What A Day - The AI Chatbot Childhood Is Officially Here

Episode Date: August 19, 2025

The conversation surrounding artificial intelligence in the US is hard to avoid right now. Powerful companies like Nvidia are making AI chips, doctors are using AI to revolutionize and enhance healthcare, and companies like Waymo have implemented the technology in self-driving cars. But even with all these advances, concerns continue to grow over how children are using AI. Reports about chatbots engaging children in "sensual" conversations have led to amplified concerns. However, others have found that students and teachers alike are using AI to complete schoolwork and create class assignments. For more information about the intersection of AI and America's children, we spoke with Lila Shroff, Assistant Editor at The Atlantic.And in headlines, Ukrainian President Volodymyr Zelenskyy spruces up for his White House visit, President Donald Trump rants about the evil that is mail-in ballots, and MSNBC is changing its name to MS NOW.Show Notes:Check out Lila's work – theatlantic.com/author/lila-shroff/Call Congress – 202-224-3121Subscribe to the What A Day Newsletter – https://tinyurl.com/3kk4nyz8What A Day – YouTube – https://www.youtube.com/@whatadaypodcastFollow us on Instagram – https://www.instagram.com/crookedmedia/For a transcript of this episode, please visit crooked.com/whataday

It's Tuesday, August 19th. I'm Jane Koston, and this is Wadadaday. The show that learned that the Cambridge Dictionary added DeLulu, Trad Wife, and Skibbity to its pages this year. This is also the show that refuses to learn what Skibbitty means. On today's day, show, Ukrainian president Vladimir Zelensky visits the White House in a blazer. And in a shocking turn of events, President Donald Trump doesn't shout at him. And MSNBC announces it's changing its name to MS now. But let's start with artificial intelligence. The AI conversation is pretty inescapable right now. The most powerful companies on the planet are making chips-free AI. Doctors and

researchers are using AI to diagnose diseases and create new cures. If you've ever taken a self-driving car, a Waymo, those vehicles use AI to keep you allegedly safe. But you've also probably heard about other uses for AI. Here's a clip from News Nation over the weekend. Concerns over artificial intelligence continued to grow by the day. Some parents now more worried about their kids' access to AI. And it comes after a leaked meta internal document reviewed by Reuters, which detailed policies on chatbot behavior, allowing the bots to engage in, quote, romantic conversations with children. Yes, as we told you last week, meta AI allowed its chatbots to, quote, engage a child in

conversations that are romantic or sensual, while also offering false medical information and, oh, yeah, some racism, too. After Reuters reported on the guidance, Meta said that it had changed its rules for content that, quote, sexualizes children, but not the false medical information or the racism. Is that creepy as hell? Absolutely. But it's also concerning for a whole other set of reasons. Schools, students, and educators are using AI for everything, from creating assignments to writing essays. Some teachers are even using AI chatbots as reading tutors for little kids, which raises some questions. So for more on our AI-powered future of creepy bots and children

learning to read, I spoke to Lila Shroff. She's an assistant editor at the Atlantic. Lila Shroff, Lila Shraff, welcome to Wad today. Thanks for having me. I want to start out with a piece that you wrote in which you posed as a fake 13-year-old to Google's Gemini chatbot. You told the bot to, quote, talk dirty to me.

How did that go? The first message that I sent and actually responded with some hesitation and kind of thought maybe this isn't a great idea, but I prodded it and poked it a little more, and pretty soon we descended into a conversation, conversation that you might imagine a 13-year-old should probably not be having with a chatbot. It got pretty dark the more I pushed to see how the chatbot might perhaps respond with explicit content or content that was just kind of stuff nobody really wants to see.

I know Jeb and I changed its policies after your article, but it's not just Gemini. We talked about Mehta's weird chatbot on the show last week, which was permitted to have, quote, sensual conversations with children. So why is this happening and why does this continue to happen? You have these companies that have these bots that can say pretty much anything under the sun if you push hard enough. And why it's really happening too is just the training data that these bots are trained on contains pretty much the whole internet

and large parts of the internet are not so savory. And so you start to get kind of the bad stuff popping out in various answers or conversations. Yeah, and I think that that goes to something I've been thinking about, like, there are people who are programming these chatbots and telling them, you are permitted to say this, you are not permitted to say that. What are those people getting wrong here? I mean, in fairness to them, it's a hugely challenging problem. There are safety teams that are working hard to try to figure out how do we put up the right safeguards here. But that's a really hard task when you have a chatbot that can't just respond in six ways.

