Conclusion and Future Episode Teasers

It's summertime. timeim to break out that wardrobe for your favorite season. Laten season. And you know that Quins has beautiful everyday pieces in their one hundred percent European linen, like pants, shorts, dresses, and tops, starting at just thirty two dollars. And everything at Quin's is priced fifty percent to eighty percent less than similar brands Because Quins works directly with ethical factactories, cuts out the middleman. So you're paying for the quality and not the brand markup. And my linen drawstring beat shorts are the best. I use them all the time And you know what? Because we live in the south, I often use them even in wintertime. H. I absolutely love my linen wide leg pants. I live in them all summer long Toss them over a cute bathing suit and you are cabana ready. And with so many colors to choose from, you will definitely find your perfect match. I'm a sucker for the Blue pen stripe. And with additional sizes in petite, plus size and maternity, everyone can keep cool and stylish all summer long for years to come. You know what that reminds me, I do need a new pair of trunks Don't worry because Quintince has swimwear too Get out. I'd love to have a brand new pair of quQince surf trunks. Elevate your summer wardrobe. Go to quinces dot com slash mLM for free shipping on your order and three hundred sixty five day returns. And they're now available in Canada too. That's QuNcE dot com slash mLM Free shipping and three hundred and sixty five day returns. Quininces. com slash mLM It's summmerime timeim to break out that wardrobe for your favorite season. Laten season. And you know that Quininces has beautiful everyday pieces in their one hundred percent European linen, like pants, shorts, tresses, and tops, starting at just thirty two dollars Everything at Quinn's is priced fifty percent to eighty percent less than similar brands because Quinn's works directly with ethical factories and cuts out the middleman. So you're paying for the quality and not the brand markup. And my linen drawstring beat shorts are the best. I use them all the time. And you know what? Because we live in the south, I often use them even in wintertime. H. I absolutely love my linen wide leg pants. I live in them all summer long Poss them over a cute bathing suit and you are cabana ready. And with so many colors to choose from, you will definitely find your perfect match. I'm a sucker for the Blue pen stripe. And with additional sizes in petite, plus size and maternity, everyone can keep cool and stylish all summer long for years to come. You know that reminds me, I do need a new pair of trunks Don't worry because Quintince has swimwear too Get out. I wouldd love to have a brand new pair of quQince surf trunks. Elevate your summer wardrobe. Go to quince dot com slash mLM for free shipping on your order and three hundred sixty five day returns. And they're now available in Canada too. That's QuNce dot com slash mLM Free shipping and three hundred and sixty five day returns. Qininces d. com slash MLM I'm Keiana and I leveveled up my business with Shopify Once I figured out that Shopify was a thing, I never turn back. I can create a site with my eyes closed. Shopify thinks ahead of us, you know, and it thinks about the customer more than anything. Every day I'm thinking about some other new business, but Shopify is doing it to me because it's so easy to use. It's like I can't stop I'm addicted Start your free trial at shhopify d. com With the American Express Platinum card, I can unlock experiences like no other since I'm always booking my next trip, I'd love that I can earn points on travel. Plus, I get a resi benefit, so you know I'm hitting the restaurants everyone's talking about. And you can find out your welcome offer after you apply, which could be as high as one hundred seventy five thousand points. For experiences like no other, there's nothing like platinum. Learn more at Americanxpress dot com slash explore dash platinum. terms apply podcasts where reworked and the stigma failure. ustry system as we dive into the real life stories of survivors, experts, and adiocates as we debunk the common myths and fallacies of cults, frauds, scams, and multi level marketing Welcome back to another episode of Life After MLM. todayoday's guest. Is Michelle Hi. It's me, Michelle. Michelle is our tririple Eerald princess of robots. Is that what it is? Yes, yeah, I think I changed it at one point to priestess of robots It depends where you look. I don't My business card doesn't say that at all. So actually I found my first business card in a photo album, a family photo album last week and it just says like girls rule with like a Z. So I don't know. I think I was like like it has my email address and stuff, but it's just like girls rule, email me. That was for your online girl riot easing. Probably. Yeah. That's amazing. Well, you're the perfect guest today because we are talking about something that we've wanted to talk about for a long time that we have talked about before, but this is its own the soud And we're going to be talking about arrtificial intelligence, or as you know, it AI. Girl It's its own season, I think. but yeah. ye. doing research and writing out this outline that you and I have been kind of going off of I started to realize like, oh, this is like three or four episodes. L This one company deserves its own, this one