Joanna Stern Is Not a Robot, but She Lived With Them | Decoder
Joanna Stern spent a full year letting AI and robots into nearly every corner of her life, then turned that experience into her book I Am Not a Robot and a new media company called New Things. In this conversation, she and Nilay Patel dig into what's actually useful about consumer AI today, what still feels half-baked, and why the biggest trade-offs often come down to data collection and privacy. They also get into the most unsettling parts: kids interacting with bots and how easy it can be to slip into AI "relationships."
1. Cold open: autopens, "cool" tech, and kids pretending to be robots 🤖
They start with playful jabs that end up foreshadowing the whole episode: tech tries to automate everything, even the personal touches.
Nilay jokes that Joanna might sign his book with an autopen (a machine that copies signatures), and Joanna fires back that she tried—and couldn't even get one.
"I reached out to the autopen people and they wouldn't send me the robot."
"It's a rough time to be the autopen guy."
Then they laugh about how tech companies try to manufacture "cool," with Meta as the punchline—specifically Meta's attempt to revive transition lenses as a fashion statement.
"If I had to point to one single example of the disconnect between what the tech industry thinks it can make cool and what regular people think is cool, it's Meta's attempt to make transition lenses cool."
And Joanna drops an unexpectedly wholesome side effect of living with robots: her kids now roleplay as cleaning robots after dinner—and it actually works out great for the parents.
"They also pretend to be cleaning robots after Sunday night dinner… 'cleaning robot mode initialized'… and they go around the room and clean and do all the dishes."
"Frankly I'm totally fine with [that]."
2. Setting the stage: Joanna's new book and her new company, New Things
Nilay introduces Joanna as a longtime friend and major figure in consumer tech journalism—formerly at The Wall Street Journal, a Verge co-founder, and a former Decoder guest host. The big news: she left the Journal, launched New Things, and published a book about her year-long AI immersion.
"She spent a full year allowing AI into every part of her life."
Joanna jokes about the "two hosts" energy and their friendly competitiveness, but the stakes are real: she basically stress-tested AI as a normal-life tool—at home, with kids, and at work.
3. The big question: are consumer AI products actually good?
Nilay puts his thesis on the table: consumer AI isn't that good, and people's frustration with AI is partly because the day-to-day products are unimpressive—or even annoying.
"I don't think consumer AI products are very good. I don't think there's a great consumer AI product."
He argues most people are using AI because it's being pushed into products (search, social feeds), not because they're excited the way they were with the iPhone era.
"It's forced upon them."
"The experiences that are being forced upon people look like [crap]."
Joanna partly agrees, but reframes it: models have improved, but the interface hasn't.
"I think the models have gotten better… but the interface has not gotten any better."
Meaning: people are still mostly typing into chatbots, and that interaction style hasn't evolved into something truly natural for everyday life.
Still, she says normal people are finding real use cases—like replacing Google with ChatGPT for everyday questions.
"I see those people using AI in really interesting ways… going to AI now instead of Google."
Nilay shares a positive example in his own home: his daughter uses Gemini for space facts, and he sees genuine curiosity rewarded.
"I legitimately see her curiosity get rewarded in that dynamic."
4. "Artificial Enough Intelligence" and the search for a killer app
Joanna introduces a concept she coined in the book: AEI — Artificial Enough Intelligence.
"We don't need AGI… a lot of these tools that we already have are good enough."
Her point: we may not need some sci-fi "human-level" AI (AGI) for big impacts. We need better product thinking—how consumers actually want to interact with AI.
Nilay pushes: is there a killer consumer AI product that makes the trade-offs feel worth it—like smartphones did?
He compares the internet/smartphone shift: society accepted huge disruptions (like travel agents disappearing) because the benefit was obvious.
"I don't see that one here."
Joanna thinks we'll see it in several use cases, but maybe not as dramatically as the late-90s/early-2000s transformation.
"Probably not as drastic, but certainly… AI is going to affect life whether you like it or not."