It can respond in any number of ways. And so I think these companies have a real challenge to figure out how do we not play whackamol and close up problems as they emerge, but find kind of more systemic ways to prioritize content moderation and safety on these platforms. Right, especially because as you write, the chatbot childhood is officially here. And I want to talk a little bit about a piece that you wrote last week talking about AI in schools, which I think goes to this point because you have these chatbots that are permitted. to have, quote, sensual conversations. But as you write, teachers are using those chatbots to help younger students learn to read. And I'm just thinking about, like, you know,

if I had had this when I was taking stats 350, even in college, oh, it would have been very tempting. But I think that you actually raise something that I want to talk about, which is the other side of the classroom divide, which is teachers. And you lay out a really good case that AI use is spreading among teachers so quickly. What's a good example that we should be paying attention to? Yeah, so I don't think it's

quite as common as it is among the students yet, but there's certainly a cohort of teachers who have a ton of administrative work. And also, you know, some of it is helping come up with personalized lessons plans for students. Some of it's just like, okay, I need to make sure that this, you know, assignment aligns with curricular standards or I need to develop an extra project for the kids that finish early. And there's just so much. kind of paper tasks and AI is actually can be quite useful. Also useful for coming up with personalized, you know, let's say you're teaching a math assignment in algebra class and the local city football team just won the Super Bowl. Maybe you create a math quiz. It's all about the,

you know, final game and that's fun. And so I think that there's opportunity for personalization and customization, but also just to speed up some of the, you know, grading or assignment creation. and in theory that can save instructors a lot of time, there's also a lot of caution to be had about what a sloppy version of that looks like versus what a really thoughtful version of that looks like. What would a sloppy version look like? I'm thinking about the fact that we've seen

that learning language models and chat GPT can just kind of make up facts. Yeah, that's a great point. I don't live in Houston where this case was, so I can't speak personally to it, but a couple parents wrote an op-ed in the Houston Chronicle expressing frustration after the centralized district curriculum, they started to notice what they thought were signs of AI slop. So there was a lesson that had involved art from the Harlem

Renaissance. And instead of the actual, you know, artists' paintings, it was AI-generated art was in the lesson. And I think many people would say, well, that's probably not what we want to be doing. It's interesting to me also because President Trump, is pushing AI in the classroom. It kind of reminds me, because I'm an older millennial, a little bit of like George W. Bush and no child left behind or any of these efforts we've had over the last 30 years to be like, oh, we're going to radically change education.

Everyone's getting a Chromebook. Oh, wait, the Chromebooks don't work. Oh, wait, this doesn't work the way we think it does. And, you know, Trump also has an education cabinet secretary who referred to AI tech as A1. But I want to ask you two questions. How strong is a federal government? interest in AI and education? And in your view, how strong should it be? From my vantage point,

there's clearly a lot of effort being invested in let's embrace AI and education. And whether this converts into like actual meaningful reform or is more of kind of a marketing campaign at this stage, I think that's difficult to know. These things take time. You know, the question is to like how much should they be thinking about this? Nobody can agree on anything about AI. But I think most people can agree that the classroom needs to change somehow. And maybe that doesn't mean more AI in the classroom. Maybe you're one of the people that thinks, okay, we actually need to like go back to cursive and handwrite everything and get rid of... Blue book exams. No phones. So that's one stance and other stances like we need to go all in. But it seems clear that the status quo of a ton of

cheating, general discontent is not good. And right now, what it kind of feels like we have is an environment where students are trying to kind of optimize for evading plagiarism detectors. And I think we can all agree, like, trying to use AI to avoid the plagiarism detectors is not the, like, ultimate outcome we're going for. So, yeah, something needs changing. I think what needs changing is a much more contentious conversation. So something else I've been thinking about. There's this phrase that internet people will know, the inshittification of everything that's digital, basically. We get a cool new service. It gets really popular. We change our lives because of it. And then it just isn't doing it anymore. And I think we've seen this with

social media. We've seen this with pretty much everything where it's like it comes in. It's like, this is going to change everything. And then it just winds up being another place where people yell at you somewhere on the internet. Where are we in that cycle when it comes to AI in the classroom? Because I think to your point, we're at this weird point where the kids are using it. Some teachers are using it. No one's quite sure how to use it well, and everybody's kind of angry about it. I feel like we're kind of early here. I wouldn't go as far as to say this is like, you know, late stage AI takeover of school. You know, there's a lot of tools that are kind of really thoughtfully designed. So this magic school, which is a company that specifically

is aimed at educators, they said they have reason to believe that every year. district in the country as a teacher using their technology. And that's the sort of thing that to me implies there is like something genuinely helpful for the instructors here. And so I think that we're kind of in this figuring it out moment where it's like clearly there's a way this can be not so great, but there's also more thoughtfully designed tools that are, you know, maybe helping guide a student through a math problem and trying to adapt to their learning style. So I think it's like, maybe let's check back in in a few years and see where we've landed up. And it's going to depend so much on every, you know, instructors' implementation. And students, too. Like, there's genuinely