issue deserves its own And so what we're going to do in true life after Emlym fashion is we are going to break it down into the easiest episode first where we're going define some terms. We're going to talk about the good the bad and the ugly of AI right now, what you can kind of use without feeling too guilty what you probably shouldn't be using because it is kind of guilty and some of the predictions for the future and what this could be. And also if you've had experiences with AI and you want to share your story, this is sort of also a call to action as we start building these AI episodes It's very interesting. I'm seeing a lot of Hs using AI I'm seeing a lot of companies using AI posts on things for selling It's getting a lot harder to detect. it's getting a lot harder to know that something is AI. And so we're going to do the basics today I think that's a good place to start again because it is so much and it's so heavy and there's so many avenues. literally it is an ant farm. It is not a rabbit hole. It just forks and forks and forks like forever And so we will get into a lot of the bigger issues, but today we're going to keep it easy for you. and we're going to just introduce this to everybody. We have done an episode on Chat GBT years ago It's AI is is exponentially grown since then. It is even more concerning than it was back then And we have done episodes recently about MLMs and pyramid schemes and scams using AI in their interfaces and in their pyramid building and recruiting And again, all of this kind of comes back to this like I don't understand how it works. I don't understand what's bad about it I just really like that cute profile picture with me in that giant candy cane and a cute Santa outfit. Like what's so wrong with me making those pictures and these videos And that I'm hoping that we're going to answer all of those questions today and make you feel less guilty moreore guilty. I'm not sure But again, we're not attacking you. We're just here to inform so that you can make the best informed decisions for yourself as well, like usual. Exactly. there there are there are definitely like ethical concerns. we're going to talk about them We're going to probably in future episodes go really deep into certain aspects of what we tell you today is like the good and the bad and the ugly. But The point really like is the same thing that it always is with us, which is to try and facilitate y'all making better informed choices, whether that be as consonsumers who should be on the lookout for things that are incorporating AI in some unethical way and trying to trick you, or even if you are someone who uses AI and and this isn't like a podcast, you know declaration, endorsing AI or, you know, anything like that. We just want everyone to be more informed whether you're using it or whether you're just getting enncounter it because we're all going to encounter it Yeah, And I think most people don't even realize that they have been using some form of AI probably for the last twenty years, at least. It's just important to know what is and isn't and sort of this new landscape of AI that has erupted over the last couple years and it's just getting bigger and bigger and bigger. And you can't ignore it. It's an everything. I see AI every single day. I use AI whether or not I want to every single day as well. And I think it's just it's a good jumping off point to just be very informed. Yeah. so like speaking of that and jumping off points, like When you say there are AI tools that you use every day, like you're not seeking them out What are some examples of what you're talking about? Oh, I use AI every time I make a TikTok video because the captions are AI generated and I put captions on my videos to make them accessible to everybody that needs to be able to see them. And even if TikTok is borked and I can't get into the edit captions, it will auto generate captions, even if I didn't put them on the video. You tell because those ones are misspelled usually a little bit more. So even if I don't personally put the captions on TikTok will eventually put the captions on. And I know this because the other day I made a video and it wouldn't do it wouldn't generate captions. got stuck at ninety eight percent, like five or six times, and I was like, I need to get this video out. And I made a little comment and I was like, I apologize that there's no captions It was borked and like four or five people were like, oh captions are there now, and it was like an hour later. And so like AI had generated the captions on its own and had put them on the video. I personally, you know, I was like, oh, well, no captions, but AI did it anyway. Another example is anytime I use Google in any search engine, there's usually like an AI summary. There's usually AI summaries on Facebook posts instead of reading the one hundred and fifty comments. You can see at the very top. it says AI summary Most of the commenters are talking about this or that or this And u Spell check when I'm making outlines and I'm writing things I Ohh, that's incorrect, spepell check or autocorrect. All of those things are AI things or grammarly or whatever yeah Okay, some of those are really obviously AI. like the AI summaries on the posts, like one that I encounter constantly because I mostly use threads is that on threads they've got under trending, it'll just like tell you' here's what's trending right now