not good writing, even for light verse.
Hawkins:
All right. Well, maybe. Students bring their laptops to class nowadays, right? So, maybe let's say, six students could bring up Bard and type in the prompt instructions to write a poem along the line of their favorite poet. Let's call this classroom experience, Six Students in Search of a Good Poem. So you have six different poems. And then maybe you could discuss why it's bad poetry, like what's going on with these poems that makes it so bad - or conversely, what's, surprisingly, working?
Schwartz:
Yes. That would work and might actually be useful.
Hawkins:
Okay, let's move on to AI and classical music products. Several weeks ago, I read your WBUR piece on Fall classical music doings in Boston and noted that you were touting Tod Machover's "shimmering electronic opera 'VALIS' (an acronym for Vast Active Living Intelligent System)." It seems like a crazy little opera, full of Philip K. Dick's inimitable and, perhaps, pathological melancholy and paranoia. Dick would be a good writer to turn to in a query about the Age of AI, thinking of the androids in Blade Runner, with that signature line, the android when he finally meets his maker: "If you could only see what I see through your eyes." I fear a little of that is ahead for us with AIs. But, anyway, your mini-preview of the Fall offerings reminded me of your years at the now-defunct Boston Phoenix. I seem to recall a piece you were writing about the wunderkind then, Peter Sellars. Your Machover delight would seem to suggest that you are at ease with the New and this whole AI thing. Is it a case of not taking the phenomenon too seriously or yet again of art finding a way to intelligently cope with its still unknown features?
Schwartz:
Blade Runner is one of my favorite movies. Well, I'd like to think that I'm open to anything new, anything that really kind of gets to me. And this opera, VALIS, really seemed [to take seriously] its original source. And the music was very inventive and very beautiful and very moving. And you're sort of finding your way, feeling around in it, because you're not entirely sure what's going on. That's a good thing. And then by the end of the opera, you're really very moved by it and moved by the main character's plight, and the [electronic] music really captures that. And what's music for, if not that? And so, it was an inventive opera. Technically daring. But some of the technical stuff in the new production, almost got in the way of what I like about the opera . It was almost overly complicated and self-conscious, though I still like the opera
Hawkins:
Can we expect that a kind of dynamic response in the future in relation to AI-created material?
Schwartz:
I don't know. I hope so. I mean, I think it would be great if that happened, and it would be a great excuse for AI to exist. If artists could really use it, respond to it in a serious and interesting way.
Hawkins:
All right. Let me ask you about the symphonies and the musical pieces that have been completed by AI. I've been wrestling with the Devil lately. I've just learned that AIs have initiated or completed symphonic music, old and new, and I'm trying to figure out how I feel about what I've read about so far. I am reading about AIs that can finish where classical composers left off -- Schubert's "Unfinished" Symphony; AIs that create new symphonies -- Beethoven's 10th -- or an endless string of new works by Bach. But, ethically speaking, do we want to hear an AI complete the Unfinished? Do we want AI hoons out there pushing bootleg quality algorithmic takes on Bach (I've read there's even a plan to produce endless "new" works by Bach, using AI). You know, personally, when I listen to that last movement of Beethoven's Ninth, I wonder why you would need a 10th. But when you hear the 10th that some gormless AI put out, you don't know why they bothered. Where do you weigh in on this phenomenon? Have you heard any pieces that blew you away?
[ Beethoven X: The AI Project: Complete (Bonn Orchestra)]
Schwartz:
No! I just found the AI Beethoven imitation Beethoven. And that It's kind of obvious, and you recognize a lot of themes from earlier Beethoven symphonies. And one of the things that was so great about Beethoven is that every symphony did something really new and something very different from what came before. And this AI piece was just rehashing all the old stuff instead of coming up with something new from Beethoven. I just thought it was kind of a pastiche of old Beethoven devices. So, imagine a lost first symphony of Beethoven, rather than a 10th Symphony, right? It was a kind of young Beethoven trying this and trying that, and then later on figuring out what to do with this stuff and that it wasn't sophisticated. It was really amateurish. And it didn't do much for me.
Hawkins:
What about the Schubert? Finished. No way. Huh?
[ Schubert's Unfinished Symphony no8, powered by Huawei AI at Cadogan Hall, London (Finale part only) ]
Schwartz:
No way! The Schubert was just the opposite thing. it just didn't sound like Schubert to me. I mean when you listen to Schubert, every few seconds there's something so beautiful and so incredibly inventive, but gorgeous. The AI Beethoven at least sounded like an imitation of Beethoven. But the AI Schubert didn't sound anything like the real Schubert. It just seemed a lot like a big orchestra just kind of pumping away. And I didn't, I just didn't buy it at all. I mean, I don't think I would have guess it was even imitation Schubert.
Hawkins:
Do you think it'll be something they can improve on, or is it a fault in concept? The flaw in programming that is trying to imitate humans?
Schwartz:
It seemed to me that the AI productions could have done a better job. And that It probably will at some point, and they'll probably write a better Bishop poem in, I don't know, in six months. Or five years or next week. You don't know how fast this stuff is developing. But surely there will be a better Bishop poem, and surely there will be something that sounds more like Schubert than this piece of music and something more inventive. So, I think it will inevitably be approved on. But I also think there's a flaw in the concept and an ethical problem.
Hawkins:
Shall we close with a reading from Who's on First?
Schwartz:
Sure. Shall I read that title poem?
Schwartz:
[Reads poem "Who's on First?" available at the Poetry Foundation website.]
Let's Talk About A Poem with Lloyd Schwartz (Malden Public Library Series)