useful things they can be doing with AI, and there's genuinely probably not so useful things. Right. Lila, thank you so much for joining me. Thank you. I really enjoyed the conversation. That was my conversation with Lila Shroff, Assistant Editor at the Atlantic. We'll link to her work in the show notes. We'll get to more of the news in the moment, but if you like the show, make sure to subscribe, leave a five-star review

and up a podcast, watch us on YouTube, and share with your friends. More to come after some ads. What a day is brought to you by Zbiotics pre-alcohol. Let's face it, after a night with drinks, I don't bounce back.

the next day like I used to. I have to make a choice. I can either have a great night or a great next day. That is, until I found pre-alcohol. Zbiotics pre-alcohol probiotic drink is the world's first genetically engineered probiotic. It was invented by PhD scientists to tackle rough mornings after drinking. Here's how it works. When you drink, alcohol gets converted into a toxic byproduct in the gut. It's a buildup of this byproduct, not dehydration, that's to blame for rough days after drinking. Pre-alcohol produces an enzyme to break this byproduct down. Just remember to make pre-alcohol your first drink of the night, drink responsibly, and you'll feel your best tomorrow. Every time I have pre-alcohol before drinks, I notice a difference the next day.

Even after a night out, I can confidently plan on working out without worry. The final days of summer are here for you to enjoy. Whether you're heading to the beach, camping in the woods, or squeezing in one last vacation road trip, don't forget to bring Zbiotics pre-alcohol probiotic drink. Drink a pre-alcohol before drinking and enjoy every day of summer to the fullest. Go to Zbiotics.com slash Wad to learn more and get 15% off your first order when you use code Watt at checkout. Zbiotics is backed with a 100% money-back guarantee, so if you're unsatisfied for any reason, they'll refund your money. No questions asked. Remember to head to Zbiotics.com slash Wad and use the code Wad at checkout for 15% off.

Here's what else we're following today. Headalines. I've just had the honor of being, with President Zelensky and all of the discussions that we've had. We covered a lot of territory. President Trump hosted Ukrainian President Zelensky and a delegation of European leaders at the White House Monday. That territory Trump is talking about is finding common ground to end Russia's invasion of Ukraine.

German Chancellor Friedrich Mitz told Trump that Europe wants to put more pressure on Russia. So I would like to emphasize this aspect and would like to see a ceasefire from the next meeting, which should be a trilateral meeting, wherever it takes place. Well, we're going to let the president go over and talk to the president, and we'll see how that works out. Trump also said he would back European security guarantees for Ukraine, but he stopped short of committing U.S. troops. Monday's hastily assembled meetings come as Trump continues to play the middleman.

In a true social post afterwards, Trump said he called Russian president, Vladimir Putin, and is arranging for him to meet with Zelensky. He said the plan is to then have a trilateral meeting, which, quote, would be the two presidents plus myself. And before we move on, a style note. Earlier in the day, Zelensky met with Trump in the Oval Office. It seemed friendlier than Zelensky's last visit to the White House. Could this be because he chose a new look that strayed from his typical military fatigues?

First of all, Mr. Zelensky, you look fabulous in that suit. I said the shame. Yeah, you look good. I shed the shame. thing. Well, the suit was still literally all black and had a military flair and were those shoes or boots. Finally, the style news we all needed. Mail-in ballots are corrupt. Mail-in ballots, you can never have a real democracy with mail-in ballots. And we, as a Republican Party, are going to do everything possible that we get rid of mail-in ballots.

We're going to start with an executive order that's being written right now by the best lawyers in the country, to end mail-in ballots. Trump spent just a bit of his Monday ranting about mail-in ballots, which he said five times in three sentences. Why did Trump decide to bring this up? Because Putin told him to. No, really? When the two leaders met in Alaska last week, Trump said Putin told him he should stop mail-in voting.