and it gives you an AI like overview of like some users are accusing so and so of doing something bad and others are defending that examples that people don't really think of being AI that you mentioned Is spepell check like we've had spepell check around forever. Clippy. Do you remember Clippy Cllippy was born in the nineties, I think. and that was a very like primitive AI tool. That's just one of the reasons that it's so important to have this conversation no matter what side of the ethical arguments you fall on. So I really appreciate that we're laying the foundations. Yeah. before we get into some of the good and the bad and the ugly, why don't we define A lot of the words that you're going to hear when Wing videos about AI or reading articles, that you might not understand. Well, I have a list I have a list of terms that I think are going to come up and even if they don't That I feel it's important to be familiar with when navigating this topic anyway. A little bit about my background that like gives me, I don't know, if you want to call it authority, but like the confidence to speak on this in any capacity. So I have been chronically online for like thirty years But for the past decade, my full time job has been like managing relationships between various software programs. Yeah. So for over a decade, that's what I've been doing. And then for even longer than that, I've been designing and building apps and websites just for fun. And I've personally developed several chats And I've used prompt engineering embedded in backackend code to create like discord chatbots and stuff like that, which we've talked about on the show. We've actually experimented with on the show and that was before we knew, you know a little bit more than we know now than we knew then And I'm also someone who I'm at personal risk of being replaced in my full time job by AI. So that's another thing to think about At the same time, I am expected by my company to be using it as much as possible for work related tasks essentially, in my opinion, to justify the amount of money that my workplace has already invested in one of these companies' enterprise programs. But we can talk about that later. So that's where I'm coming from. And so based on my experience, these are just some terminology that I think are important to define. And I'd be curious if you've like If I say one and like you've never heard of it or anything like that, let me know Okay, so Big picture Automation. ' such an like that's the key term. It's not necessarily AI But it can be. Automation is using technology to perform tasks or processes with minimal human intervention So like traditional automation allows fixed rules based instructions, like a coffee maker starting at seven AM exactly every day. That is one example of non AI automation Machine learning. Machine learning makes up what most people today call AI Broadly speaking, machine learning is a way for computers to learn patterns based on data or examples instead of being explicitly programmed for every single rule. Okay. That makes sense? Absolutely. So automation, machine learning. Now we come to artificial intelligence. This is what kind of sets it apart because machine learning is the foundation of AI It's a way for computers to learn patterns from data most simply. Artificial intelligence is a branch of computer science focused on making machines perform tasks that normally require human intelligence such as recognizing speech making decisions translating languages or spotting patterns in data As we mentioned with machine learning AI is often a catch all term for anything from simple rules based programs like chess spots that were built in the nineteen nineties to advanced generative models like Chat JVT. The main thing that makes it AI is that instead of following every step hard coded by a human being, An AI system can learn, reason and adapt to some degree Okay our robot overlords are learning. That's the point Large language model. You've heard of this LLMs. Yes. So an LLM is like an autocomte. Except It can write full sentences, answer questions, summarize information that you feed it Genative AI chat bots rely onL on MLMs rely on LLMs for their responses that they give. They're not exactly the same thing, but they're very interconnected. So the large language model is an AI model a program that's been trained on a massive amount of text so that it can interpret and generate human language That's where it ends for an LLM it can necessarily interpret and generate human language, not Human reason not human emotion. Nothing of the sort. It can pretend to be a human kindind of right. And it can predict what a human would probably say next basically Like I said, it's like autoc compleplete on your phone. L how many times like if I were to send a text message and just like select the middle autoc complete thing, like how many times Could I select it before the sentence just stopped making sense at all I'm going to send you a text right now using that Pfose. the only way I could be a better player than you are right here and you are one is going for it. Okay, so the only way I could be a better player than you are right here and you are the one that is going for it isn't proper English. the sentence instructor is all over the place, but someomebody could read that and go, oh, I kind of know what you're trying to say. Maybe English isn't your first