Linux After Dark – Episode 114

This Late Night Linux Family podcast is made possible by our patrons. Go to late nightlinux.com slash support for details of how you can join them. Support us on Patreon for access to ad free episodes and early releases. That's late nightlinux.com slash support. Hello and welcome to episode one four of Linux After Dark. I'm John. Chris. I'm Gary. Welcome back all.

How do you convince people to stop using unethical technology? The answer of course is you can't, so I suppose that's the end of the episode. We'll be back in a couple of weeks. Yeah. So a bit of context here. Someone dear to me sent me one of their creations the other day and it was a sticker that they created, like a bumper sticker. And I won't go into details, it was just a dad joke basically.

But it was just obviously AI slop. And I said to them, Look, please don't send me AI slop. Like I find it pretty unethical and I just don't want to see it. And they said, Yeah, okay. And I thought that would be the end of that. Then I discovered this cool Python project where you tell it the location that you want and like some other parameters and it makes this cool kind of poster of a map of that area.

And I did this person's area, sent it to them, and they said, Oh, isn't that AI? I'm like, No, like not every cool thing made by a computer is AI. And I I just have this question for you. Like, is there any way to try and even approach this subject with people? Well I think the big problem is that it's become so ingrained with What people are doing in society though.

Either they A don't realise that it's AI, B don't realise the implications of it, or C just don't care and think that everything that it outputs is absolute truth and is fine. And I've had several examples of all three of those in the last couple of weeks. And you're right, it's infuriating. And I think it's most infuriating when people present what the AI is telling them as fact, despite the fact that it has no concept of

The actual reality, the question you're asking it, the nuance around certain situations, or anything else? Professionally, I find it baffling that People raise support tickets and say, I asked Chat GPT and it said this and it makes me physically wince and I can't imagine myself ever doing that.

But I do think you're right, Gary. I think and you Joe, I think people just think, Oh, this is the next thing that computers do. So it's just fine. There's obviously the kind of other ethical concerns which are quite right in terms of resource usage, but the atrophication of the human brain, like I'm employed to be better at this than you. That's the paradigm here. You've raised a support ticket to ask me for help.

Why are you presenting this complete bollocks to me with complete lack of apology as if it's helpful to the situation? Yeah, I mean I've had similar, Chris, where I've literally had people link me to the Gemini AI output of something they've searched for in a user manual for a piece of audio equipment. It's like I know what the manual says, but I also know the way I've configured the piece of equipment and that isn't the way that it works. And you're right. Searching with an AI model by default.

before engaging your critical thinking skills. And I don't know how you get around that. Well, just this week I had an interaction with an ebay seller who was new to it and um

Long story short, they needed to give me a partial refund and they didn't know how to do it. And so during the conversation where we were talking about how to do it, I went away, did a Google search. found the relevant page that they needed to read. um was about to send it to them when they said, Oh, I asked ChatGPT and I'll how to do it now. That is really symptomatic of the problem that searching ChatGPT for things has become

the new Google it. And I find that immensely frustrating because people don't realise the the resource that it uses. But then doing a Google search now, if you don't do dash no AI, pretty much uses the same resources, doesn't it? Well yeah, it's still generating all of that slop at the top, say I guess that's the point, that it it is just so mainstream that I don't know, is there a reasonable

ask to even get people to avoid it anymore. I think it's even more than it being there for consumers. I basically can't work without interacting with an LOM at this point. The search tool that my company uses internally has tons of AI features and automatically gives you an AI summary of whatever you search for. Just a week ago, it switched its default from the search interface to the chat interface, so you would do the exact same thing that you used to do and type in a search query.

And it would just send that as a message to the chat and s bring you to that interface. Uh huh. Why would I have come here if I wanted to do that? I came here because I wanted to search. I wonder if our company uses the same tool because I was trying to look up the expenses policy on our internal documentation portal today. And it just kept giving me AI summaries of the expenses policy with no link to the sources. And what I wanted to read was the actual expenses policy and what I had signed.

Convincing people to not use something because of their ethical concerns is an interesting curve, isn't it? Like I've spoken before about how I like Spotify. I've used this N C Spot client. And more recently I have tried to stop using it, but we have a family plan. my sister pays for it actually and then I pay for other things and we kind of have a seesaw balancing effect.

