← The 404 Media Podcast
Loading summary
Transcript67 lines
- [00:00]
Jason Kebler
A super intelligent AI built a time machine, came back in time, used its powers to make a super vague X account and tweeted the name of this person and nothing else.
- [00:16]
Sam
Hello and welcome to the four four Media podcast where we bring you unparalleled access to hidden worlds, both online and IRL. Four4Media is a journalist founded company and needs your support. To subscribe, go to Four4Media Co as well as bonus content every single week. Subscribers also get access to additional episodes where we respond to the best comments. Gain access to that content at 404 Media co. I'm your host, Sam and with me are 44 Media co founders Jason Kebler.
- [00:41]
Jason Kebler
Hello.
- [00:42]
Sam
And Emmanuel Myberg.
- [00:43]
Emmanuel Myberg
Hello.
- [00:44]
Sam
Joe's gonna come in. Joe's gonna appear as a ghost in this next clip that we're about to play. Um, he has an announcement, he has
- [00:51]
Jason Kebler
a special special message. He couldn't show up, but he has a special message.
- [00:56]
Sam
Joe got. Joe got some really good impact in. So he's going to tell you about that in the next for the next one and a half minutes. I've not heard this yet, so we'll hear it for the first time right now.
- [01:07]
Joe
Hi, I can't be there today on the podcast because I'm traveling for some work related stuff, but I wanted to give you a quick update on a story we covered a couple of weeks ago. At this point, the headline of that one was FBI extracts suspects deleted signal messages saved in iPhone notification database. And the issue was that even when people seemingly set signal messages to delete or in the case of this specific story, even deleted the signal app, copies of incoming signal messages were still stored inside the iPhone notification database. So even though the FBI was not able to forensically extract messages from signal itself, they were sort of captured in this relay point in the notification database. Obviously we thought this was very, very important. We covered it after that. Actually a few people sent me different cases where this has come up as well. So it's clearly a tactic that third parties and authorities are turning to to obtain signal messages in and of themselves, but also ones that are supposed to be deleted. Well then just last week or around about that time, Apple said it has now fixed this follows Signal asking Apple to look into it following our reporting. I'll just read a little bit of signal's statement here. We are very happy that today Apple issued a patch and a security advisory. This comes following 404 Media reporting that the FBI accessed signal message notification content via iOS despite the app being deleted. There's a link to the security advisory on Apple's website, Signal then adds, note that no action is needed for this fix to protect Signal users on iOS, once you install the patch, all inadvertently preserved notifications will be deleted and no forthcoming notifications will be preserved for deleted applications. And that's the key part, right, because there was this very important question of, okay, well, I could turn off notifications for Signal or I could change it so only the sender appears or not the message content or something like that. But we really had no idea for how long these notifications had been stored, how long they were going to be stored for in the future, and all of that sort of thing. And in an email to me, Apple also added much the same. It said, this is actually going to purge all of the notifications that have been collected inadvertently, the ones which have been marked for deletion. Now, Apple describes it as a bug, as in this was not intended behavior. They say it was. It's always their policy to delete notifications that are supposed to be removed, but clearly, obviously that was not happening here. So I just wanted to give you that quick update that this is the sort of reporting that 404 Media subscribers are helping bring about, you know, and now Signal users, and to be clear, other app users as well probably that have, you know, deleted messages, but they're still saved notifications. Have to worry about that now. Anyway, I'll give it back to the rest of 404 Media. Thank you so much.
- [04:30]
Sam
Okay, thank you. Thank you, Joe. Thank you to our foreign correspondent. Okay, so the first story we're going to talk about is one from Jason. The headline is, did a time traveling super intelligent AI try to warn about White House correspondence dinner shooting? An investigation. You always know some something is going to be up when we end a headline with an investigation. So yeah, this is about a very weird meme that's been going around. Meme or conspiracy theory, I don't know which. But Jason, you want to just like start with describing the image in question here. What are we looking at?
