Jon Wiener: From The Nation magazine, this is Start Making Sense. I’m Jon Wiener. Later in the hour: Can AI write poetry –good poetry? Katha Pollitt decided to find out. but first: Harold Meyerson with today’s political news – that’s coming up, in a minute.
[BREAK]
First up: the Democrats’ strengths, and the Democrats’ divisions: for today’s political update, we turn to Harold Meyerson. He’s editor at large of the American Prospect. Harold, welcome back.

Harold Meyerson: Always good to be here.

JW: I want to start with the political news from LA. The primary for mayor has been finally resolved. Incumbent Karen Bass came in first with 34%–just a third; Progressive City Councilmember Nithya Raman got 29%. So the two will go into a runoff in November. In third place, Republican Spencer Pratt, supported by Trump, 26%. The New York Times described Nithya, as we call her, as “a progressive Democrat who has drawn comparisons to Zarin Mamdani.” How would you compare Nithya and Mamdani.

HM: Somewhat unfavorably for Nithya. To begin with, Mamdani really united the entire New York City left. And running against Andrew Cuomo, that was not all that hard to do. And he also had the most powerful and united DSA local behind him, but not just the local. If the local provided almost all of its 10,000 or 12,000 members to walk precincts and work phone banks, Mamdani’s total corps of volunteers was an amazing, just over 100,000, which means roughly 90,000 nonmembers of DSA also worked for him. He united that, and sort of like the victory of Fiorello LaGuardia in 1933, he sort of united all of this progressive ferment with a rising ethnic group, East Asians and South Asians in in New York City.
Nithya Raman can’t really claim that kind of a groundswell and that kind of support. The LA political landscape is somewhat divided. It’s not clear, actually, that she is running really that much to the left of Karen Bass, who has almost no nothing in common with Andrew Cuomo. Karen Bass, like Nithya Raman, emerges out of the LA left, a  long time one of the leading advocates and police critics in South Central. Over a decades- long career, having worked as a 20-something on a Venceremos brigade in Cuba, a background that is actually more to the left than Nithya Raman’s background.
And Nithya had then sort of compromised on some of the issues that DSA in LA supports, voting for increased police funding, backing away from all the provisions of the mansion tax on high valued property when it’s sold. So that really, I think, on issues, as was not the case in the Cuomo race, there isn’t really a hell of a lot of difference between Bass and Raman. The difference is that Bass is a relatively unpopular mayor, and Raman is still largely a tabula rasa to Los Angeles voters.

JW: I would add one other thing. The three other DSA-endorsed members of the L.A. City Council have all endorsed Karen Bass. That is a significant move on their part.

HM: It is indeed, and it’s reflective of what you might call the growing left establishment in LA has been feeling and prioritizing – in the sense that Nithya, who had also endorsed Karen Bass just two weeks before she entered the race. It sounds that Nithya is in this particular instance more, and, shall we say, more an opportunist than the personification of a left upsurge.

JW: We should say something about Spencer Pratt. Trump has declared that there’s quote, “big cheating” by the Democrats, and his evidence is the vote count, what he calls “delay.” Some see this as Trump’s strategy for the midterms in November — to attack the results when his candidates don’t come in first. And we saw on Meet the Press how he made the case: He just said the fact that the count is slow is his evidence of cheating. “All I have to do is look,” he told Kirsten Welker when she asked for evidence of cheating. What’s your assessment of the Spencer Pratt candidacy in Trump’s case here?

HM: All of this partly results from the undue hype that Spencer Pratt received while running for mayor. I want to make one point very clear: He’s polling at 26%. Do you know what percentage Donald Trump won in Los Angeles, In the city of Los Angeles when he won the presidential election in 2024?

JW: What percent did Trump get in L.A. in 2024?

HM: 26.5%, while Kamala Harris got 70.1%. In other words, there was no imaginable way that Spencer Pratt was ever going to top maybe a quarter of the vote in Los Angeles. The percentage of Republicans within the city of Los Angeles is below 25%. And when they’re added to the Republican leaners in Los Angeles, you more or less get to 25%. But that’s it. That’s what the population of the city politically is composed of.

JW: Trump’s complaint is that Spencer Pratt was in second place for a few days, and then, as the week went on, he fell into third place. Trump considers that to be evidence of fraud, cheating, and “you’re all crooks.” Do you have another way of understanding this?