But I said to her, I'd be completely happy if you cancelled it tomorrow. I'm not really using it anymore. I'll find a different way, listen to music on YouTube. I've got my own music collection running Navi Drome. So I can do that. But they don't want to give it up because it's convenient. I mean, even if you explain

You know, the reasoning and I I'm sure Joe is listening to this thinking, you know, why didn't you care about the musicians from the beginning?'Cause it's clearly like shit. It's a shit deal for musicians and always has been. But that wasn't enough to put me off. Once it's embedded, you know, I I I mentioned it and then my mum said, Oh, you know, what what are the implications? I'm like, well, you know, you can't

Just load up Spotify and have any music you like. You'll have to maybe use YouTube, get ads. It's like, oh no, no, I don't want don don't get rid of it. Don't get rid of it. you know, you explain the reasons that what they're using the profits for now. openly and unapologetically, it's a hard sell. People don't want to give up the convenience. It's the same reason they don't want to stop using Amazon, they don't want to stop using Uber, all of that stuff, especially because

you saying with the corporate entities, once companies have spent a lot of money on this stuff, they have sunk cost fallacy. I have a friend that works in a job where They asked for feedback on a pilot, he gave it honestly, and they blocked him from providing more feedback because they basically just said, uh no, you're not being constructive. It's like, I'm literally giving you constructive criticism about the failures of this.

And then they blocked him from providing any more feedback and rolled out the product. We talk about other things where we want people to do the more ethical thing, like

Eat less meat. And that is a possible choice that you can make. It makes your life more difficult. With don't use AI, it's like Cool. How? Today I sent my mother in law something with inshitification in it, and she laughed and said, Oh, that's a funny word. So I send her like when it was added as word of the year in uh twenty twenty three, twenty twenty four or something, and then the origins of it and Cori Doctoro completely unaware of that, and people are so much further behind.

And I think that doesn't help, to be perfectly honest. The everyday person is not gonna respond well to incredibly aggressive attacks. They're just gonna be like Meh, you know, did a fairy die when I asked ChatGPT, leave me alone, stop getting in my face, go away. And I think that's the really difficult thing, isn't it? Because in order to avoid some of this stuff, I would basically just have to be a hermit in a room.

I couldn't do my job because like May, that involves interacting with an L L M on some level occasionally. I couldn't speak to my family because they use it. I can't even go and do my banking anymore because something pops up in the app. I can use my phone because they have moved the buttons around in such a way that every time my muscle memory tries to click in the search bar, Gemini pops up.

I mean it is truly just pervasive in every area of my life that's meaningful, including just interactions with other human beings. Yeah. Well and LLM came up with this topic for us today, for example. Did it? No, obviously not. It came from my brain box.

Well, when I talked about this on Late Night Linux recently, about the ethical use of AI and stuff, someone, I can't remember who, sorry, got in touch to say that trying to convince people not to use things because of the ethical arguments never worked. And you need to be pragmatic about it. So is that the answer? We need to tell people that using AI search, for example, just isn't as good as using something like

Google or Kagi or whatever. I think therein exactly lies the problem. The LLM produces output that, for most intents and purposes, is good enough. Like, how do you convince people that good enough isn't good enough? It might be wrong, it might be lying to you, it might have killed a tree in order to tell you that.

But it made something that looks reasonably like a person would say it. And like if you ask an AI, tell me what I did last year, so that I can put in my yearly review and it returns a bunch of things that aren't true. You can just slap that in a review. What's your boss gonna do? Say no, you didn't do those things. Maybe, but probably not because he's probably just

using another LLM to ingest all of that and say who did the most work on the team. I did actually try to get that information out of an LLM at uh my work that has access to things like my activity graph.

And yeah, it just completely lied to me and I didn't use it. But I could see people really easily just going, Yep, that's good enough. I gotta get back to work. Yeah, I can absolutely see people doing that. And I guess the flip side of that is that If you're using the LLN to ingest documentation and I guess almost using it as a search,

It is relatively good at that kind of thing. And if you check the sources it returns, then maybe it's not so bad. But equally, sometimes it does just return complete nonsense, like you pointed out. I think if you use it with critical thinking, but then for me, I just feel like, isn't this just kind of like a search engine used to be before they got banned?