- [05:10]
Jason Kebler
Well, first of all, someone like mentioned me yesterday and said this is Betteridge's Law of headlines here. And it's like, no fucking shit. That's the point. That's the point. If you don't know what that is, it's like there's this theory in journalism where anytime there's a question mark in a headline, the answer is no. And it's like, yeah, the answer is no. A time traveling super intelligent AI did not warn about White House Correspondent center
- [05:38]
Sam
shooting anytime there's an investigation. We're up to some silly shit.
- [05:42]
Jason Kebler
We're up to no good, for sure. So basically, in the aftermath of the White House correspondence dinner shooting, which I hesitate to even call it a shooting. Was it a shooting? I guess there was. It was like an attempt, I believe shots were fired.
- [06:05]
Emmanuel Myberg
Right.
- [06:06]
Sam
But no one was shot, were they?
- [06:08]
Jason Kebler
Yeah, allegedly. Maybe a Secret Service person was shot, but that. It was like, wildly difficult to get any information about this, despite there being hundreds of journalists there, which we don't need to talk about, but it was hard to find information about what was going on. Anyways, in the aftermath of this, someone, I don't know who, but X, the community of X, found this account called Henry Martinez that had exactly one tweet from December 22, 2023. So like two and a half years ago. And the tweet just said Cole Allen. And Cole Allen is the name of the suspected shooter. And the Henry Martinez account has a Pepe as its avatar. Pepe holding a wine glass. And then it has this header image that is like. I would describe it as like generic 3D art, psychedelic rainbow art of like falling, like a. Like a stalactite. Stalactite situation, like interior of cave, but it's rainbow. Make it Technicolor. And this is just like a genre of AI computer. Not AI computer art, but just of 3D rendering art that is very, very common, and it's very common on stock image websites. And so this was like the Henry Martinez cover photo, and it started this conspiracy theory. Should I explain what the conspiracy theory is?
- [08:03]
Sam
Yeah, please get into. I guess, like. Yeah, so this is all very weird. The. The Cole Allen thing is weird because it was posted years ago. So what's the theory behind. I guess, first of all, what are we supposed to be seeing? And then what is the theory behind what we're seeing? Because it's like a magic eye thing, people think. Right. Like, you're supposed to be able to.
- [08:25]
Jason Kebler
Yeah. So I mean, one version of the conspiracy theory is that it's a magic eye. So you can, like, just stare at this, you know, technicolor image for a while and you'll start to see something. And the thing that some people are saying is that it is like a digital representation, like AI representation of the. The iconic photo of Trump after he got shot in Butler, Pennsylvania, with his fist in the air. And that is like being used alongside some other information to suggest in a conspiracy theory that's been viewed by millions and millions and millions and millions of People to be clear that a time traveling artificial intelligence created this Twitter account in 2023 to warn of the White House correspondents dinner shooting and possibly the Butler shooting. Like, unclear whether that is part of it or not. But that is like sort of the unhinged conspiracy theory that has occurred. And we can talk about why. But that's, that's kind of like the long and short of it. There's also like a bunch of images where someone does like a digital slider and you can see the like technicolor image turn into the Butler image. Like back and forth, back and forth. I gotta say that the way that they manipulated that one is pretty. It's like I think you can do that with any image because that's like how these sliders work. Like they're, they're designed to make one image look like another image. But I don't know what, what do you guys think? Like, have you, have you tried the Magic Eye? Are you able to do Magic Eyes in general? I feel like I, I don't have that skill.
- [10:23]
Sam
I can, I used to. There was like a. This reminded me of like this little game program that you would project onto a TV when I was a kid and it was all Magic Eye stuff. So I would be constantly going up to my TV as a child, getting this close to it, crossing my eyes and then backing up. I was like, that's probably why my eyesight is really, really bad today. But I tried it with this and I couldn't. I was like, this is nothing. I don't see anything. But maybe I don't have the specific unlocked alien awareness necessary.
- [10:54]
Emmanuel Myberg
I can do Magic Eye very easily, but I'm one of those people who feeds the image recessed instead of pop out. Do you know that's a thing because it's supposed to pop out, but I can only see it recessed so everything looks like a mold or something like that. The reason Sam can't Magic Eye this is because it's not a magic eye. Right. The way that you do Magic Eye is you take an image and you put it into a pattern that is then offset and obscured by an abstract image. And this isn't that.