HM: Yeah. California has a system which is, you know, could stand some tinkering with. But a system by which mail ballots–and most of the ballots cast in California are mail ballots–mail ballots that are postmarked no later than Election Day can be counted up to a week after Election Day. So long as they’re received by a week after Election day, which happens to be the day we’re recording on Tuesday of the following week which allows for late counts.
Now, in this particular election Democrats were filing their mail-in ballots disproportionately late because they had to vote tactically, looking at the last polls in the gubernatorial election.  This is all if you’re looking for sort of a meta culprit here, you might just pick the goddamn jungle primary, which under this current alignment of gubernatorial candidates, compelled late voting among Democrats. But that is essentially the full explanation of what went on here. And anyone who really understands California politics and Los Angeles politics, looking at the raw figures immediately after the polls closed, could have concluded that the fact that Spencer Pratt was initially running second was, you know, in no way a guarantee that he’d end up finishing second. Quite the contrary.

JW: The biggest and best poll in California before the election, conducted by UC Berkeley, found that 55% of California Republicans say they are not confident in the results. So Trump does have an effect on how Republicans in California understand what’s going on.

HM: That’s because they haven’t looked at voter registration figures. Republicans in California have been falling beneath, you know, the Democrats.  About 40% of the registered 45% of the registered voters in California are Democrats. About 20 high 20% are independents. Low 20% are Republicans. If they had they had a rational complaint, it would be that they’re living in the wrong state. They are a marginal political group within California, that has so estranged the huge number of immigrants who came into the state beginning in the 1980s, and continued with such unpopular positions as opposition to abortion rights and so on, that they understandably fell to a marginal minority. That’s who they are. And it’s not because of the way the elections are structured, it’s because they have been brilliant at self marginalization.

JW: I have one more quote from the New York Times about the mayoral election in L.A. they described Karen Bass as “a grandmother and a former physician’s assistant.” Now, you’ve pointed out there are some other things in her resume, starting with the Vinceramos Brigade, including chair of the Congressional Black Caucus.

HM: Speaker of the California Assembly, and before all that, leader of Community Coalition in South Central. She’s a community organizer. She’s been speaker of the California State Assembly. She’s been head of the Congressional Black Caucus. She was someone Joe Biden considered for the vice presidency in 2020. And so on. So that’s a rather dismissive New York Times evaluation.

JW: Moving on to other California election results that have national significance: There’s a key test of whether Democrats can flip a Latino district that switched to Trump. This is the congressional primary in the Central Valley, a district between Fresno and Bakersfield, California 22.  A fascinating contest in the poorest, most Latino district in the state. Trump won this district in 2024 by 1.8 percentage points. Before that, Biden had won by 13 points. So this is one of those places where Latinos switched to Trump in 2024.  And there is a Republican incumbent running for reelection: David Valadao.  Valadao’s district has the highest share of Medicaid recipients, the highest per capita share of households receiving food stamps of any district currently represented by a Republican. Nevertheless, Valadao voted to slash federal funding for Medi-Cal, even though two thirds of his constituents rely on it for health care. So he has become one of the top targets for Democrats to flip this seat in November.
This is also significant because it’s the site of a battle between the progressive wing of the Democrats and the establishment of the party. The party has pretty much endorsed– The DCCC has endorsed– State Representative Jasmeet Bains, who’s also supported by pro-Israel money and is described as a “moderate.” And she’s being challenged in the primary by a grassroots progressive endorsed by Bernie and AOC and Dolores Huerta: Randy Villegas.
The results as of now, as we speak. 72% reporting: David Valadao 42%; progressive Randy Villegas 31%; centrist Democrat Jasmeet Bains 27%. Not yet clear which of the Democrats will go into the final, but total of the Democratic vote is itself kind of significant.

HM: It is. Valadao is at the moment holding a support in the low 40s. It means he sort of has put a bull’s eye around himself. This does not augur well for his prospects in November. And of the five districts that the Democrats redistricted in response to the five districts that the Republicans redistricted in Texas, this is the closest to a swing district, and I think these results suggest that it will, with the other four, swing Democratic.
I should also add that the raw numbers of voters in this district is far lower than that of any of the other 52 congressional districts in California, which reflects the poverty of most district residents, which is reflected in the rate of Medicaid recipients, and also the fact that there are a hell of a lot of undocumented farm workers in this district who don’t vote. Contrary to the Republican belief that undocumented immigrants flock to the polls and cast ballots, they don’t. And so if I were betting, I would bet that Randy Villegas would win the second slot and go up against Valadao in November and handily win the district.

