Spring TV has been awash in artificial intelligence.

No, not the utilization of AI, but the inclusion of AI as a plot device, as fear of the burgeoning technology has made its way from the subconscious of paranoid writers to the screens of differently paranoid viewers.

Not surprisingly for shows written in the aftermath of the dual industry strikes in which concern about AI’s encroachment into the creative process was a key point of dispute, AI needs to get its publicist on the phone because these aren’t the headlines it needs.

On The Pitt, HBO Max’s Emmy-winning drama about good doctors doing good work under the most trying of circumstances, Sepideh Moafi’s Dr. Al-Hashimi was introduced as a contrast to Noah Wyle’s Dr. Robinavitch. Ultimately, she wasn’t presented as an adversary, but Dr. Al-Hashimi’s dedication to AI innovation in medicine has consistently run afoul of the show’s deeply humanist ethos. Although Dr. Al-Hashimi boasted that AI is a major time-saving device — particularly when it comes to charting and transcribing — the characters have rapidly discovered that generative AI isn’t close to 98 percent accurate, at least not in this intense one-shift sample size.

On HBO’s The Comeback, meanwhile, Lisa Kudrow’s Valerie gets her latest chance at a television comeback, but it comes with a catch: The show she’s been recruited to star in will be written largely by AI, with a disgruntled husband-and-wife team (Abbi Jacobson and John Early) trying and failing to shepherd a technology capable of spewing out 50 alternate punchlines, all of them hacky and derived from decades of accumulated (stolen) schtick.

On Amazon’s Scarpetta, the main character’s niece, Lucy (Ariana DeBose), is mourning the loss of her wife (Janet Montgomery’s Janet), and she seems to have become a shut-in. The truth, however, is that  Lucy is spending most of her time in front of her computer because she has a sentient AI version of Janet capable of serving as confidant, therapist and virtual partner. The other characters distrust this Janet 2.0, but as the season progresses, they learn that Janet’s value might be greater than anybody other than Lucy realized.

Each of these shows — not to mention an AI subplot on Scrubs, the dysfunction of AI creators on The Audacity and the regular onslaught of AI moguls who seem to get periodically murdered on broadcast procedurals — understands that AI isn’t a thing to be trusted, but they each approach or even define AI in different ways.

The Pitt, as a well-researched show proud of its layers of accuracy, is giving a reasonably accurate rendering of how generative AI in various forms is being used to modernize the medical space, with the predictable assortment of evangelists and skeptics. The Comeback is just using “AI” as shorthand for “ChatGPT,” finding punchlines in the myriad ways that the program has been found to be glitchy and eccentric, even if the trepidation feels straight out of the industry strike negotiation manual. As for Scarpetta, its version of AI is pretty much science fiction, the latest way to handle the all-too-family “protagonist talks to dead wife/son/goldfish” trope, not wholly divorced from reality but basically a less nuanced version of a dozen different Black Mirror episodes.

On The Comeback, the network executive played by Andrew Scott laughs off the idea that there’s anything wrong with doing a television show written completely by AI: “In our business, AI, well, it has bad branding, right?”

Of course, The Comeback gets something right about AI and Hollywood. The industry is currently in the “catch me if you can” phase, the one where people at various stages of the creative process have been trying to sneak in applications of the technology, either hoping or assuming that nobody will notice and then issuing limp explanations or justifications when obsessive fans notice. Which has begun to happen.

Marvel’s Secret Invasion, a limited series that managed to squander the collective talents of Samuel L. Jackson, Don Cheadle, Olivia Colman, Ben Mendelsohn and more, wasn’t getting particularly good buzz surrounding its 2023 release even before fans accused its image-morphing, green-tinted credit sequence of being AI. It was, indeed, AI-generated, and producers backed into candor by saying that the decision to utilize the off-putting imagery was to capture the alienating and identity-hopping nature of the show’s Skrull-infiltrated world. Sure. Humans have never been able to produce art capturing shape-sifting unease.