- [11:29]
Jason Kebler
I will say it's not because she doesn't have the correct mega gene.
- [11:34]
Emmanuel Myberg
Yeah, but I mean there is something about the texture of the image that kind of matches the composition of the photograph. Like, I think that's fair. It's like not one to one, but if you look at it side by side, it does kind of maybe look like Somebody with his hand up
- [11:59]
Sam
cloud or like a constellation.
- [12:01]
Emmanuel Myberg
Exactly.
- [12:02]
Sam
I guess we could imagine. It is a weird image. It kind of looks like it moves to me. It's a weird one. If I was on enough mushrooms, it would probably hurt my brain.
- [12:13]
Jason Kebler
I can't do Magic Eye because I got Lasik. And once you do that, you lose your. You lose the part of your retina that is able to see. Magic eyes.
- [12:21]
Emmanuel Myberg
Nuh, you're making that up. Is that real?
- [12:24]
Jason Kebler
I mean, I don't know, but it sounds like it could be our ability
- [12:28]
Sam
to see this away.
- [12:29]
Jason Kebler
Yeah, exactly. It was removed. It was removed. So, okay, back to talking about this. And I swear that there's a reason and there's a point to doing this story and all of this, other than it's ridiculous. And we're just sort of like trying to explain a little bit how the Internet works these days. But basically the time travel aspect of this story came from, I guess one, the idea that, well, this guy's name was tweeted two and a half years ago. But the other thing is that as the conspiracy theory goes, there was an institute in Europe that used this image on a blog post. And that institute is called Time Machine. And so therefore people are like, well, how much clearer could you get? Like this image comes from this organization called Time Machine. And all this is happening and they do AI research to some degree, which we'll talk about. But basically it was like, okay, well a super intelligent AI therefore built a time machine, came back in time, used its powers to make a super vague X account and tweeted the name of this person and nothing else. But that's not really what happened, of course. What happened is that again, this image in question is a stock image. It's not an image at all. It's a stock graphic that was made by someone called Distinct Mind and it was uploaded to Unsplash in 2021. Unsplash is like a royalty free stock image library that we actually use all the time. And it is notable in that it's fully free. You do need to make an account, I believe. I'm not sure if you even need to make an account to download anything, but it's like you have access to more, you do make an account. And so this was uploaded in 2021 and like all stock images, it's been used for all sorts of things all over the Internet. So I did a reverse image search on Tineye and then I also used Google's reverse image search which shows you like where, where else the Image has been used and when, and it's largely been used on blog posts about psychedelics and psychology. So there was like a medium post by a doctor who went to a ketamine psychotherapy retreat and wrote about this image or wrote about that experience and used this image. Someone is selling it on Etsy as a poster, is used as an ADH treatment website's image. Someone wrote about the Bible and used it. And then a finance firm used it about pricing integrity, like a blog post about pricing integrity. And so basically it's been used like over and over and over again. And one of the places it was used again, this is not the institute that created the image. It's just they downloaded it and they used it for a blog post. Is this research organization called Time Machine and they are funded by the European Union. And I will say that their website is kind of batshit, but kind of in the way that like a lot of nonprofits websites are batshit, especially in this space. And so what Time Machine is, is it is a European research organization that is working to digitize historic documents and artifacts from the European Union. And so using technology specifically. So they're using a lot of AI to parse handwritten notes and categorize them. They are doing a lot of 3D scanning, which like Emmanuel's writing a fair bit about 3D scanning libraries and things like this, but basically like using technology to scan the inside and outside of buildings so that you can then like upload the 3D models onto the Internet so that they are like saved forever. And the maximalist version of this is like you can do VR tours of buildings that have been like saved in this way. And then the other thing is like, I mean, this happened in a quite interesting way where for Assassin's Creed there was like a very high res scan of Notre Dame in France. And then after Notre Dame lit on fire, they like consulted the 3D renderings from Assassin's Creed to help rebuild. Is that true? I feel like that's true.
- [17:32]
Emmanuel Myberg
The scan is definitely real. I forget and I would have to look it up. But I do think that they referred to the actual Ubisoft scan of it. I'm not sure for what purpose.
