AI Is Breaking Our Brains

0:00In today's world, data breaches happen all the time. And even the most secure companies can't always protect their employees' personal information from ending up in the wrong hands. That's where Delete Me comes in. Delete Me is a service that removes your employees' sensitive information from hundreds of data broker websites. Sites where hackers can find phone numbers and emails within seconds. Rachel Toback, CEO of Social Proof Security, says attackers use this data to target employees with phishing messages and AI-powered phone scams. but Delete Me makes it harder for these bad actors by scrubbing your employees' details regularly.

0:38It's simple. Attackers are lazy. If it's too hard to find contact info, they'll move on to easier targets. Delete Me takes care of this for you, doing the heavy lifting so you don't have to. And over time, they keep removing the information so it stays down, protecting your team from constant exposure. If your business has a social presence or deals with clients, you need DeleteMe. Visit DeleteMe.com slash 404media and start safeguarding your team's information today. That's DeleteMe.com slash 404media.

1:18Hello and welcome to the 404media podcast where we bring you unparalleled access to hidden worlds both online and IRL. 404media is a journalist-founded company and needs your support To subscribe, go to 404media.co. As well as bonus content every single week, subscribers also get access to additional episodes where we respond to their best comments. Gain access to that content at 404media.co. I'm your host, Joseph, and with me are 404 Media co-founders, Sam Cole. Hi, Joe. Emmanuel Myberg. Hello. And Jason Kevlar. Hey, good to be here. Was that your Jason voice? Yeah, it was. How was my Joseph voice?

2:04It was perfect. It needed to be a little bit more British, but pretty good. Yeah. Yeah. As you can probably tell, Joseph is not here. I'm Jason. But you get the B team today. We are going to start out with Emmanuel's story. Microsoft study finds AI makes human cognition, quote, atrophied and unprepared. this is one of those stories that we published that the response to was kind of mixed I feel where there was a bunch of people saying hey dumbasses of course we already knew that however I thought it was quite interesting and I think that anytime that a company is pushing a technology but then also publishing studies saying that their technology is causing some problems, I find it to be very interesting.

2:54So what is this paper? How did it work? What did they find? So I want to start by reading a quote from the paper itself, which I think really sums up the point. And it says, a key irony of automation is that mechanizing routine tasks and leaving exception handling to the human user, you deprive the user of the routine opportunities to practice their judgment and strengthen their cognitive musculature, leaving them atrophied and unprepared when the exceptions do arise. So that is what they found. This is the gist of the paper. I think that's very well phrased. In terms of how they actually measured this, they recruited 319 knowledge workers from one of those this is an interesting story on its own but there's these various platforms where people can sign up and volunteer to participate in this kind of research so they use one of those to recruit 319 knowledge workers and then those knowledge workers reported 936 first-hand examples of them using generative AI.

4:07And then the researchers provided them with a questionnaire to reflect on how using AI impacted or did not impact their critical thinking. And the paper goes on quite a bit on how they prepped the people who participated in the study, which I think is good because a lot of these terms are vague and complicated, but they explain what their definition of critical thinking is, made sure that the people who participate in the study understand that, and explain what they're looking for. And a lot of interesting stuff to talk about here. I want to start with some of the actual data, which the study did not make a big deal out of, but I found to be probably the most terrifying thing about it, which is just what people report that they use AI for, and their daily jobs.

5:07So they cite some examples. So there is a teacher who is using DALI, which is an AI image generator, to make a presentation about handwashing in school. There's a commodities trader who is using ChatGPT to kind of train themselves on their job and become better commodity traders. And there is a nurse who verified that a chat GPT generated educational pamphlet for newly diagnosed diabetic patients. So that alone, I think, is already kind of shocking that all these professions are using generative AI like that. um but yeah what these people reported and what the study found is that the more a worker relies on generative ai the less critical thinking they use in their job and also there's a correlation between workers who are more confident in themselves are less confident in the generative AI output and therefore feel like they have to use more critical thinking and the inverse is ultra true.