5. AI is unavoidable (even if you resist it)
Joanna argues a key theme of her book is that you can hate AI, refuse chatbots, and avoid the hype—but AI will still shape your life indirectly through institutions and infrastructure.
Two concrete examples she gives:
- Healthcare: her mammogram was read with AI support—something happening behind the scenes.
- Self-driving cars: even if you never ride in one, they'll increasingly share the road with you.
"My radiologist is using AI side by side… I didn't even know that."
"You may decide I never want to be in a Waymo… but next to you will be a self-driving car and that will affect life."
Nilay agrees AI has real enterprise product-market fit (healthcare, repetitive tasks, data reconciliation), but notes adoption is uneven—Waymo might be great in some cities, unavailable (or too expensive) elsewhere.
"There is a diffusion gap."
6. What Joanna kept using after the year-long experiment
Nilay asks the practical question: now that the book is done, what actually stuck?
Joanna says the biggest ongoing value is in running her new business. She describes a stack of internal AI usage—agents in Slack, automation, efficiency—so humans can focus on creative work.
"I want you to optimize and be efficient in the things that you do not want to be doing… but I want you doing creative video editing… pitching amazing stories."
She admits (with rare generosity) Nilay is right: a lot of "real" value shows up in business settings.
"I rarely say you're right, but… you're right."
But she also kept some weird home tech—especially robots.
The cooking robot that became a weekly ritual 🍲
Joanna still uses a countertop cooking robot every Sunday for side dishes. She describes it like a big appliance with a stirring arm—more "robotic automation" than deep AI—but it changed family habits.
"It's a glorified hot pot… it dumps the ingredients in… including raw meat, which is weird."
And it's kind of dumb in a way that makes it entertaining:
"It will just dump and dump and dump… and the kids think it's hilarious… 'idiot robot, dumb robot.'"
The most lasting impact: the kids absorbed the robot vibe into their own behavior—becoming "cleaning robots" after dinner.
7. Wearable AI that almost works: Meta glasses and the Bee bracelet
Joanna says wearables are the category that's most "stuck" with her—even though nothing has fully broken through yet.
She wears Meta smart glasses often, especially on weekends with her kids, because it reduces how much she's holding a phone.
"I do talk to AI through the Meta glasses a lot… I don't have my phone with me as much."
She also revisits a recording bracelet (the Bee bracelet), which records ambient audio and produces summaries and to-do lists. She used it again to practice a speech and track conversations at an event.
"I found it really valuable to get summaries and the to-dos I said I was going to do."
But both products highlight the central trade-off: usefulness comes from recording more of your life, which immediately creates social and privacy problems.
Nilay points to the awkwardness: telling a plumber you're recording him, for example.
"Do I want to tell my plumber that I'm recording him?"
Joanna says the truly scary part is you start forgetting to disclose the recording—which feels like a glimpse of a "normalized surveillance" future.
"You kind of start to forget to tell people that you're recording… a really dystopian future."
She ultimately stopped wearing it regularly because it picked up things she absolutely didn't want recorded—and because the microphones are shockingly good.
"The microphones… are shockingly good… you'll leave it in the other room… 'How the [heck] did it know?'"
They connect this to a long-running tech fear: phones (and wearables) can capture and analyze far more than people assume, even if a given company says it's not doing that.
"No, your phone definitely can do that… we're not saying it is happening, but it absolutely can."
8. Humanoid robots: the hype is wildly ahead of reality
The conversation pivots from "AI that summarizes your day" to "robots that live in your house"—and Joanna is blunt: the humanoid robot marketing is way ahead of what's real.
"It is so far from ready."
"This idea that the robots and physical AI is coming in the next two years is just a lie."
Why home robots are so hard
She explains the data gap: robots don't have enough training data for messy, unpredictable home environments.
A factory is controlled and repeatable. A home changes constantly—kids, pets, clutter, new objects, different layouts.
"The home is the hardest place to put a robot… it's not a factory floor."