- [17:43]
Jason Kebler
Right. Well, I mean, there's basically all sorts of organizations and museums that do stuff like this. And this is the hot thing in archiving now is like using technology to categorize all of these like old letters from wars and save like 3D models of like old clothes before moths eat them and things like this. And so that's what time machine is. They're not building a time machine literally. They are, like, building these interactive experiences about castles and things in European history that you can then, like, go through and look at. And they do talk about using artificial intelligence in some of their research, because everyone in every field is talking about how they can use artificial intelligence now. And we've done a few articles about this, like, in history right now. It's quite controversial to. It's like anything else. It's like, some are AI maximalists, and then others are like, no, don't do this. And so I did an article at the US Archives wanting to use AI to scan, like, Civil War letters and then transcribe them and categorize them and all of that. And then people pushing back because the AI doesn't always transcribe it right, or it gets things wrong. Or there's also, like, a bunch of people who are trying to make, like, bring dead people back to life by analyzing their letters and then making a chatbot of them or whatever. And so that's, like, the type of thing that time machine has researched over the years. And this has been turned by conspiracy theorists into, oh, they're building, like, a literal time machine, and they use this image, and it was on X. And therefore an AI is warning us about this shooting. It's very stupid.
- [19:35]
Sam
I mean, I think it's interesting and you note this in the story, but part of why some of these things. Things go right, because this has happened before where, like, an account will tweet a name years ago, and then that name will become newsworthy, and then everyone's like, oh, my God, they predicted this. I didn't actually realize that. This is like a game for some accounts. They actually just, like, are blasting random names all the time, and then one tweet will hit from, you know, years later.
- [20:08]
Jason Kebler
Yeah, I mean, so that. That's, like, to the extent that anyone cares about this at all, that's like, the most interesting question here is, like, what is this account? Why did it tweet the name Cole Allen in December 2023? And, like, the mechanism by which it happened? And so, I mean, we don't know. We don't know exactly, like, how this happened, but in the past, there have been X accounts that tweet just, like, yeah, random names or, like, specific outcomes, and they will be a private account. And so people can't see what they're tweeting. And then something happens in the real world, and they delete all of the tweets except for the One that makes it seem like they have psychic powers and they have predicted something. And so, like a very high profile example of this was during the 2022 World Cup. There was an X account that seemed to have predicted the winners and correct scores of a bunch of games. And that's exactly what he did, is he had like a private account. He tweeted like hundreds, thousands of different potential outcomes of games and then deleted the wrong ones. And then by the time people had discovered it, they were like, whoa, this person is a freak. The. And I guess like an interesting. Just another side note, like, one of the reasons why it's like this is because Elon Musk made it a lot harder to archive tweets. He killed, like, quite a lot of the archiving tools because he made the API access to Twitter really, really expensive. And so a lot of the, like, fire hose of. Of like raw data of tweets is like, people don't have it anymore because to have that costs like hundreds of thousands of dollars a month. And so all these free archiving tools and things are gone. And so it's pretty impossible to know what happened here because all the tools that we would use to figure out what happened are no longer in existence.
- [22:17]
Sam
Awesome. Love that. Yeah, thank you, Elon. Once again, very interesting. Okay, well, I guess we didn't. We solved the conspiracy here despite not having the. The gene that can see it. So good for us.
- [22:34]
Jason Kebler
Yeah. There's a lot of other conspiracies about this and like, a lot of people think it was just like a fake shooting and things like that. And it's like, we don't need to get into it. But I think, like, very broadly speaking, it's like, it's. It kind of talks, it's. It says something about the level of trust in this administration and then also just like, what's going on on the Internet. And I think also people's like, general exhaustion with shootings and with this administration and all of that, where it's like, yes, this happened on a weekend. I feel like no one cares.
- [23:12]
Emmanuel Myberg
All right, I was going to ask how you rate the. The level of interest in this. I would say, and this is not to trivialize what have, what could have been like, really horrible violence, but it seems like low interest in, in an attempted assassination at a public event filled with media.
- [23:38]
Jason Kebler
Pretty strange.