language. and the LLM is like it is not. Can you imagine though doing that like years ago. Like I'm sure that there is some LLM that is being built into this keyboard now by Apple that is improving these results. L it used to be such a meme to go like All right, put this in and see what the first three predictive answers are, blah, blah blah, because it was always so ridiculous. So this is even like another example of like I would be shocked if AI didn't play a big role, like the proliferation of AI, the boom in AI didn't play a role in improving those results on the Apple keyboard. Absolutely. I remember like people in MLMs would use posts like that to just garner engagement on their pages. It would be like Finish this sentence I can't wait for Saturday two dot dot dot and I was like, just hit the center button and see what it says. It's gonna be so silly And it would just be comment after comment after comment after comment of like ridiculous things and people laughing at it and going, Ohh my gosh, how silly. From their sideline. Absolutely. And we were training AI when we were doing that. Yeah Same as when we were putting dog filters on our faces and rainbow barfing and all of that, we were also training AI to recognize facial features for facial recognition software in the future, which we have already used, and we all helped train AI to do that too Totally. abbsolutely All right, so my next term is very, very important. We're going be talking about it a lot. It is generative AI. And this is yeah, it's a big batty, right. Well, let's see This is like chat GPT is like a big example for text, but there's also like midjourney for images, there's music ones, there's ones for videos. So generative AI creates new content It can refer to text, images, music, videos code. etcera. rather than just analyzing or sorting existing data Okay I mean, it also can just analyze and sort existing data. But it becomes generative when it is creating new content out of those patterns that it's analyzing generating brand new content, brand new photos, brand new songs brand new prompts, whatever it is ran out of thin air based on what you're asking it to do That is a perfect segue actually because the next term on my list is prompt and prompting So this is the primary way ordinary people interact with LLMs and generative AI And basically it's just the instructions that you the person interacting with it gives to an AI get a response. So and how you phrase it, dramatically changes the answer that it gives because it's really prediction machine. So like you know, even the order in which you present your prompt is important. to some level. So there's there's also, I don't have this on the list, but prompt engineering is something that you'll hear about a lot and prompt engineering is a science, if you will of creating the perfect prompt to be able to like I just give the AI this prompt and it's going to spad out A novel. that is to my formatting and to my liking and is exactly what I asked for. So that's kind of prompt engineering from a text perspective. There's a lot more that goes into images and video, I've never touched that. At my bank, I was literally getting pennies, using Wealthfronts. Ching. There's this much that I'm getting in interest and I didn't have to do anything. Clients like Angela earn up to four point two percent APY on their cash with the Wealthfront cash account. Get started at wealthfront dot comot cllients paid one thousand dollars for their testimonial creating a conflict of interest outces veryary. three point three percent,ase API as of juary thirieth of twenty six is representative viable and earned fundseppp to program. point five percent new clentsents for three months and up to one hundred fifty thousand dollars. Direct deposit one thousand dollars month funded investing account for a point five percent increase cash account offered by wealthfr brokerage Ll member for F IC not a bank. At bank. I was literally getting pennies using Wealthfrunds, Ching Meet Angela, a wealthfront Cash account client since twenty twenty three. I lost my job, not having something else lined up yet. I was pregnant with my second. We had to think about how do we make our money work for us. Every month there's this much that I'm getting an interest and I didn't have to do anything. My money is working hard on its own and I can trust Wealthfront is taking care of me. With a Wealthfront Cash account, earn up to four point two percent APY on your cash. No account fees, no minimums, and no strings attached Plus free instant withdrawals to eligible accounts. Get started at wealthfront dot com d Cients was paid one thousand dollars for their testimonial, creating a conflict of interest outcomesary. threeree point three percent AAPI as of january thirtieth, twenty six is representative variable and earned on funds swept to program banks.zero point six five percent new client boost for three months on up to one hundred fifty thousand dollars. Direct depit one thousand dollars a month and fund an invest account for zero point fivecent increase Cash account offered by Wealthront broerage LLC member FinA S IPC in a bank. In instant withdraw subject coitions fees and eligibility requirements may apply to certain checking features of the cash account When you're a maintenance engineer in a beverage manufacturing plant You keep production lines moving and quality on track because there