JW: Meanwhile, the Senate is “in play,” as we have been told many times. And I want to talk a little bit about Maine, because Maine has a primary this week. Tuesday. We do not yet have the results of the primary as we speak, but Graham Plattner is the Democratic candidate. The Democrats cannot take the Senate without winning Maine, without defeating Susan Collins. It’s the only blue state with an incumbent Republican senator.
Now, she does have a few notable anti-Trump elements in her record. She did vote to impeach Trump, but she supported him something like 96% of the time on everything else in the Senate. And just now, she has signed on as the 50th and decisive supporter of the pernicious SAVE Act. That’s the voter suppression proposal Trump has demanded that the Senate pass, which requires proof of citizenship to register to vote. So Susan Collins has made sure this will come to a vote, probably be a tie, and probably JD Vance will break the tie in favor of the Republicans.
Graham Plattner is the Democrats’ candidate. He’s been leading slightly in the polls, but in the last week, he’s been attacked by the New York Times and The Wall Street Journal. Tell us what this is about.

HM: Initially, this was about his sexting several women in the period immediately following his marriage. And then it became the accounts of several women who said he came on to them way too much, was abusive in what dating they had with him. And one of those women–who is now a Republican Politico–said, alleged, that he had gotten really physically abusive with her. No one else has said that. And given where it comes from, there has to be some degree of at least skepticism about this.
But let’s also keep in mind that Susan Collins voted to confirm all three of Donald Trump’s Supreme Court nominees, who voted to essentially revoke Roe v Wade. Graham supports Roe v Wade and supports reinstating American women’s right to abortions. So if the question is, who can you rely on to support women’s rights? There’s no question it’s not Susan Collins, it’s Graham Platner.

JW: Now it’s time for news of the class struggle in California, a regular feature of this broadcast. Let’s talk about the World Cup. The opening for the U.S. team is scheduled to take place in Los Angeles on Friday at SoFi Stadium in Inglewood, and there’s been intense negotiations for the last two days about the threat of a strike at the World Cup opening for the U.S. team at SoFi Stadium in L.A. Where do we stand on that at this hour?

HM: Well, this was the local 11 of the Hotel and Restaurant Workers, which is a storied local and a storied local in defending immigrant rights, in significant part because its members are disproportionately immigrants. And 2000 of them are the food servers and preparers and sellers at SoFi Stadium. So they were having the regular contractual issues with the contracting company about wages and such. And as we speak on Tuesday, they have managed to settle that. However they reserve the right–And this has been their issue from the beginning of this process–They reserve the right to go off the job if ICE agents appear in SoFi Stadium. That is both a principled issue and an existential issue, since there are doubtless some undocumented immigrants among the people who will be preparing and selling and serving food to the people that the attendees at the stadium.  And there will be a lot of immigrants, believe me, this is a Los Angeles crowd–There’ll be a lot of immigrants in SoFi Stadium for what will be the first match featuring the US team in the World Cup.

JW: Harold Meyerson—read him @prospect.org. Harold, thanks for talking with us today.

HM: Always good to be here, Jon.
[BREAK]

Jon Wiener: Can AI write good poetry? Katha Pollitt wanted to find out. She’s an award-winning poet herself, and also an essayist and columnist for The Nation. She also writes for The New Yorker, The Atlantic, and The New York Times. We reached her today at home in Manhattan. Katha, welcome back.

Katha Pollitt: Hi, Jon. Thanks for having me on the show.

JW: Obviously, AI is good at some things, especially finding and summarizing information, which is kind of the opposite of poetry. I asked AI, “do people think AI can write good poetry?” And the answer is yes. I learned that a 2024 study found that non-expert readers could not tell the difference between AI-generated poems and human written ones. And actually, they like the AI-written poems better. So, by this measure, the answer to the question at least “”do ordinary readers think AI poetry is good?” The answer is “yes.”

KP: Well, I just want to say that if people don’t read poetry, as most people do not, then they’re not going to be able to make a good judgment about what poems are good. So, I completely discount the results of your research.

JW: Okay! You set out to run your own test of the ability of AI to write good poetry. Tell us about that.

KP: I did. Well, I was really mad because Anthropic, as you probably know, stole the writing, like 500,000 books and articles, it stole them to train its large language models. And I am due $15,000 from them because of that. And I just am furious that they don’t ask. I would have said, “no, it’s plagiarism to do this without asking.” So anyway, they have to pay all this money, which I mean, I’ll be surprised if I ever actually see this check, but it just got me to thinking about the capacities of AI. So, I asked her to write poems in the style of Katha Pollitt.

JW: A good test.