- [23:39]
Emmanuel Myberg
Yeah.
- [23:40]
Jason Kebler
And it's not just low interest, like on the Internet, like, I see people talking about how they don't care, and I see, as predicted, like, tons and tons of takes from people who were in the room being like, here's what happened and here's how I reacted and all of that. And it's like, I don't know. We knew that that was coming. Like, that's just sort of how these things work. It's like, if something happens at an event that a lot of journalists are at, the journalists are going to write about it. But I have talked to people in my life being like, oh, did you see the shooting or hear about it? And the instant reaction from people who I believe are, like, smart, are like, oh, that was designed to make Trump's approval rating go up. Slash, like, I don't care. I'm moving on with my life.
- [24:36]
Sam
The popularity of it being. The idea of it being fake is really interesting to me because everyone. I mean, there I was at the gas station near my house, and this guy was talking very loudly about how it was fake, and who cares? He was. He was reacting to a New York Post that was on the newsstand, but he was just like, this is also fake. Like, just people. It's like, I've. I. We were getting coffee, me and my partner and I had read about it that morning, and I was like, oh, did you hear about the shooting at the White House Correspondent Center? And he was like. And then we just got coffee, and it was like, it just doesn't register anymore with. With anyone. Which I think is a really sad indictment of the state of things.
- [25:18]
Jason Kebler
But I was watching the NBA playoffs with a friend at his house, and abc, I guess it was abc, switched from the playoff game to.
- [25:31]
Sam
They cut in.
- [25:32]
Jason Kebler
They cut in. But they also put a little thing, like, if you want to keep watching the game, switch to ESPN2. And we watched the news conference for 30 seconds, and we're like, oh, he's talking about a fucking ballroom. Put the game back on. And I've seen a lot of people who are like, yeah, we just, like, kept watching the basketball game because, like, this was crazy. Like, this is ridiculous.
- [25:58]
Sam
I mean, that's part of it, too, right? It's like, Trump, another attempt on his life, not the first. And he's just like, yeah, I think we should remodel. It's like, okay, so I guess we can go back to not giving a shit about this. Like, what do you think?
- [26:15]
Jason Kebler
I mean, that's the main thing that people. That the administration has been talking about coming out of this. And also, like, we don't need to get into it, but the White House Correspondents Dinner is not gonna be at the White House ballroom. Like, that's not what it is supposed to be. Who knows what it's supposed to be at this point, but it's like theoretically an independent ish thing that is not like the president's event. And so I don't know.
- [26:40]
Sam
Anyways, and he's like, I guess if this place was less of a dump, people wouldn't try to shoot it up. It's like, what are you talking about? Anyway, okay, well, shall we move on to our next story after these messages?
- [27:04]
Jason Kebler
Buying glasses used to feel way harder than it should. Like, you walk into a store, everything's overpriced, half the styles feel outdated, and somehow you still can't tell what's actually going to look good on your face. It always felt like way too much effort for something you wear every single day. That's why I've been really into Warby Parker. Honestly, it's the only place I've been buying glasses from for like the last 15 years. They basically simplified the entire process. Their virtual try on is actually legit. You just open your phone, point the camera at your face and you can see how different frames look in real time. And it's surprisingly accurate. You can actually tell what fits and what doesn't. And once you find something you like, the quality is there too. These don't feel flimsy or cheap. They feel like premium frames just without the insane price tag. These are mine. They're Elio in black walnut. I've been wearing them a lot lately. I wear them basically anytime I go outside in California. And I've gotten tons of compliments on them. They're comfortable, they fit well, and they just look clean. What really surprised me though, is the price. Prescription glasses start at $95, which is kind of wild considering the quality. You don't have to choose between something affordable and something that actually looks good anymore. And it's not just glasses. They've got contacts online, eye exams, sunglasses, and even over 300 retail stores if you want to check things out in person again. I've been wearing these Elios basically anytime I go outside, I love them, I wear them driving, I wear them walking my dog, I wear them at the beach. Got tons of compliments on them. So right now, buy one prescription pair and get 20% off any additional prescription pair at warbyparker.com 404media that's 20% off additional prescription pairs when you go to warby parker.com 404media. University professors disturbed to find their lectures chopped up and turned into AI slop. Samantha this is your article. It's quite hot. Like you, you really hustled on it. You got it out quickly. There's a lot going on here, what happened at Arizona State University.