6:26So if you have not a lot of confidence in your abilities, you just trust the generative AI more, rubber stamp it and let it go. Yeah, it's interesting. they have a few little excerpts in this study where they have quotes from people. And one of them is, quote, I can be confident that everything is spelt correctly. I don't need to second guess myself. I can get the reassurance I need without having to bother another person to check it for me. And that's obviously a very small thing where they're just asking it to spell check something. But I did find it interesting to see what people were using it for.

7:03the other thing that this study says I believe if I understand it correctly you wrote this story but like as I understand it the people who are using generative AI are rather than being creative and you know creating things from scratch which is how we used to do things and how we still do things here they're like correcting the AI so they're sort of like managing the output of this other of this technology and that's like a different type of work um can you elaborate a little bit on on that like it's a different type of work more or less yeah maybe one of you can look up the exact word that they use which i thought was really good but essentially what you're referring to is the difference between someone doing the job and someone's job being verifying the output of a generative AI tool, which is, yeah.

8:06Yeah, here's the quote. It's, the data shows a shift in cognitive effort as knowledge workers increasingly move from task execution to oversight. Oversight, right. Yeah, and that's kind of a scary vision of the future where all these jobs, you're not actually doing the job. You're kind of making sure that the AI is doing the job correctly that you previously did. you emmanuel had this um like allegory more or less uh where you said like i don't remember anyone's phone numbers anymore and you just use you know you're they're saved in your cell phone and that's not the cognitive you don't have that like cognitive load anymore you've like offloaded it to technology and that's not artificial intelligence but it seems like you know saying that someone's cognition is atrophied and unprepared, that seems quite bad.

8:57But it does seem like it is not fully a dire picture that they're painting here. Can you elaborate a little bit more on the phone number analogy? Yeah. So I think you and I were texting about this study that we both saw over the weekend. And I think the reason that we thought it was interesting is because the study itself, let me see what the actual, I'm going to read the title of the paper. It's The Impact of Generative AI on Critical Thinking, Self-Reported Reductions in Cognitive Effort and Confidence Effects from a Survey of Knowledge Workers, which is obviously many words, but that is like a pretty spicy title for a paper from Microsoft, which is heavily invested in the success of generative AI.

9:48And we can talk more about what that means for all this to be coming from Microsoft researchers. But to your question, and the researchers point this out at the top, and this is something we talked about before we wrote the story, is that there is a need, we both felt the need to hedge here and make this seem maybe less dramatic that even the title says because there is a long history of various technologies that we offload cognitive tasks to, and it seems outrageous at the time, but then becomes completely normal, and we all assume is good. So the researchers point out calculators, right? It's like calculators didn't used to exist.

10:36People thought that it would make us bad at math, and obviously that is not the case. Math has advanced tremendously. People are probably overall better at math than they used to be before the invention of the calculator. People said the same thing about the internet itself. This I didn't know, but they noted that Socrates really objected to writing becoming commonplace. And that would make people dumb because they wouldn't have to remember stuff, I guess. I did not know that. Right, me neither. But it's like, yeah, we've all felt this. And I think the millennials in the audience probably remember a time before smartphones and not having Google Maps on your device.

11:16And I talk about in the story how when I moved to San Francisco, it was right before smartphones caught on. And I would leave the house with a little pocket map. And I would learn how the city works and public transportation works. And I would remember all of that. And in a few years, I kind of knew the city very well. And since then, every time I moved to a new city, I never went through that process because I just had Google Maps. and I never learned... I really can't navigate New York that well without Google Maps. And is that bad? I don't know. It's probably not great that my navigational skills have atrophied, as they say in the paper.

11:59But overall, people are better at navigation. And one could easily imagine generative AI, should it catch on, should it work as well as the people who run AI companies tell us it will, offloading various tasks and us getting completely used to it and then it becoming an overall positive. One can imagine that world even if the prospect of offloading all these tasks right now seems scary, especially if you're a nurse or whatever. Yeah, I mean, I find it to be pretty scary regardless. I do. I take your point. And I think just as a counterpoint, I didn't memorize my partner's phone number until like six months ago.

