The other day I did a post about Artificial Intelligence. In the comments on the accompanying Instagram post, CC DeVere of Frenchtown Confidential hipped me to this commiserating post of hers, wherein she’d asked ChatGPT about French influence in early Los Angeles, and was rewarded with a bunch of nonsense.
Apparently that sort of thing is pretty common. Don’t worry, though, because we’re told when ChatGPT is “hallucinating” the absurdly counterfactual claptrap it produces accounts for only a mere 40% of its output. Mankind’s relationship to ChatGPT is thus:
Inspired by CC’s post, I gave ChatGPT a whirl, and, um, wow. Actually, a mere 40% of its output being nonsense would be a huge improvement.
I asked it some stuff about Bunker Hill, which spat out answers that — if I’d written them? If I’d written them as satire, attempting to lampoon how bad ChatGPT is? — you’d be like “Oh, Nathan, stop hating on ChatGPT with your hyperbolic, preposterous, melodramatic depictions of modernity’s decline!“
So, I’m just going to provide you some screengrabs, and you be the judge. I asked ChatGPT about Bunker Hill’s houses, and architecture, and some of its history, and … well, when wee Timmy and Sally have a report to write, they’re going to receive a big fat “F”, should they rely on ChatGPT…oh wait, since teachers are uniformly terrible (well, regularly terrible; my apologies to the good ones) and use AI for their confirmation bias “research,” I’m sure Timmy and Sally will get A+s and gold stars. And so the cycle continues.
And since I have previously delved into Chavez Ravine and Cooper Donuts and the Great American Streetcar Conspiracy, I ChatGPT’d in that muck too, being awarded more ludicrous falsehoods, proving ChatGPT is an equal opportunity fabrication machine.
Anyway, let’s begin with—
Bunker Hill
First I asked ChatGPT about some of the families who settled the Hill. Let’s see what it had to say!
Now, there were some fine families on the Hill; they had names like Brunson and Rose and Bradbury and Crocker. So, who the hell are these people? Well, William Workman was a real person, but he lived out in the Rancho La Puente. But then there’s…John MacArthur? For whom the “nearby park” is named? The MacArthur Park Mansion on Bunker Hill? If you’re on this blog in the first place I’m going to assume you know that’s all manner of crazy business. And as long as we’re taking crazy pills, what about “Henry Slauson”? Do they perhaps refer to famed L.A. developer Jonathan Sayre Slauson, who lived down among the mansions around 23rd & Figueroa?
It goes on:
Again, a mixture of real people (Griffith J. Griffith and William Mulholland, neither of whom ever lived on Bunker Hill; while the chatbot admits this, the song and dance about their influence on Bunker Hill is silly, e.g., in the mid-80s when Bunker Hill blossomed, Mullholland was a mere superintendent at LA City Water Co.) and completely random made-up people, like “William G. Perkins” and “Eugene Spence.” So then I moved on to architecture:
Damn, that’s some wild hallucinatin’. There was a Mills family on the Hill, but their house at 327 South Olive was a Queen Anne. The rest is just…weird.
Holy cow I love ChatGPT so much. I shouldn’t have to tell you that City Hall and the Bradbury Building are not on Bunker Hill, but the Huntington Hotel? Yes there’s a Huntington at 8th & Main but … tremble! at the authority with which it states Cal Plaza (Grand to Olive, 2nd to 4th) now sits on the site.
I asked it some other general questions, about the history of redevelopment, much less hallucinatory, but still flawed. ChatGPT kept telling me Bunker Hill’s redevelopment began in 1955, though its redevelopment began in 1949 and there were dozens of major milestones between 1949 and ’55; it also kept telling me Bunker Hill’s redevelopment disproportionately impacted the “notable population” of Black families, making it “an important chapter in the history of housing injustice” despite there being no African-Americans displaced from Bunker Hill.
I could have gone on and on with Bunker Hill, but I was eager to head over to
Chavez Ravine
Less “hallucinatory” overall, but still wrong. (You’ll note that ChatGPT gets much of its information from Wikipedia; as I’ve demonstrated on this blog a dozen times, Wikipedia is pathologically unreliable.)
People believe, and will repeat, that on May 8th 1959 Aurora “Lola” Vargas was dragged from her home, a belief that persists largely due to this photograph:
It is one of the most influential, if not the most influential and widely-seen images in all Los Angeles history. But Aurora “Lola” Vargas wasn’t dragged from her house because she didn’t live there — and in fact never lived there. In the 1940 census (before her parents even moved to 1771; they were renters on the next block north through the 1920s and ’30s), she lived with her husband Porfidio at 1201 Curtis Street. After Porfidio died in the war, Lola remarried Eliason Escobar Colón and moved to his place at 1841 Malvina. Lola Colón & family were at 1841 in the 1950 Census, but soon moved out of Chavez Ravine and bought a house — in fact, two houses — on Simmons Avenue, about midway between East Los and Montebello. Come May 1959 it was from her East LA home, ten miles away, that she drove to her parents’ house in Chavez Ravine, on this famous day. She arrived late, after all the media circus had arrived and announced to all in earshot “You’re going to have to carry me out!” and ran up into the house.; the deputies just shrugged and got on with it, while the television cameras gleefully whirred. It may seem a small and trifling point that her parents’ 450-square-foot house was merely “in” her family and not hers, but, given the absolute media saturation surrounding the belief she was carried from “her” home, I think it’s an important point to make.