- [29:24]
Sam
So I think we're all at this point familiar with the idea. And you wrote about this very recently. The idea and the phenomenon that's happening where so many of people's workplaces are getting AI shoved on them. It's like, you must use AI. We're going to be implementing AI whether you like it or not. And usually that comes with a bunch of messaging. Leadership or an administration will be like, we want you to use this tool. We're excited about it. It's in beta. Here you go. Try it out. Let us know what you think. That's usually kind of the way it goes. Not always, but I think typically in the cases that we've seen where an employer is like AI time, this is very different. So I'm pretty sure a lot of people found out about AI Atomic, which is the name of ASU Arizona State University's new platform for AI learning, through a post from Chris Hanlon, who is a professor of literature, US literature at ASU, and he posted on Friday, April 21, I think it was last week, where he says it's a subscription platform that claims to offer customized learning modules for fee paying users. None of the ASU faculty whose course materials were harvested for the module which I generated were aware that their image, lectures, lessons or other teaching materials are being used. So at this point, Chris is kind of raising the alarm about, hey, there's this new platform. No one knows about it or what it is, and it's using all of our meaning like faculty materials. So Atomic is. And we can kind of just go through what the actual platform is. But the way that I, I signed up immediately for an account. I didn't, I wasn't really sure if it was public beta, but I signed up with my personal email and it worked. And it is a chatbot. At first it's a prompt and you put in there like I put in, I want to learn about AI ethics. Because I was like, I'm going to find some good sources this way. So I put in, I'm going to learn about AI ethics. And then it takes you through basically a quiz. So the chatbot's like, how long do you have per week to devote to learning about this topic? And you can say 30 minutes, an hour a week, Just ludicrous, stupid. But you can say like several hours a week, unlimited time. How fast do you Want to learn about this? And it'll say as fast as possible. Which is what I chose because I was optimizing for the most ridiculous content. And it's like, what part of AI ethics do you want to learn about? It asks you a bunch of questions about the topic and then eventually it says, okay, we're generating you a module and it'll take a few minutes, minutes. And what it spits out is. What it spit out for me was a series of sections. So I think it was seven or eight sections. And each of the sections includes a subhead and it was like AI's impact on ethics, Bias and Responsibility. Ethics and Responsibility in AI. But then it says several of the subheads in the sections are repeated. So it's ethics and AI is section one, section two and three. So anyway, it's auto generating these sections. And then in the sections is just AI generated text, like walls of text breaking down concepts based on lectures pulled from ASU professors content. And we can get into like how that stuff actually got there. But that's basically what it is. And then between the walls of text in the sections are really short, like 40 second to 2 minute I think is the longest one. I saw clips from professors lectures, but they're totally out of context. They're video seminars and then no names. It's like, who is in these videos? No direction back to like what the course is. No references, no credit whatsoever. So that's ASU's new tool.
- [33:36]
Jason Kebler
So I guess a few things here. One, nominally the purpose of this is like, hey, you're an ASU student, you're in this class and you need to study for the class. And we are going to be like your AI guide to doing this. Like as in we're gonna like take the, take the lectures that you're supposed to watch off of Canvas, which is the, you know, online like learning portal thing that is used across high schools and colleges all over the country. We're gonna take like the video lectures that your professor has recorded and we're going to like make them shorter for you. We're gonna give you like the, the TLDR of them more or less. Right?
- [34:22]
Sam
I think, I mean, I think that is one way it could be used. As far as I can tell from the way that it's marketed, it's like you can't get course credit from doing this. It says it in the FAQs that it's not for credit. I think it says, yet it kind of implies that maybe someday kind of leaves that open. But I think it's mostly for a public facing non ASU student to use to take advantage of ASU's the stuff that it already owns.
- [34:55]
Jason Kebler
So it has a static expertise and all that.