12:51And we've been together for seven years. and that's very bad. But I can still remember a lot of phone numbers from my childhood. Like my next door neighbor's phone number, I still remember it. And I do think that the maps one is a good thing to bring up. And I find that when I purposefully try to navigate without Google Maps or something like that, I feel my brain working in a way that it doesn't normally work when I just blindly follow the map. And I also find that if I go somewhere without Google Maps one time, and I have to figure it out myself, it downloads into my brain in a way that doesn't happen if I do it 30 times with Google Maps.

13:40I'll go to the same place and just blindly follow the map, and I'll have no idea how I got there, what happened during the ride. There's an entire blank space in my brain. And I do wonder if you start offloading all sorts of different things to generative AI in some way, what that will look like over time when you just start taking all of these things off, like what is left. and the paper does bring that up like it says effects on writing and they're like well people know how to write now and the people who use generative ai for writing who already know how to write they will you know like be able to edit what what is what's coming back so they'll be able to do that oversight process but they sort of worry that in the long term like it says however there are concerns that novice writers may become overly reliant on these tools, potentially impairing their long-term skill development by bypassing critical writing processes, such as constructing logical arguments and understanding subject matter.

14:46That seems like a real potential problem. I just like, so two things. One, which is addressed in a paper. And that is, that goes back to what does it mean for this to come from Microsoft? And I think that is a company that is heavily invested in AI, recognizing potentially a real and big problem. And eventually, at the end of this paper, coming around to a solution, right? Saying, we recognize... And we see this with other things, right? Like, when we write about Google doing research about AI misinformation, the point of that, at the end of the day, is for Google to say, we recognize this is a problem.

15:33This is a result of a technology that we are developing and deploying. Here's what we suggest we can do to mitigate this problem so we can develop this technology further. And in this case, essentially, what Microsoft says is that the generative AI tools need to, first of all, encourage critical thinking, even though they are doing the work. So they say the AI tool can suggest areas for user refinement or offer guided critiques, like invite the user to examine the output, which is not something we see AI tools currently do. When OpenAI gives you an answer, it is kind of doing it in a definitive voice with authority.

16:19And that's where things can get really hilarious when we talk about Google AI overview telling people to eat glue. That would seem less bad if Google in some way was like, hey, maybe you should look into this further and not just trust the output. And then the other thing that is not addressed in the paper, but I think is something that has come up in our reporting a little bit. For example, when I wrote about the CEO of Suno talking about how nobody likes making music actually, and we should just let AI do it all. And that is a more fundamental question of like, what do we actually want AI to do?

17:01Let's assume that generative AI will eventually be able to do all of our jobs. Does that mean that we want it to do it all? Are there things that we don't want it to do because we want to do them just because we enjoy them? Or we think we get better results even though an automated tool can do it? I think these companies that are developing AI know this is a problem. And it's like Jason said, people replied to this story and they were like, oh, duh, like everyone knows this is a problem. But something being studied is much more powerful than people just like kind of anecdotally knowing something.

17:31And when I was writing about Anthropic, Anthropic recently put in all of their job applications, which they have like 150 different open roles right now. So it's a lot of job applications or job, you know, like descriptions

17:49that they're telling applicants not to use AI to fill out the application. And it's like they know that this is a problem. And I was trying to find a study that kind of backs up the idea that everyone knows, which is that relying really heavily on these conversational AIs makes your brain kind of mushy after a while if you're just kind of never using critical thinking. And in their application, it says, don't use AI assistance during the application process. We want to understand your personal interests in Anthropic without mediation through an AI system. And we also want to evaluate your non-AI assisted communication skills.

18:30And Anthropic, if people don't know, makes Claude, which is a very popular conversational AI that you would use for that purpose. Like if it was any other job application, you were like, I just need to whip off why I want to work at Anthropic. I would use Claude to do that probably if I was going to like rely on a chatbot to do those sorts of things. But they're saying, we want to see your non-ai-assisted communication skills which i wonder is is that going to be a sought-after skill in the future if people don't have it if people are losing that ability to like put together critical thought without the use of a computer um is that going to be the new kind of sought-after uh skill in the workforce i don't know that feels like a very far away thing but maybe it's not that far.