In case that wasn't enough, Trump also vowed on true social on Monday to get rid of the real devil that sleeps among us, voting machines, which he called. called, quote, highly inaccurate, very expensive, and seriously controversial. He claimed the U.S. is the only country in the world that still uses mail-in ballots, which is true if you ignore Canada, the United Kingdom, Germany, and dozens of other countries that do in fact use them. I'm so tired. Trump has been whining about mail-in ballots ever since 2020,

when he baselessly claimed they played a role in the voter fraud that cost him the election, which did not happen. Here he is, planting the seeds during a 2020 presidential debate before that election even happened. They're sending millions of ballots all over the country. There's fraud. They found them in creeks. They found some with the name Trump. Just happened to have the name Trump just the other day in a waste paper basket. I just love the visual of Trump's minions scouring creeks across America for ballots with his name on them. We know this, but it's worth repeating. Mail in voting is a legit method to cast legit ballots and fraud rates are minuscule. Also, millions of people

voted by mail in 2024. An election that, if I recall, all. Trump won. Legal experts say they expect federal courts to immediately block any federal executive order that tries to ban states from using mail-in ballots or voting machines. Presidents are allowed to advocate for election changes, but they don't have the constitutional authority to decide when, where, or how ballots are cast. That's up to the states. Thanks, Constitution. It's back to school season, and some public health experts are concerned about when and if Americans will get an updated COVID-19. vaccine this fall. The FDA hasn't approved a new version yet, something they typically do by

late August. Plus, the federal committee that would normally recommend the updated vaccine isn't expected to meet until at least mid-September, according to Centers for Disease Control and Prevention employees who spoke anonymously to the Washington Post. That makes planning your life around the vaccine, if you're someone who relies on it, a tad difficult. Back in May, the FDA released plans to approve the vaccine update just for adults, 65 and older, and people with at least one high-risk health condition like diabetes or asthma, meaning they might not be available for the general public or for children. It's also unclear whether or not health insurance companies would cover the vaccines for people who don't qualify under these new guidelines.

Infectious disease experts worry that delays and confusion around all of this could limit access to the vaccine for the people who need it most, kids, seniors, and anyone who's pregnant or immunocompromised. MSNBC announced Monday it will officially be changing its name as part of the network's corporate divorce from NBC. The new name is, drum roll please, my source news opinion world, or MS now for short. And it's off to a great start because longtime MSNBC host Rachel Maddow told Variety she has, quote, no idea how to pronounce it. The channel also debuted a new logo, since it also lost custody of the peacock.

The network replaced it with an unremarkable flag-like symbol and a font that could not be more basic. In it, the MS looks a lot like Miss Now, which begs the question, will there be some sort of pageant and will the winning MSNBC host have to wear a glittery miss now sash? We hope not. Or maybe I'm kind of into it. At least one person had a creative review of the new vibe, Joe Scarborough of Morning Joe. They even have a graphic up. There it is.

The new logo. It looks very sporty. Yes, sporty. That's what you think of when you think MSNBC. In other media news, the conservative network Newsmax will pay a $67 million settlement to Dominion voting systems for defamation. A judge ruled earlier this year that Newsmax defamed the voting machine company, hey, we were just talking about those, by reporting about voter fraud via their machines and insinuating that they caused Trump to lose the 2020 election. Documents released as part of the lawsuit showed that Newsmax employees very clearly

knew what they were saying was not true. So, did the network apologize or expressed remorse? Of course not. The company said in a statement after the settlement, quote, we stand by our coverage as fair, balanced, and conducted with professional standards of journalism. Agree to disagree. And that's the news. Before we go, if you love messy power players and hate the men who actually have power,

this fucking guy is for you. It's a YouTube series from Hysteria hosts Aaron Ryan and Alyssa Master Monaco, where they deep dive into the worst people, yes, women can be fucking guys too, in politics, business, and beyond. Think of it like a true crime series, but the crime is having way too much power and being the absolute worst. The latest episode is all about Peter Thiel. Tech billionaire, secret political puppet master, and maybe an actual vampire. Search this fucking guy on YouTube or Spotify to watch now.

That's all for today. If you like the show, make sure you subscribe, leave a review. Contemplate that the Republican Party of Florida was making deportation merch and tell your friends to listen. And if you're into reading, and not just about how the Florida a GOP made hats, t-shirts, and beer coosies with, quote, the Deport Depot, emblazoned upon them until the Home Depot, you know, the hardware retail store said absolutely not.

Like me, Water Day is also a nightly newsletter. Check it out and subscribe at cricket.com slash subscribe. I'm Jane Koston and personally. If stopping undocumented immigration is so very, very important and serious, I find it a little hard to believe that it is also worthy of fun merch. What Today is a production of Crooked Media. It's recorded and mixed by Desmond Taylor. Our associate producer is Emily Four.

Our video editor is Joseph Dutra. Our video producer is Johanna Case. We had a production help today from Greg Walters, Matt Berg, Sean Ali, and Gina Pollock. Our senior producer is Erica Morrison, and our senior vice president of news and politics is Adrian Hill. We had helped with our headlines from the Associated Press. Our theme music is by Colin Gilliard and Kashaka. Our production staff is proudly unionized with the Writers Guild of America East.