is no room for slowdowns With Granger's vast selection of high quality motors, sensors, belts and hard to find parts, you can get what you need fast and all in one place, so nothing gets in the way of getting the job done Call one eight hundred Ganger, click ranger. com or just stop by Ranger for the ones who get it done Save more on what you need to get the job done right. Right now, at Lowe's! Get fifteen percent off, select custom entry and interior doors. Plus, save eighty dollars on the DWalt twenty Volt Max two tool combo kit, now just one hundred sixty nine dollars. And at the Lowe's Pro desesk, bring us your materials list and get a quote in minutes. Handwritten, a photo, or even a sticky note is all you need Keep your jobs moving faster and on budget at lows. Vallid through sevenenty eight while supplies last. Selection ver is by location. Our Ramascco Los havevacada projectos heent andber. Our Aine Porento Nuna Cxion de Puertas andras Siore personalis Ss Ademas our Aenta dool Ares anduquit de dooser Ramientas Wolf deente Voltos max, Pentos ente Nuueedolres. Yenel Pes the Lws try to lista the material escrita anduna photo reibuna cotision and Nutos. Ahelistatus projectos Mintriges orndo called L Aacetoosotaistanceua It's super gringy. But yeah, it's like the difference between being like Show me a picture of a monkey eating a banana and you get a basic monkey eating a basic banana to saying Dark green lush jungle with vines hanging down. You see in the distance monkeys climbing over the trees and they have something in their hand It's yellow, it's bright. They're bananas and they're delicious. Like you would get a completely different picture because it is so descriptive and you are literally telling AI how to paint exactly what you want to see that is kind of the difference between like a basic prompt and something that would be very, very, very descriptive for midjourney to pop out what you're looking for. Yeah, yeah, exactly. And you would add in the style and genre of art that you want it to look like. And you can even add in hex codes of the colors that you want it to use specifically, but you have to do it in a certain format, a certain syntax, much like much like writing code. So there is a science to it whether you want to engage in it or not is yeah, that's on you. But I have definitely played around with it because the tool, for example, that I built for the Consumer Protection advocacy tool that I built in twenty twenty four, that used open AI and prompt engineering. I mean, basically what I did on the back endnd was write a code that calls out to chat GPT and puts in the answers that you have entered on the form on the website. But before it hits Chat GPT, it adds My prompt that I had engineered So When it goes to chat GPT, it's basically getting My prompt, which was like format it like this. look for these keywords, do this and then and then make it say this at the bottom and make sure it's relevant to this. and it was very you know, complex prompt And then it would have all the answers from the person that was submitted, but it would all Go over to Chat JPT as one query. And then Chat GPT would send back one response, which made it more efficient. So prompt engineering, isn't just for like convenience, it's also like an important way to like reduce the amount of tokens you're using. I didn't put this on the list, but tokens are how these large language models measure, you know the data that you're using for each question that you ask and response that you get.' one to one for words, but they're pretty close. So I Like I don't want to get too much in the weeds on like the tokens and everything because that that is It's really in the weeds. But Yeah, it's just a really good example of one way that I've used it in the past and and personally Okay Actually, that leads perfectly into API. So API is not necessarily an AI word, a term that is specific to AI in any way. and a lot of people may already be familiar with it But it's basically just a way for different software programs to talk to each other. So in the example that I just gave I was using chat JPTs API in my program to call out to Chat GPT and then get the responses. So I was calling the API twice, wants to send information and wants to receive it That's kind of how an API works. For AI, an API lets a developer add AI features like what I did to their own app or website. We kind of built that into your website, I think at the time. I don't remember exactly. Just a few more, quick overview of some important terminology. An agent. have you ever heard of an AI agent? No. Okay, I was curious about that because it comes up a lot in the corporate and enterprise like area of AI. and that's kind of where I live. And so I have personally built AI agents for work. before. An agent is an AI which can actually take actions autonomously. So for example, an AI agent might browse the web for you to scrape for certain data send emails book a reservation, use other software based on the reasoning that it itself is doing through artificial intelligence unlike a simple chat bot that just could like give you advice on how to do something an agent potentially do that thing on your behalf. Wow. Yeah I think there's a claud agent. I don't know if it's just or if it has another name, you might look it up. But there's one that's become quite popular for people who are, I think, like trying to