So, we’re told that after the elder Arechigas and daughter Aurora had their house demolished (though, again, Aurora never lived there), the elder Arechigas lived in a tent “for several months” because “the eviction made them homeless.” ChatGPT further informs us that this fact fuels important discussions about displacement and housing justice, but don’t worry, ChatGPT can help you phrase your parroting of these points so they don’t sound romanticized or misleading.
The problem being, of course, said statements are in fact romanticized and misleading, because they’re not true. Don’t get me wrong: after the May 8, 1959 eviction and house demolition, Manuel and Abrana did pitch an old army tent in the ruins. But why? They weren’t homeless: like their daughter, they also owned two very nice houses elsewhere. And they weren’t camped in the Ravine for “months”; rather, it was ten days before they folded up their tent and decamped, after losing all public support when it was discovered they owned more property than literally 98% of Angelenos. When they left Chavez Ravine, they made a beeline straight to one of their other homes, specifically, the one on Ramboz Street, a house they’d bought brand new in 1956, and which, by 1959, they’d already paid off more than half the mortgage; that’s why they’re listed here in the March 1960 phonebook. (As long as we’re on the subject, daughter Victoria, living in her parents’ other Chavez Ravine home at 1767 Malvina, wasn’t made homeless either: she just went to one of her three houses on Allison Ave.)
The Chavez Ravine story has about a million moving parts, and I could go on, but instead let us move on to —
Cooper Donuts
Did you know ChatGPT can actually draw maps for you? Neither did I, until it offered to, and I said yes please! Here’s one:
Ummm. So if you’re familiar with the streets downtown, you’ll know this makes no sense whatsoever. I mean, it doesn’t make sense a lick of sense regarding the Cooper Donut story, specifically: because if we’re at the corner of Second & Main, we’re nowhere near any drag-friendly or hidden gay bars, least of all Harold’s and the Waldorf, which were some blocks south. But more to the point it doesn’t make a lick of sense geographically in general: if you run north from First Street there’s no Second and Third, rather, it’s Temple and then the freeway. That’s all the Land of City Hall north of First.
This ChatGPT-provided timeline is even more fanciful. Remember, there’s only one person — John Rechy — who ever claimed to have been a witness to this event. His tale and his alone sets canon, because there exist neither contributing stories, nor contributory bits of evidence. Not surprisingly, ChatGPT’s timeline contradicts Rechy’s account. Rechy stated the event occurred on Main between Fifth and Sixth, which is like six blocks from Main and this mythical AI-concocted “Second Street north of First Street” in the timeline. Rechy said it happened after 2:00am, after the bars had closed, not at midnight, thus there were no “quiet streets” at 1:30, especially after, canon instructs us, LAPD kept the area under heavy police lockdown till dawn. And of course, I hope I don’t have to mention, Rechy said it didn’t happen at a Cooper’s.
ChatGPT seems to be under the sway of the post-Rechy apocrypha, from which we have the sanctification and hallowing of Second and Main as part of Cooper Canon. The City of Los Angeles’s June 2023 consecration of Second and Main is based wholly on apocryphal texts — these apocrypha regarding Cooper Donuts are rife with doctrinal errors (analogous to, say, in 2 Maccabees and the Additions to Daniel) and historical inaccuracies (like in the books of Judith, and Tobit) … sorry, I can’t help but use this analogy since the Cooper Donuts at 215 South Main was right across from St. Vibiana’s. Point being, we can argue whether post-Rechy apocrypha is noncanonical or deuterocanonical (which would at least put the Tony Hoover narrative in an appendix) but we can at least all agree that ChatGPT is crap. Seriously, now I wonder if ChatGPT also eagerly endorses other hoaxes like the Donation of Constantine, the Vinland Map, The Protocols of the Elders of Zion, and those 1983 Hitler Diaries. I’d check but instead I’ll move on to:
The Streetcar Conspiracy
Auuuuuugh ok. I know you know where I’m going with this: it’s all baloney-wrapped malarkey. I could write a 5,000-word response to ChatGPT’s assertion here but, you know what, I already did, so go read that.
*******
Bottom line, of course, is you should rely on primary sources, not an OpenAI hallucinator, which is a symptom of tech’s arrogance, not a tool for truth. I’m sure I surprise no-one when I assert ChatGPT is useless for research and poisons education, accelerating cultural decay by making nonsense authoritative.
Having made my point repeatedly many of you may, of course, roll your eyes and spit “way to pick at the low-hanging fruit, lazybones!” so this will be the last of my “shooting fish in a barrel” attacks on Artificial Intelligence. Rather, at some point in the near future I’ll combat John H. M. Laslett’s Shameful Victory. It’s written by a PhD! From Oxford! And published by a university press! And holy hell in a handbasket — spoiler alert — it’s crap. See you then!
Bunker Hill, Los Angeles 