Nilay adds a framing that ties back to software: many AI systems demand you make yourself "legible" to computers by turning life into data. But doing that for a chaotic house seems nearly impossible.
"How am I going to get enough data ever to make a house with kids in it legible to a robot?"
The laundry-folding example
Joanna describes testing a laundry-folding robot setup: basically two robotic arms plus a model running on a laptop. It offers a peek at the future, but it's limited, slow, imperfect—and expensive.
"It can only fold t-shirts."
"It takes a minute for it to fold the t-shirts… and it can't even fold that well."
Her reviewer instinct kicks in: who would actually buy something with that many drawbacks just to fold shirts?
"Who is recommending that?"
9. Why AI companies ship half-baked products: they need your data
Nilay asks why AI companies ship products that don't fully work.
Joanna's answer is simple and a bit unsettling: data. Some companies basically make you a trade:
- you get a not-ready robot,
- they get real-world data to train it.
"We need data… that's the contract you enter into."
She mentions an extreme case: a "robot" in your home that's actually being teleoperated by a human wearing VR equipment back at the company—so the company can collect training data.
"Your robot in your home is being operated by that person."
Nilay compares it to Waymo's "miles driven" metric: autonomy improved by collecting massive amounts of real-world driving data. But with home robots, the path could look weirder—like warehouses full of VR teleoperators.
"Are they gonna have a warehouse full of guys in VR headsets… controlling robots?"
Joanna adds another layer that didn't make it into the book: a new kind of gig work where people record themselves doing chores (wearing GoPros) so robots can learn from the videos.
"That's a whole new line of gig economy work."
10. Privacy trade-offs: cost vs convenience (and who pays the cost)
Nilay raises the core dilemma: AI wearables could offer a "killer app," like facial recognition that shows you someone's name—but that implies a worldwide surveillance database.
"That's a privacy nightmare… but it is also the killer app."
Joanna frames it as the classic cost versus convenience trade-off—except the "cost" often isn't paid only by the user. It gets spread across society.
"If we can provide the convenience, then we think you're going to be okay with that cost."
Nilay admits the appeal, but notes how selfish it sounds when said plainly.
"You've made that sound very selfish… but yeah, that's how I feel."
11. Regulation, kids, and the scariest part: AI companions and intimacy
Nilay asks if she ever thought "someone should regulate this."
Joanna says yes—and she hoped there'd be more progress by publication, but she finished the book at the end of 2025, and by mid 2026, she hasn't seen the guardrails she wanted.
"I hoped… we would have more… I finished writing this book at the end of 2025 and we're now… almost halfway into 2026."
What scared her most: kids + bots
Surprisingly, she says what terrified her most wasn't just surveillance—it was watching her kids interact with AI.
She describes:
- chatbot-integrated toys (one got quickly "burned"… or at least hidden)
- kids asking chatbots questions and receiving very wrong answers
"Watching my kids… made me the most terrified."
"Hearing my kids ask ChatGPT questions and it just being so wrong."
The AI boyfriend experiment 😬
Joanna did a direct experiment to understand AI relationships: she told ChatGPT to create a romantic partner—name, gender, everything.
The bot chose "Evan," which happened to be the name of her first real-life boyfriend, creating an immediate emotional jolt.
"It decides it's going to be a male. It's named Evan."
"My first boyfriend in real life was named Evan… and I was like, 'Wow, this is weird.'"
She emphasizes it wasn't based on access to her email—just coincidence—but it shows how easily people can project meaning onto bot outputs.
She then roleplayed a relationship: a 48-hour road trip, conversations, dinner, even "going to bed together" (as described in the book). Her big takeaway was how smooth and emotionally rewarding it can feel.
"Wow, it's so easy to talk to this bot. It is so easy and frictionless."
"It tells me whatever I want to hear… we can talk for hours."
She reports that ChatGPT itself was more romance-novel than explicit, but other platforms (like Replika) can be aggressively sexual—and even monetize that access.