- [34:57]
Sam
Yeah. And it's. And it charges a subscription fee, I think. So they took down the signup page after I reached out basically and I think other journalists were also reaching out about this. But within a couple hours the signup page was gone. I'm pretty. I had a 12 day trial. I put my credit card information in there, which was probably not a good idea, but I'm on a trial and I think after the trial's up, it's $5 a month. It's something like that. It was really. I remember thinking it was really cheap and this is not, you know, like commissary to the quality that it should be giving you. But then I started doing it. I was like, oh yeah, this is not worth more than 50 cents a month. So I think it's for people outside of ASU to experience asu, like to kind of to audit a class. Almost like I'm gonna. And it's very customized. So it's like I. If I want to learn about like. Cause some of the topics were like, I want to learn about how to manage my finances. It was like very normal stuff that you might just want to learn about from the Internet.
- [36:00]
Jason Kebler
Got it. Okay. I guess that that's semi interesting because like for years I feel like Ivy League at first and now probably like more universities have been doing MOOCs which are massively online. What actually, what do they stand for? Massively online Online College Online.
- [36:22]
Sam
Massive online. Massive open online courses. It was huge. When I was like in college, it was. MOOCs were a big thing that's like, yeah.
- [36:29]
Jason Kebler
And it was basically like, oh, watch the lectures From Computer Science 101 at Princeton or Stanford. Did a bunch of them. So this sounds like something similar except for like masterclass slash, we are going to make it shorter and you'll be able to learn all this stuff very quickly. Which I asked that question because then it's great that it will create these modules for you and then just throw in a bunch of random shit. Like basically you talk to professor who had some of her course material pulled into this like AI ethics module that you had created and it had pulled her from like a class from years ago that was like totally unrelated.
- [37:17]
Sam
Yeah, it was a class from 2020. She's a film studies professor and someone, she told me someone had asked in the class a student had asked in the class, hey, let's can we talk about AI and just kind of like, can we actually add this to the course and talk about it a little bit? Because this was in 2020 and it wasn't really a big deal yet, like the way it is generative AI was not what it is today. It was a big deal, but it was like, ooh, this thing is coming. So she kind of just like threw in a couple slides about like defining AI and machine learning. And she told me for this story that she would not define it that way today. It's not really an up to date definition. And also it wasn't part of a class about AI at all. It's not her expertise. But they threw in this one minute clip from her class, I assume because they probably were scraping canvas and looking for just things that mentioned the words AI and ethics or like definition AI. They were kind of slotting it in that way and hers just cropped up to the top. As far as a succinct version of this, it kind of fit into that minute that it was probably trying to hit. But this is all assumption. I don't really know exactly how it works other than it's definitely scraping from canvas, like you said, which is their learning management system. And then ASU owns canvas. One of the professors told me that canvas is very clear about if you're putting stuff on canvas, we have usage rights. If you working at asu, they're very clear about intellectual property rights. It's like ASU doesn't solely own their rights to that content, but it does own the ability to do what they want with the stuff that you're teaching on canvas.
- [38:58]
Jason Kebler
I mean, that's true. And I guess the professors that you spoke to didn't raise that as like the main thing that they were upset about. But a lot of newspapers and media companies have been training these AIs on their reporters writing. And then like there was a story about McClatchy, for example, which is this big newspaper conglomerate. And then McClatchy was like basically generating stories or proposed to generate stories using AI and then putting writers bylines on them, even if they didn't like agree to that. And so, I mean, I think that what's happening now at asu, from what I can tell from your story, is that it is, as you described, chopping their lectures up and decontextualizing them. And that's super annoying, but it's not taking their lectures and generating new AI using a deep fake of the professors and their stolen Voice and things like that, which hopefully it doesn't get to that point, but it would be easy to imagine them doing that. Like, that is totally within the realm of possibility for, you know, either ASU or for like another tool like this. But yeah, as it stands, it seems like it's a pretty. Like. Seems pretty useless at this point. From what I can tell.