19:16This study is not exactly about that. And yet I found myself thinking the same thing. What does this mean for humans? And what does this mean for jobs? And what does this mean for cynically companies who want new ideas and who want to continue making money? and we talked about this before but there was this period of time where a bunch of journalists were getting laid off and as part of the culture war there was like a type of software engineer who would go into people's twitter uh mentions and say learn to code like the the way that it worked was someone would say oh my god i just got laid off from my job advice and then a bunch of dickheads would respond to them, say, learn to code, saying that software engineer was going to be the only place that jobs would exist in the future.

20:14And in a somewhat interesting fashion, software engineers are being automated right now by generative AI. And I think there's a lot of creativity in software engineering. There's a lot of problem solving in software engineering, But a lot of the sort of like commodified coding is being replaced by things like Copilot, Microsoft Copilot. Mark Zuckerberg is talking about replacing software engineers with AI and turning a bunch of these coders into people who just like oversee AI coding bots and things like that. And then in this story we reported on last week, like the federal government wants to replace a lot of its coders with AI coding bots.

20:59And meanwhile, there's like tons of journalists and writers who have been replaced by AI or who have, you know, are competing with, you know, AI that steals content and just, you know, regurgitates it out there. But what we do and what a lot of really good journalists do is they find new information and share new information and share new ideas. And I wonder if the only way that you'll have a knowledge work job in the future is if you can create new ideas with your own brain. Because a lot of the little quotes in here about what people are using generative AI for are like, I just wonder if companies are getting what they paid for because a lot of companies are asking people to do stuff and they're like, Oh, I typed it into the AI and I gave whatever, whatever it spit out.

21:56And so I wonder about like the sameness of everything that's going to start happening as AI does more stuff in like an office setting, I guess. I think, what is the OpenAI research tool? Is it just Open Research that they recently unveiled? Deep Research, I believe. Deep Research. That is the AI tool that I think is closest to what our jobs are. And I haven't used it yet, but I will use it and report back and tell you if we're cooked or not. I suspect not. I was watching the demo and initially I was really freaked out. I was like, damn, that's what I do. But then I thought about some of what you, Jason, said, which is a lot of our stories are really the foundation at its core as an original thought and also talking to people and getting information that inherently is hard to get out of people.

22:58And as far as I know, Deep Research is still unable to do that. But before we move on, can I just very briefly talk about Pod Committed? Yeah, please. I wrote in the story, I used the phrase that Microsoft is Pod Committed to the rapid development of generative AI tools. And I've never gotten more emails about a mistake in my article before. And I was like, did I make this up? And I didn't. It's a real term. Pod Committed is a poker term. I don't even play poker, but I like the metaphor, which is you're sort of so deep in to a bet that the only way forward for you is to keep placing the bet or to go deeper into the same bet.

23:45And I was very surprised that people are not familiar with that. So it's like in Hold 'em, in Texas Hold 'em, for example, like you get your cards, you make a bet, someone else calls, someone else raises, you call or you raise and then like you see some cards. And then by the end of like the hand, your options are like fold and lose all the money that you've bet. or you're pot committed, meaning you put in so much of your money that giving up is going to fuck you at this point. If you don't win the pot, you're done. Yeah. So you have to keep going. So you have to keep going. You have to keep investing in it because even if there's a low percentage chance of being able to win at this point, a card comes up in poker on the river or whatever, the last card, and you're not going to win.

24:37or you only have a 5 % chance of winning or something like that, you still kind of just have to bet because you won't have enough chips left to play if you fold, more or less. It's how I understand it. Yeah, yeah, yeah. And I'm saying, I mean, that is basically my read on where Microsoft and a lot of the big tech companies are with AI. They have put so many billions of dollars into it already that to back out at this point would be disastrous. They need us to accept it as a society. Right. I always say that they're shoving it down our throats. And I mean, that is what's happening. But it's like, even though people are showing in many ways that they don't want to use it, or they only want to use it in certain contexts or for certain things, it's like, they're like, you will keep eating these vegetables until you're healthy.

25:29Even though we can see you turning purple or whatever. I just wanted to say, it's a real term, I promise. It's pot committed. It's a real thing. Man, you wanted to say he's right. That's all. Yeah. We had like an hour long conversation about pot committed yesterday though afterwards. Sam, you knew it though, right? No, I didn't know it. I had to look it up. And then I asked my fiance who plays poker and he didn't know it. But he's not very good. That means he's not good. That means he's not good at poker. He said, oh, I would never do that though. And I was like, okay, yeah, for sure. Only win.