KP: Last year I did this, and it came out with something like with like a hallmark card greeting. But since then, it’s been taking poetry workshops, and this is where your study shows that it can imitate conventional styles and modes of feeling. Do you want to hear one? Sure. Okay, so, this is what it gets from its imaginary poetry workshops. Free verse and wistfulness are in. So, here’s how one begins:
The meeting runs long as meetings do. / A table of voices, / mostly baritone,/  interrupted by a careful soprano / that learns to fold itself between commas. / Outside, the city conducts its own debate,/  sirens insisting. / Busses sighing dissent. / A woman on the corner /Counting tips like a rosary of small survivals.”
 It’s so bad, but it has the tics and conventions of magazine poetry of our time. The careless mixed metaphor. “Looking out the window.” The knowing nudge. Like saying, “he meeting runs long as meetings do.” As we all know. That kind of thing. It just makes me want to cringe, and if not, throw up. And its subject was also infuriating, which is the subtle thwarting of women in the office and at home. Because later there’s a line “at home, the sink stages its quiet rebellion.”

JW: And pardon me for laughing.

KP: Yeah, I know, and women’s inability to do much about it. And you get one cliché after another. There’s just nothing fresh or original, no wit or zing, no pressure on language. And it’s full of these decorative phrases. I hate this, like “the stubborn verb of living.” That’s it. Sounds poetic, but it doesn’t do any work.
What really upset me, Jon, and still does, I’m getting all upset just thinking about it again, is that this is what ChatGPT “thinks” my poems are like: obedient, sad-ish, “feminist,” but defeated, you know? And then I wrote, you know, “get that woman a dishwasher.” What’s her problem? And this is not how I write, but it has enough echoes of my writing that I was anxious and unsettled all day, and I still am. And you think, is that really me? The me that the world can see, but I can’t? No, I decided.

JW: No!

KP: Perhaps that surprises you? No! It’s like the uncanny valley version of me. It’s like a Stepford writer. And it’s so disturbing that this is happening. So, then I tried another poem. I tried Grok. Now Grok, as you may remember, I was already familiar with it because I asked Grok, which does these horrible deep fakes, to put Queen Elizabeth, Melania Trump, and me in bikinis. That was another shocker, you know, I really don’t want to see. Melania looked great. It’s not surprising. So, I asked Grok to write a poem. And this this was even worse. Okay, Get this:
“We fold the laundry / like it’s a treaty with the state / of being female.”
Now, first of all, “like it’s a treaty” should be “as if it’s a treaty.” And do you fold the treaty?

JW: I don’t think so.

KP: What does that mean? And then it says it often says “I’ll revise this for more urgency and intensity.” It said, and it did that several times. And each time the poem was worse. So, then we get “we fold the laundry like it’s the last thin barrier before the state storms the house.”
And believe me, you know, if the state was about to storm the house, no one would be worrying about folding the pillowcases. I mean, it’s just ridiculous.
And then this was really disgusting: “Somewhere a judge is sharpening his pen like a scalpel, writing rulings that feel like fingers inside us.” Oh. How disgusting. And you don’t sharpen a pen. I mean, not since the 19th century. Or a scout. You don’t sharpen a scalpel. So, you know, it’s just all kind of word fog.
So, then I tried Claude, because that’s the AI of the moment. And so, this is what they get: “outside.” It’s the same thing. It’s “out the window / the street does what streets do./ A woman holds a hauls a stroller up the curb alone./ A man in a good coat looks at his phone./ Two pigeons share a grievance near the trash./ The usual arrangements,/ the usual cost.”
 Now, what does that mean? “The streets does what streets do” — Go from here to there? I mean, what does that mean?
And then okay, so then I asked a whole bunch of other things. So, Grok on the death of Genghis Khan. I thought, let’s get away from this feminism, fake feminism. And so that was “the tents fold. / The horses move on. / The grass forgets.
And I said, okay, let’s try a classic topic of poems, including some of my favorites, daffodils. And then you get, “So here they are again, the yellow hatted hooligans of spring.” Oh, and you know, it’s like this faint, sour fog of exhaustion and depression that hangs over every poem. And what is wrong with them? And I say, you know, did a Claude have its heart broken by a washing machine? Would Grok put ChatGPT in an ugly bikini? It’s where you realize that these machines do not have human feelings, so they’re just faking it.