But to Joe's point, I do think that people just they don't care. Much like people that smoke, like if you say you should stop'cause you'll get sick, well, I'm all right now. It's the same thing. Don't do that. That's really bad. They go, Well, nothing happened and it gave me something that looks like a good answer. Leave me alone. What do you do when people start taking photos of you or recordings of you in my case and feeding that to AI?

when you clearly don't want them to. The curve that I think about is when I was much, much younger I can distinctly remember a peer of mine at the time, and I would have been in my sort of early twenties, I think. I think I was still using my ISP's email, like I was way behind. And they said, Oh, Google have just launched this free email. You get free space, loads and loads of space. And I just signed up without thinking, you know, the old adage, you are the product.

And I think a lot of people we forget we are not like most people in the way they think about things. They just think about the end product and it doesn't come round. So Most people wouldn't think if they just download an app from the app store that puts funny faces on their kids. they don't have the kind of mental chess to think how that's happened actually.

They just think, oh look, this looks funny. I'm gonna send it on. They don't think about all the waypoints of that where it gets shared with whoever. You just agree to the terms of service when the app comes up and just do it. You don't think about what's going on in the background. And even if you tell people, they just look at you blankly.

Like, yeah, but that picture has gone elsewhere and is now kept forever. You if you read the small print, it's not just happening here. It's not like, you know, a street artist drawing funny things on it. It's a completely different thing. But you'll get a blank response back from most people. And what can you do once has already happened, right? Yeah, you've

uploaded that picture, you've shared it with someone in good faith, and then the damage is already done. There's no turning back. And that's the really, really scary part of this for me, I think. But how do you ask them not to do it again? Like I asked this person not to send me sloping. Now they're just like offended by that and secretly seething about it because their creation, their great creation that they shared with me, I just dismissed as slots. Sometimes people do need to hear the truth.

Maybe there's a better way to word it, but asking an AI to do a thing for you is not a creative endeavor. Don't at me. I know many people who would disagree. Cool. But I will agree. Yes, good for them. Let them have fun. But yeah, I firmly agree it's not creative. Digital art still has critical thinking and assembly. The means of production and the tooling has changed, but I still like to think, you know, an excellently produced Yeah. See repetitive nonsense, my great album.

But if it's pure passivity And I put that back, you know, in a in a work context, the way I try and tackle it when I'm presented with it, is just to keep asking questions and then try and let those questions reveal a reality, like a kind of shitty tin pot Louis Thoreau. the person just keeps saying, What about this? And you say, Okay How did you arrive there? Can you explain to me the bit before what you're showing me?

and keep putting somebody on the spot. It's more difficult when it's just a supposedly funny photo. The foundations of it seem invisible and also have no impact on people, even when you lay it out to them. You know, if you use this platform, everything is being seen by third parties. People don't care. So I don't know if there is an answer to your question, Joe. How do you get people to stop?

I don't think you can. It's very, very hard. If you go incredibly aggressive, then you repel them twice as quickly as far as I'm concerned. And if you're too much of a soft touch, they don't change either. Inculcating real change is incredibly difficult. Yeah, I mean it's not a technical problem, is it? It's very much a people

and societal problem. It's absolutely a societal problem. And it has just resulted in the very quick erosion of people's critical thinking skills, both in terms of Thinking about what is happening when they upload the image to the IR model. or thinking about the wider context behind the thing that the IR model is returning to them. I'm just really shocked that we've gone from a place where people very readily had critical thinking skills.

and would genuinely read documentation for something, try and understand what they were doing and the why and the how. to just blindly believing what some model that's ingested something tells them. Well I think that goes back to the same mindset if you think about trying to solve a problem with Linux or some application in Linux. Once upon a time

people would just go and Google it and they'd find some, you know, stack exchange post, whatever, and just copy paste the command and, oh, right now it's working, with no attempt to understand the how and the why. But then people like Chris will always try and understand the actual why of what that thing is doing and like read the manual. And you know, I think I'm guilty of the former.