25:57He only plays poker hands that he wins. I think we should do a 4-4 poker club. We should. Poker match. versus rovers? No, maybe not. That's a bad idea. We can bet Bitcoin. Okay. We'll take a quick break. And when we come back, we're going to talk about Wikipedia.

26:26Do you plan your vacation locations based on the local language? With Babbel, language no longer has to be the barrier. This year, speak like a whole new you with Babbel, the language learning app that gets you talking. Learning a new language is the pathway to discovering new cultures. So before you embark on a trip, embark on learning something new. Babbel's quick 10-minute lessons handcrafted by over 200 language experts get you to begin speaking your new language in three weeks or whatever pace you choose. And because conversing is the key to really understanding each other in new languages, Babbel is designed using practical, real-world conversations.

27:06Spending months with private teachers is the old way of learning languages, and nothing screams tourist like holding a phone translation app up to your face all day. Babbel's tips and tools are inspired by the real-life stuff you actually need when communicating. With a focus on conversation, you'll be ready to talk wherever you go. Last year, I used Babbel to learn Italian before my trip to Rome, allowing me to better connect with the people I met there. This year, I'm brushing up on my rusty Spanish before a trip to South America. This year, get talking with Babbel. Babbel is gifting our listeners 60 % off subscriptions at babbel.com slash 404.

27:44Get up to 60 % off at babbel.com slash 404. Spelled B-A-B-B-E-L dot com slash 404. Babbel.com slash 404. Rules and restrictions may apply. I don't know about you, but I like keeping my money where I can see it. Unfortunately, traditional big wireless carriers also seem to like keeping my money too. After years of overpaying for wireless, I finally got fed up from crazy high wireless bills, bogus fees, and free perks that actually cost more in the long run and switched to Mint Mobile. Now I pay just a fraction of the price with Mint compared to big wireless carriers, and you can be doing the same.

28:27Say bye-bye to your overpriced wireless plan's jaw-dropping monthly bills and unexpected overages. Mint Mobile is here to rescue you with premium wireless plans starting at$15 a month. All plans come with high-speed data and unlimited talk and text delivered on the nation's largest 5G network. Use your own phone with any Mint Mobile plan and bring your phone number along with all your existing contacts. Ditch overpriced wireless and get three months of premium wireless service from Mint Mobile for 15 bucks a month. If you like your money, Mint Mobile is for you. Shop plans at mintmobile.com slash 404media.

29:04That's mintmobile.com slash 404media. Upfront payment of$45 for three-month five-gigabyte plan required, equivalent to$15 a month. New customer offer for first three months only, then full price plan options are available. Taxes and fees extra. See Mint Mobile for details. Starting any business becomes so much easier when there's another business you can rely on to make things easy. For our merch, that business is Shopify. And for millions of online and brick and mortar stores, well, they rely on Shopify too. Nobody does selling better than Shopify, home of the number one checkout on the planet.

29:44Shopify's not-so-secret secret, it's ShopPay, which boosts conversions by up to 50%, meaning way less cards going abandoned and way more sales going. So if you're growing your business, your commerce platform better be ready to sell wherever your customers are scrolling or strolling, on the web, in your store, in their feed, and everywhere in between. Businesses that sell more sell on Shopify. Upgrade your business and get the same checkout 404 Media uses. Sign up for your$1 per month trial period at shopify.com slash media, all lowercase. Go to shopify.com slash media to upgrade your selling today.

30:25Shopify.com slash media. Hackers and cyber criminals have always held this kind of special fascination. Obviously, I can't tell you too much about what I do. It's a game. Who's the best hacker? And I was like, well, this is child's play. I'm Dina Temple-Raston, and on the Click Here podcast, you'll meet them and the people trying to stop them. We're not afraid of the attack. We're afraid of the creativity and the intelligence of the human being behind it. Click Here, stories about the people making and breaking our digital world. AI machines. Satellite. Engine ignition. Click Here. And lift up.