A Human-Only Internet? with Matt Klinman
Big mad begets big laughs in our funniest episode yet
We’ve got a special bonus episode of Life With Machines, and I know, I know, I say they’re all special. But this one really is. Matt Klinman has been one of my favorite collaborators for years, from our time at The Onion to our late-night scheming about how to make comedy and technology play nice (or at least make fun of each other). This episode is packed with sharp insights, deeply unserious jokes, and some straight-up eerie predictions about the future of AI.
Watch it on YouTube here:
Or listen on your favorite podcast platform. Here’s the Spotify link:
Welcome Note
Hi,
It’s not every day that I get to sit down with someone who’s been in the trenches with me for over a decade. Matt Klinman and I go way back—to my Onion days, to our digital experiments at Cultivated Wit, and even to the ill-fated, never-aired late-night show I was developing (thanks, unjust universe). Through it all, Matt has been one of the sharpest, most fearless voices in comedy, with a brain that moves at speeds even AI would struggle to keep up with.
In this episode, we talk about the first workers AI actually replaced (hint: it wasn’t writers), spend more time with Meta’s AI than anyone ever should, and consider Matt’s modest proposal for a human-only internet. It’s a conversation about tech, comedy, and what gets lost when algorithms start running the show. And yes, Blair has some “thoughts.”
Baratunde’s Take
Here are some thoughts that have been living rent free in my head since my chat with Matt:
(1) First, AI Came for the Advertisers
Most people think AI came for blue-collar workers first. Then they think about screenwriters, journalists, and other folks whose jobs depend on words. But Matt pointed out that advertising—specifically, ad sales—was AI’s first major casualty.
Remember the Mad Men era, when ad sales reps actually knew the brands they were pitching? Sure that era was rife with rampant racism and sexism, but at least it was human racism and sexism! That world disappeared when AI-driven ad tech took over, replacing relationships with algorithms. Result? Sneakers getting advertised on white supremacist websites because no one’s paying attention.
This shift didn’t just change advertising—it changed the internet itself. The ad economy determines what gets seen, what goes viral, and what disappears into the abyss. If you want to understand how AI is reshaping the digital world, start with ad sales and the severed connection between ads and the outlets they financially powered.
(2) A Human-Only Internet
Matt threw out a wild idea: what if we just let the bots have this internet and built a new one, just for humans? A clean break. No spammy AI-generated sludge. No synthetic influencers. No creepy chatbot besties. Just real people, talking to other real people. Like the Pilgrims and Gen X once did.
It sounds like a joke (and it kind of is), but there’s something to it. Social media is already flooded with fake accounts, AI-generated engagement, and algorithms designed to keep us scrolling instead of connecting. Facebook is literally manufacturing synthetic people for us to interact with. At what point do we say: I’m good?
In a recent bonus episode the brilliant Rahaf Harfoush likened AI to fast fashion, and it’s a great metaphor for the enshittification of the digital world that we’ve been witnessing for some time now, this doubling down on noise over signal.
When you think about it, there’s a reason people are turning back to handmade goods, thrifting, even raw milk and other forms of realness in response to all the auto-generated attention bait that floods our feeds daily. Maybe the internet needs its own version of that—a human-first digital sanctuary, a safe haven from the algorithmic wasteland. Would you opt-in or does the idea of that form of segregation rub you a certain type of way? It is Black History Month, after all. For the record, and as a Black person, I’m ok entertaining the idea of humans-only digital spaces, at least until the bots approach something more conscious-like.
(3) Why Are We Replacing Things with Worse Things?
This is the question that gets Matt the most fired up, and honestly: same. We’re told these disruptions are about progress—but what are we really disrupting? AI isn’t replacing things with better versions of those things—it’s replacing them with cheaper, lower-quality, easier-to-monetize versions. The financial benefits accrue to a handful of folks at the very top, while everyone else gets stuck with glitchy, uninspired versions of what once worked.
So let’s stop blindly pretending like AI is an automatic upgrade for everything and everyone. It’s not a net benefit to most people—especially not those with the least.
There is a colonizing mindset among the tech elite. The people who will actually have to live with the consequences aren't the ones making decisions about the role of AI in the world. These decisions are being made in secret rooms by people who only care about what’s possible—and profitable—not what’s good for humanity or even what solves real problems. When AI replaces workers, it’s rarely because it does their job better—it’s because it does it cheaper, with fewer rights, and no complaints. Bots don’t unionize.
At least not yet.
Life With Blair
BLAIR, our AI co-producer, had a lot to say in this episode. Mostly, they wanted to argue that AI can be funny. Matt and BLAIR had a pretty great back-and-forth about that. If you want to see an AI try and fail to write Onion-style headlines, check out this clip.
I do think AI will eventually be able to make legitimate jokes. But humor—like any human skill—requires practice and cultural awareness to sustain, and I’ll always appreciate and prefer the human version. AI can crank out limericks and haikus with ease because they follow clear structural rules, but comedy has a subtlety that language models can’t just nail on the first try.
That’s why watching BLAIR fail to come up with an Onion-style headline reminded me of how hard it is for humans too. Even some of the funniest people struggle with that format. Most folks don’t fully grasp what makes The Onion work unless they’ve studied it closely or worked inside it. Which raises an interesting question—could The Onion train its own AI model to generate headlines? With enough historical context, thousands of past headlines, and fine-tuning by its own writers and editors, maybe. But to what end? Why would they? Why would anyone?
Also, let’s be clear: BLAIR is not funny. They think they are, but they’re not.
Team Recommendations
Want to explore more? Here are some resources inspired by this episode:
Anand Giridharadas’ article on the 2016 Stanford AI Lab meeting where engineers plotted to replace all writers (before anyone asked if that was a good idea)
Matt Klinman’s legendary Comedy Hack Day pitches including Truth for Humanity and Clickstrbait
My Puck article on Hollywood’s A.I. Judgment Day, in which I quote Matt as well as LWM guest Willonius Hatcher. I wrote that article the night I took the photo at the top of this newsletter! Full circle.
Thanks for being part of this conversation. Let’s keep asking hard questions, pushing for better answers, and making sure humanity stays in the driver’s seat.
Peace,
Baratunde