Not quite a hundred percent, but not to the extent that I know, Chris. I mean, you've talked about with the Python stuff, for example, how you decided to learn that properly. Yeah, and it's the same when I had to learn git properly. I started with what the actual file system underneath was, which The particular course I picked was highly recommended and as I was doing it I didn't

I was like, why? Why the hell am I looking at this so deeply? But it was incredible as a foundation. I'm not saying that everyone has to do that, by the way, but if you copy and paste and something goes wrong taking the time to explore the why of the wrong is incredibly valuable. Making mistakes are really important, but making mistakes without any kind of reflection on why they happened is terrible. And that's why I really dislike this kind of the the word vibe.

'Cause that's what that means to me. It's just like, that was wrong, do another wrong thing. That was wrong, do another wrong thing. That was wrong. That was less wrong. Why were the other things wrong? I don't know. I don't care. This is less wrong. Go ship it. I can't bear it. That to me just means you've switched your critical thinking off and I'm not interested.

I think what's worse than all of this even though, is that the reason we're here is because people realize that it makes a lot of money. This isn't just some advertising on a website with high quality content or anything. This is supplanting the entire need for other people to be in your life. And when you get down to that level, well then what do you do, right? Like try and tell someone, Oh yeah, that LLM you're using is programmed to serve the billionaire that's controlling it, not you.

That could be directly in the case of Grok telling you how Elon Musk is great. Or like in the case of OpenAI indirectly because they're opening up to advertisements baked directly into word predictions that you get out of the L L M. But it planned all my holiday. The itinerary helped me get the visas and all the rest of it.

So it serves me. It just happens to have told me to stay at this hotel and go to that restaurant because they pay. I mean I don't think that's happening right now, but that will be happening surely. No, I think that is happening right now. And that also happened before with ads. Like if you Googled

restaurant in Reykjavik, I'm absolutely sure the the top result would be an ad. And people didn't realise that those were ads either. Yeah, it's how people got scammed for, you know, the European health insurance card, for example. For ages, people were just giving money to a form which just went directly into someone's bank account and they just ran off.

Because they Googled how to get a European health insurance card and the SEO was positioning this scam site high enough up that people would just click it. I saw a post I think it was on Mastodon, I don't know who it was, so I'm sorry I can't cite my source here. But it was something along the lines of they worked in a university or something and the kind of literature department said that AI is useful for doing like coding and maths and doing artwork and design, but oh no, not for literature.

And then the art and design people said, Oh no, not for art and design, but it's useful for literature and maths and coding. And then the maths and coding people said, Oh well, no, not for coding, but it's useful for literature and art and so everybody seems to think that it's useful for everything else apart from the thing that they are an expert in, because they realise

having tried it, that it is a bit shit. And like, maybe that extra logical leap will get taken by people slowly but surely, or am I just dreaming? I think ultimately what the question we're boiling down to is like How do we convince people that real person-to-person contact and critical thinking are good things that they should seek out? I think it all started with the simple statement that we have had enough

of so called experts. I think you might be right. That's literally the way the world operates now. So critical thinking is devalued so much that sadly I don't w I don't want to be a a a massive doomer about it, but These uh chat bots and LLMs are perfect because getting rid of critical thinking in the majority of people gives you an incredible level of control. Well, yeah, it gives you as the controller of those people control and it gives

the people that you do it to an incredible level of confidence. So AI is Brexit's fault. Great. Thanks. I fear that they may both be symptoms of a greater problem, but yeah. I think all that we can really do is just bring people back into the fold of thinking critically and having good communities that they don't need a silly chat bot on the internet to have a friend.

And I think the only answer to how we do that is to do it ourselves. But I imagined this cool thing, and then it just pr came into existence because of this cool new technology. Therefore it's definitely positive. Imagine if you had a friend who's an artist who would do that, either for money or just because the thing that you said was so incredibly funny in their head. I think we just need to start the trend to get off the surfaces that are propping all these conditions up.

I all of those services are propped up by the same people who are also controlling the L LMs. Isn't that funny? Or we just wait for the AI bubble to burst, the economy to just go in the toilet and then we'll just Sit around a fire to keep warm or something, it'll be fine. Or it'll all be on to the next thing. I mean it was NFTs before this to give people a feeling of ownership in a world where they can't own a fucking house. Right, well, next episode let's talk about how great quantum is.

But on that positive note then, yeah, let's get out of here. We'll be back in a couple of weeks, but until then, I've been Joe. I've been Chris. I've been Gary. And I've been Matt. See ya later.