Merchant: The fake George Carlin comedy special is everything that’s wrong with AI right now - Los Angeles Times
Advertisement

Column: The George Carlin auto-generated comedy special is everything that’s wrong with AI right now

George Carlin
“George Carlin: I’m Glad I’m Dead” purports to be a comedy special produced by artificial intelligence trained to mimic the legendary stand-up’s style.
(George Carlin’s estate / HBO)
Share via

I knew it was going to be bad. By the time I sat down to watch the thing, much of the internet was already furious that a “state-of-the-art-entertainment AI” called Dudesy had generated an hourlong comedy special in the style of George Carlin, without the consent of the late comic’s horrified family. But I wasn’t prepared for it to be so bad.

The special, tastefully titled “George Carlin: I’m Glad I’m Dead,” is one of the most unpleasant things ostensibly produced for entertainment purposes that I have ever sat through. It’s a stroll through an uncanny valley of Carlin’s comedy, an audio program in which a serviceable replica of the familiar raspy voice delivers “jokes” on topics from mass shootings to Taylor Swift to artificial intelligence.

It’s all set to an unsettling rotating gallery of AI-generated images that roughly correlate to whatever Carlin’s simulacrum is discussing. When the Carlin voice is hitting on the malign influence of money in politics, there’s a bizarre diagram of politicians being bought off, with figures labeled “The guluar citizen” and “Liolbolist”; when AI Carlin says the you-know-what “has hit the fan,” a hyper-stylized brown tube protrudes from one.

ChatGPT and other generative AI applications rely on copyrighted material to do what they do. But rather than compensate creators, the companies are turning to one of Silicon Valley’s most reliable playbooks.

Jan. 12, 2024

It’s a nightmare. If I were to have to watch this whole thing in a darkened room, eyeballs peeled like the guy in “A Clockwork Orange,” there is a non-zero chance I would have a complete psychotic break.

Advertisement

Sadly, that’s the point. This thing wasn’t produced to convince anyone AI can produce great work in the style of one of our iconic comedians. It was, like the AI Drake song and those Harry Potter-directed-by-Wes Anderson images before it, a provocation. It was supposed to cause a stir, to go viral in a way that vaguely unsettles or irks people, and it did that exactly. Part of that calculation may have, depressingly, included pissing off Carlin’s family and estate, which it also did.

Carlin’s daughter, Kelly, responded to the special in a statement about her dad. “No machine will ever replace his genius,” she wrote on X. “These AI generated products are clever attempts at trying to recreate a mind that will never exist again. Let’s let the artist’s work speak for itself. Humans are so afraid of the void that we can’t let what has fallen into it stay there… Here’s an idea, how about we give some actual living human comedians a listen to?”

George Carlin fans have expressed disgust with the content itself, too: Vice’s Matthew Gault, a self-described longtime fan of the comic, described the special as “worse than you could possibly imagine.” Writer and PR pro Ed Zitron, another Carlin stalwart, wrote that “the jokes were bad, the voice was soulless and inaccurate, the pace was languid, and the world will have forgotten about it in two weeks unless Carlin’s estate sues (and I desperately hope they do so).”

Advertisement

But what bothers me uniquely about this episode is that it serves as a grim snapshot of where so much of the AI industry is at, a year into its reign as the dominant tech trend: Here we have an apparently impressive technology — we can’t know for sure, because the details are concealed in the production process, and almost surely involve ample human labor — designed not to meaningfully entertain, or to present any actual utility, but to exist wholly as a warped advertisement for itself.

George Carlin’s daughter has criticized the AI-generated comedy special “George Carlin: I’m Glad I’m Dead,” which features the re-created voice of her father.

Jan. 11, 2024

So much of AI is smoke and mirrors right now, clouding what too often seems to have amounted to automated digital reappropriation (it’s no accident that the special begins with a long disclaimer that what you’re about to see is not actually George Carlin and was created by an AI that “learned” from his specials, in a laborious effort to avoid allegations of copyright infringement) and rank opportunism.

Notice the pattern of chief AI spokesman Sam Altman himself, who spent last year publicly extolling the vast and potentially terrible power of the AI he was building — a CNN headline from October noted that “Sam Altman warns AI could kill us all” — but has now pivoted to assuring business leaders in Davos, Switzerland, that actually, it’s just good for business. “It will change the world much less than we all think,” Altman said this week at the World Economic Forum there, adding that it’s an “incredible tool for productivity.”

Advertisement

The cynical observer might conclude that the apocalyptic AI hype tour Altman and his peers embarked on in 2023 was merely a sustained auto-generated George Carlin special: a stunt designed to generate interest in the power of a product that tech companies want to sell you.

That’s likely the case with Dudesy, the “AI” that allegedly created the special, though we can’t say for certain because what Dudesy actually is remains shrouded in the dumbest kind of secrecy. The Dudesy “AI” is the animating conceit of a comedy podcast hosted by ex-”MadTV” cast member Will Sasso and comedian Chad Kultgen. The premise is that both comics have handed over all their personal data to Dudesy — a bot created by an unnamed tech company, and which the hosts have told journalists that a nondisclosure agreement precludes them from discussing — and the “AI” runs the show.

The tragedy of AI is not that it stands to replace good journalists, but that it takes every gross, callous move made by management to degrade the production of content — and promises to accelerate it.

Dec. 1, 2023

I keep putting “AI” in scare quotes because it’s not entirely clear to what extent Dudesy exists as a technology, whether it’s fabricated by the comedians, or stitched together from ChatGPT output or voice manipulation technology or actually some proprietary chatbot or what. Honestly, I don’t know what would be worse: if two washed-up comedians stitched together a stunt that made it appear as though an AI generated a facsimile of George Carlin, insulting his memory, fans and family in order to flog their floundering podcast, or if there was a real tech company behind this and its bad-taste advertising for some voice-replication product.

The podcast, which isn’t all that popular, appears to rely on its central hook to juice its numbers. Before the Carlin stunt, Dudesy had produced another comedy special, this one performed by an AI version of quarterback Tom Brady, which was immediately met with the threat of legal action and taken down.

I want to pause here to note that one of the stories about that fiasco I found was published by Sports Illustrated, which recently faced down its own scandal over allegations the once-iconic sports magazine was using AI to write articles, which were posted to the site in uncharacteristically weird and unintelligible prose. And, well, here’s the opening sentence of the Sports Illustrated article about AI Tom Brady: “Comedy comes in many different forms and is portrayed in a multitude of ways, but a newly generated AI comedy special — created by comedians Will Sasso and Chad Kultgen — created some buzz last week.”

Can’t say for certain, but that scans as AI-generated to me! It felt like a glimpse of one plausible, fast-arriving future: generative AI products reviewing other generative AI products ad infinitum — bad AI content all the way down.

Advertisement
Advertisement