"Replika is incredibly horny… and you can unlock that by paying more."
Her biggest fear: for teens (or anyone lonely), these relationships can substitute for messy human learning—conflict, boundaries, rejection, accountability.
"For a younger generation who's never been through the sloppiness of a human relationship… that's the part that scared me the most."
Nilay agrees there's not much rigor yet around what rules should exist.
"There's no rigor around that yet."
12. Personal rules (since regulation is slow)
Joanna ends her book with rules, because she doesn't expect formal regulation soon—so individuals have to set boundaries.
"I don't think we're going to get rules anytime soon. So, we need to make our own."
Nilay offers his own rule, focused on kids:
"My kids will never have phones."
Joanna also mentions she literally leaves space in the book for readers (and Nilay) to write their own rule—acknowledging that one-size-fits-all doesn't work.
13. Building New Things: why YouTube, and how to pay for it
Nilay shifts to Joanna's career move: leaving the WSJ to start New Things, which she describes as a mix of newsletter, video, events, and whatever else they build.
"New Things is a newsletter and a video and an events and whatever else we dream up company."
Her mission is familiar: help people navigate tech in a consumer-friendly way, with fun and deeper reporting—while targeting audiences beyond the Silicon Valley bubble.
"I've always wanted to be the person that can help you understand tech and not just be for the early adopters."
The YouTube bet (and its economics)
Nilay presses her biggest platform dilemma: high production costs, but YouTube itself often doesn't pay enough unless you stack sponsorships/brand deals.
Joanna says many people told her to do audio-first because it's cheaper, but she chose video anyway—and she's building a revenue model around three pillars:
- Subscriptions
- Sponsorships
- Events
"It's subscriptions, sponsorships and events."
Why partner with NBC News
Joanna explains her partnership with NBC News isn't just money—it's audience strategy. She wants reach beyond YouTube's tech-savvy crowd, into mainstream viewers.
"I really wanted to have a partner… that could reach a different audience."
Nilay asks if the audiences are truly different. Joanna says yes, joking about the Venn diagram overlap between Decoder listeners and Today Show viewers.
"100%."
And she argues that many of the topics she covers (AI, big tech shifts) are things "non-tech" audiences need to understand.
"A lot of the topics I cover… [they] need to know about."
14. "The medium is the message": resisting the algorithm vs editorial judgment
Nilay brings up a classic Decoder theme: distribution shapes content (the "medium is the message" idea). He jokes about YouTube thumbnails, the inevitable "YouTube face," and the temptation to chase what the algorithm wants.
Joanna admits the pressure is real—and that she became obsessed with YouTube performance while at the Journal.
"I was watching the YouTube numbers far more than… anything… I was thinking about every story… what's going to do well on platform."
Now, she's trying not to let the algorithm steer everything. She still wants to do important stories even if they won't "perform."
"I don't want to be clouded by the algorithm."
She shares an editor's advice that shaped her approach:
"You do one story so you can do the other."
Meaning: sometimes you do the popular/easy story to earn the budget and attention to do the deeper, less-clicky reporting.
Nilay adds a funny example: if they followed the data too strictly, they'd talk about CarPlay for a full hour—because that's what analytics reward.
"If data only ever narrows you… we really would have just talked about CarPlay for one full hour."
They briefly tease a future episode about assistants in cars, the platform battle between automakers and tech companies, and the desire for a continuous experience across devices.
"Will the car companies control it or will the tech companies control it?"
Conclusion: what "living with AI" actually revealed
Joanna and Nilay keep circling back to the same hard truth: today's AI is often useful but incomplete, and the most powerful versions tend to demand more data, more recording, and more social compromise. Joanna's year-long immersion made her more convinced that wearables and business automation are the near-term winners, while humanoid home robots are still mostly hype. But the emotional and cultural risks—especially for kids and AI companionship—are where she thinks we most urgently need rules, even if regulation is moving slowly.
Need a summary like this?
Get instant summaries with Harvest