JW: Well, I asked Gemini, that’s the Google AI — I pay for Gemini Pro in the belief that they’ll give you a better version, a smarter version. I asked Gemini Pro, “do you think AI would be able to write good poetry? “And their answer was–this is the paid answer no—“AI can absolutely write poetry, but if your standard is art, writing that surprises or shifts how the reader sees the world, AI falls completely flat. The mechanics of AI are fundamentally at odds with the requirements of real poetry.”
 I thought this was pretty insightful, and that’s because the core of the large language model, AI, is predicting the most statistically probable next word. Now, that’s kind of the opposite of poetry, which is to surprise and to come up with something new and original. And Gemini Pro understands that. Gemini Pro said “AI naturally gravitates towards cliches. It chooses a safe, poetic word describing the moon as silver or a heart as heavy rather than a strange word.” It can write in the form of whatever form you give it. It can count syllables, so it can write a sonnet or a haiku or iambic pentameter. But Gemini Pro said “it is only aggregating the ways thousands of other people have done the same thing.” So, AI understands why it can’t write a Katha Pollitt poem. But what I wonder is, why do you think ordinary people prefer the AI poem to poems written by humans?

KP: Well, I think the answer is that ordinary people, and here I’m going to sound very elitist, they like things that are predictable. Here’s something that happened to me, the first poetry class I ever taught, I commented on a student’s poem and said, “well, you know, this is a little sentimental.” And he said, “what’s wrong with that?” And I didn’t really have a good answer for that. But some people, most people like things that are predictable, accessible, no surprises, a vocabulary that is you don’t have to look any words up in the dictionary. And that’s because they’re not into poetry enough or at all to experience what’s original and wonderful about it. I mean, look at the books that are the best sellers. They’re usually terrible.

JW: They’re usually conventional.

KP: They’re, they’re completely conventional. They’re conventional in language, if not in, you know, another vampire novel, okay, with very conventional language. So that’s because they don’t know how to do the kind of work it takes to read something that’s a little bit out of their wheelhouse, and they don’t want that. People like things that are very familiar, like lunch. People like to go to chain restaurants because they know what they’re going to get.

JW: And then I asked paid AI what it thinks a Katha Pollitt poem is like.

KP: Oh, God, do I want to hear this? What did it say?

JW: “Pollitt’s poetry pairs precise imagery and restrained craft with intimate, often witty subject matter, turning the ordinary events of life into brilliant, poignant and often funny poems.” Now—

KP: Yes, go on.

JW: Do you recognize any of that?

KP: Well, here’s the funny thing. A poet friend of mine asked what ChatGPT or whichever one it was,  what AI thinks of their poetry, and it came back with all these flattering remarks. And her friend did the same and got more flattering remarks. And that’s because it’s just like when they’re when their computer psychiatrist, they are trained to flatter you. That’s what they do. They’re not going to say, “well, actually,” and so I think that they just have a, a bias toward making you feel good, except when they’re trying to get you to kill yourself. It also can happen.

JW: Well, this quote I asked, then I said, “do you have a source for this information?” Turns out it’s from the Amazon website description of “The Mind Body Problem” your award winning first book.

KP: Okay.

JW: They just found it because that’s what they’re very good at, searching the web. It’s not very hard to go to Amazon and look up Katha Pollitt and see how her work is described. So, it’s not really their idea, it’s what they found that is not wrong in some, couldn’t be wrong, because it’s the publicity for your book.

KP: That was my second book, actually.

JW: Oh, sorry. Okay.

KP: But yeah, I mean, that’s the thing. They are scraping the web and finding just very accessible things. I mean, they could have gone to the web and found like a terrible review and put that up. It was very discouraging about it is that, you know, people say, I’m not going to use it for this. I’m not going to use it for that. But fast forward in 20 years, will people still be saying that? I don’t think so. I mean, already, students just are cheating unbelievable quantities. Now they have people trying to submit manuscripts that are AI. People admit that they’re, well-known writers who admit, “oh, yeah, I use them to help just to help me get a first draft.” A first draft is where you do some very important work. It’s not just getting an outline. It’s thinking. And AI cannot think. So, it’s very discouraging to me because I think in the history of the world, people tend to choose the easier thing. And so, each technology, whether or not it’s as good as the one before, tends to take over. For example, CDs are out. I mean, first vinyl was out, now CDs are out and Spotify is in. Spotify is a very inferior product. But it’s easier than a CD. You don’t have to put something in something else.

JW: “AI can’t think” – Katha Pollitt wrote about AI’s attempts to write poetry in her column for The Nation. You can read it at thenation.com. Katha, thank you for your wit and zing, and thanks for talking with us today.

KP: Oh, thanks for your wit and zing also — And thanks for having me on the show.