Ultimately, the show was so negligible that the controversy went nowhere. And you might not even remember the miniature tempests that arose when Netflix’s Ted Sarandos acknowledged that the Argentine science fiction epic The Eternaut utilized generative AI to deliver special effects faster and for less money than a more traditional approach would have required.

Confirmation bias prompted only a minor kerfuffle when fans watching One Last Adventure: The Making of Stranger Things 5 discovered — or thought they’d discovered — that a shot of a writer’s computer featured a web browser with ChatGPT in several tabs. Nobody was exactly able to articulate if this was necessarily a bad thing or why it was bad, but it lined up with the countless scenes of writers seemingly unable to crack several story problems with the series’ final season.

I’m more disturbed, personally, by AI’s encroachment into nonfictional space. In 2024, Netflix’s What Jennifer Did was criticized for allegedly using AI-generated or manipulated photos, a charge that the true-crime documentary’s producers denied. Blurring of fact and fiction in documentary storytelling dates back to Nanook of the North, so this isn’t earthshaking, but added to the trend of using AI-created voices — sometimes for well-known celebrities — in documentaries, it’s getting harder and harder to believe what we see and hear.

So far, we’re in the dribs-and-drabs stage of our awareness of AI usage on television. While Chinese television saw the premiere of the wholly AI-produced Qianqiu Shisong, a series of 26 episodes of seven minutes apiece, back in 2024, nothing comparable has launched on a domestic network or streaming service.

That, presumably, is why there was a brief uproar earlier this winter with the release of On This Day … 1776, a shortform series from Darren Aronofsky’s AI-focused Primordial Soup, streaming on Time’s YouTube channel. The historically based series was the latest example of how unprepared we are for discussing what is or isn’t AI, since it combined SAG-AFTRA actors and AI visuals but utilizing human animators and required people who understand what Google’s DeepMind technology actually is or does.

In truth? I don’t know. All I know is that On This Day … 1776 was and continues to be awful, a mishmash of misapplied cinematic grammar and dead-eyed photorealistic famous characters that is neither entertainment nor educational. Episodes have tended to be shorter than five minutes and yet feel interminable and violate one of the few rules in which some people are willing to allow AI, namely if it allows us to tell stories that weren’t possible to tell using more traditional storytelling devices or technology. Instead, it asked the question, “What if Ken Burns’ The American Revolution was produced with the artistry of a video game, the humanity of Robert Zemeckis’ The Polar Express and the historical depth of a virtual puddle?”

Nobody is exactly sure in what ways Aronofsky did or didn’t collaborate with AI for On This Day, but for the AI-cautious, the sense of betrayal is real whenever somebody we think of as an analog creative comes forward saying they’re jumping on the AI bandwagon — see also under Natasha Lyonne’s dalliances with AI and Ben Affleck’s Netflix-purchased AI startup.

It’s no wonder that in The Comeback, nobody is supposed to know that Valerie’s show is AI-scripted. It becomes a vicious circle in which nobody in traditional entertainment is proud to say they’re using AI because anybody who attaches their name to a venture gets pilloried, so that makes audiences all the more sensitive and attuned to find things that point to AI, presumably aware that for every two or three things that prompt brief social media hand-wringing, there are probably a thousand smaller things that are slipping by.

When it comes to AI and television, here’s what I know, or at least what I think I know: AI can make a Tilly Norwood, but it’ll never give a performance as good as Ethan Hawke in The Lowdown, the accumulation of four decades of a flesh-and-blood actor acting and evolving before our eyes. AI can make “off-putting” credits like Secret Invasion, but it’ll never make 90 seconds of pure joy like the credits for Pachinko. AI might allow you to make special effects that look like a video game cut scene or turn around a sitcom pilot full of hacky punchlines, but the digital assistant you saved a few bucks not to hire or the writer’s assistant you decided you didn’t need could be the next Ray Harryhausen or Norman Lear. Or that’s what I think I know, today.