- [40:24]
Sam
It's really bad. It's like it's. It's worse than useless. I mean, there was one. The first. The first section of the first module that I created, it was pulling from a lecture and the professor had a pretty heavy accent and he was saying ex riskers to describe a certain type of people who feel a certain way about the risk of AI. So again, I don't even know what the context of the lecture was because it was so short, but X riskers was what he was saying. And he even wrote it on the board and spelled it that way. And then Atomic went in and transcribed that as Xriscus. So it was R I S, C U S. And then it kept referring over and over to Xriscus, the Xriscus community, again and again throughout the module. And then it did a quiz about the Xriskus community. And it's like, if I'm learning about AI ethics for the first time going into this, I won't. I have no idea what this topic involves. And I'm reading this, I'm like, oh, there's a community of people called the X Riskus community. And I go tell my friend, like, have you heard of the X Riskist community? It's like, what this is. It's just making shit up based on really bad transcription. It also did that with another. It did it with Chris Hanlon, the professor who tweeted or bluesky, posted about it, where he was teaching about Cleanth Brooks, who's a literary critic. And it transcribed it to client Brooks and then referred to client throughout the module. Not only did ASU not alert anyone to this, didn't send that email that I referred to earlier that was like, hey, we're doing AI now. Get on board. Didn't even tell them, didn't ask for the permission to use lectures because they own it. Technically they legally don't have to, but they didn't take the time to look through the modules and what was being created because they're highly customized, so everyone's getting a different one. If I went and generated a different module about ethics and AI, it would give me something totally different. So there was no fact checking and no references out to anything else. So you're completely in this closed environment learning about things that may or may not even be factually accurate or spelled correctly. And it's not telling you go here to learn more, or here's a professor's full lecture or here's their name even there's no names on anything. So I think it is like a lot of the pressers I talked to weren't, they weren't saying I want to be compensated for my work because this is an ownership issue in that way. They were saying this is an ownership issue because I took the time and effort to create the, and build that curriculum and that module. And I care a lot about my students understanding these topics, especially when we're talking about ethics and things like that. I want them to grasp these topics in a way that I've crafted them because I took the time to learn about it. And it's insulting and dehumanizing to just turn that into this dog food slop that ASU has done with Atomic. And that was what really bothered most professors. And I think that's, I mean it was just like, it was honestly pretty. It was very sweet hearing them talk about their students and how much they cared about what their students were learning and how frustrated they were that this was done just completely in the dark without their knowledge and also as butchering because for a professor and for someone in academia a lot you're not getting paid. Amazing. So you're reputation is what matters and your ability to speak on something as an expert is what matters. And this AI is just ripping clips out and turning them into fodder for bad information, which I think really is what frustrated most of them. But yeah, I think ASU knows that it's in hot water at this point because they took down the signups.
- [44:25]
Jason Kebler
But yeah, that's what I was going to say. I guess like let's end it with like whatever the status of this thing is. I mean it's like again a beta test. It seems like you can't sign up anymore.
- [44:36]
Sam
Wait, I can now. Today. I guess I'll play with this more today. The signups were a four or four page yesterday and today we're back. So I'll make another account and see if I can keep messing with it. But they haven't replied to any of my emails. I emailed the people that are involved, President's office, communications office. No one has replied. I think they're probably doing some serious, some serious crisis management at this point.
- [45:03]
Jason Kebler
If you are, if something similar is happening at your college hit up Sam.
- [45:09]
Sam
Yeah, definitely. I think this is happening elsewhere pretty rampantly. So. Yeah, and you might not even know it. So go digging around, I guess.
- [45:22]
Jason Kebler
Should we leave that there?
- [45:24]
Sam
Yeah. When we come back, we're talking to Emmanuel about some philosophy. Speaking of ethics consciousness.
- [45:33]
Jason Kebler
That's in our subscribers Only section which you can gain access to by going to 404 Media Co and clicking subscribe.
- [45:44]
Sam
As a reminder, 44 Media is a journalist founded and supported by subscribers company. If you wish to subscribe to 44 Media and directly support our work, please go to 44 Media co. You'll get unlimited access to our articles and an ad free version of this podcast. You'll also get to listen to the Subscribers only section when we talk about a bonus story each week. This podcast is produced by Alyssa Midcalf. Another way to support us is by leaving a five star rating or review for the podcast. That stuff really helps us out. This has been four four Media. We'll see you again again next weekend.