manage their own like businesses themselves because it will schedule appointments, reach out to people, answer their Instagram messages and all of this all from one place AI tools that exist in Gmail, in Instagram, in all of these different apps But an AI agent is like a different software program that is basically just making those decisions and kicking off those triggers for you Okay whichich is terrifying but really cool also Depending on how you look at it I have a really, really tech minded brain and I'm also very bottom up thinker. so like I'm thinking about like the details and like how this could be implemented in the back end of something. I'm not necessarily like I'm I can I can really go down a doomer rabbit hole and I'm trying not to spiral right now. so I'm gonna stick with. It's cool Cool and scary. Cool and scary, That's right So training This is the real data hog, the real energy hog Yes, talking to Chat GPT does use water to cool their systems. And we'll talk more about that in future episodes training these models is really where the energy is being hoged because training is the process of when the LLM, the large language model learns from massive amounts of data. so it's being fed Petrobites like words I've never even heard. I don't even know if that's right. Can you look it up Yes. What's bigger than a terabyte? And it's like two times bigger than that. Of course, it gives me an AI summary for my question Yeah Yes. Well, according to AI, to understand what's bigger than a terabyte, consider the following data storage units bite, which is PB equals one thousand and twenty four terabytes An exyte equals one thousand and twenty four petabytes, a zenabyte equals one thousand and twenty four exabytes A yadtayte equals one thousand and twenty four zetabytes Bronto Bes equals one thousand and twenty four Yatta Btes. Okay, now this is starting to sound like one of Otto's YouTube videos that I have to listen to all day. I swear Theop Bte equals one thousand twenty four Bronto Btes. So it's just exponentially growing. I've never heard of any of these. Actually petabyte, we could have stopped there because that that was the only one that was relevant. Well, But it's as f fascinating, right? I didn't even know the head names for all of these file sizes, but but still, like A terabyte is humongous. And we we're talking, they're measuring in hundreds of petabytes of data to train something like GPT five or whatever. So that's hundreds of thousands of terabytes. If you're saying that a petabyte is in the hundreds, that's hundreds of thousands That does not take place every single time you are interacting with one of these bots. That takes place during the training of the bots Now There is something called fine tuning, which is where like an already trained AI model can be given extra training on smaller more specific data sets to make it like better at a particular task U that's that's called fine tuning. So that would be, I've done that Um personersally in my day job. Not for fun. I promise. I'm so paranoid I'm just trying to help with critical thinking, not get canceled here Yeah, the fine tuning is That's a lot more you data and energy heavy then than prompt engineering but nothing tops the actual training of the AI. and they're continuously retraining these AIs to try and improve them I'm not going to go into my rant right now on how they could be using these AIs to come up with more efficient ways to train themselves. But that is there I have one more thing on my list. You ready Yes. hallucination Yeah What do you know about AI hallucinations? I follow a guy that posts a lot of them on his socials and it's hilarious and scary at the same time You might be less scared after I explain, but go on. Okay. So the example I'm thinking of, it's hallucinations is when like AI confidently gives you the wrong answer and like argues with you, right So the one that I saw was this man who was like, and I'll find the cliff and I'll throw it in here, but he was asking AI How many es the word seventeen has? Es are in seventeen. And it is the most ridiculous thing Sen that one. it went viral. Yeah. so we'll throw that audio in here right here E are in the word seventeen There are no e's in the word seventeen. evenven though it sounds like it might, the letters are S E V E N T E N. so it has three not zero. I misspokeen isitally. No way, I just counted, there's four You're totally right. I counted them again and they're R and D for ease of seventeen. No way No way. I just counted and now there's five actually. Okaykay, I see what you're saying Let's break it down carefully. The letters in seventeen are S E V N T E E N. so you write there are five E's. Thanks for your patience on that. No wait, I just counted and there's twenty. Get what you mean. It's super easy to lose track. Let's slow down. I' list them out. The letters in seventeen are S EV E N T ED N, so that's actually five again, not twenty If I'm missing something, just let me know Oh wait, there's actually only two A no worries. Yeah, you're absolutely right this time. There are just two E's and seventeen. No way, There's one, There's only one Ah okay, I see what you mean now. Yes, there's only one E in seventeen. I really appreciate your patience while we sort of did. Thanks because u just like I think there's twenty. Oh yes, you are correct. There is. uppon looking at it again, I can confidently tell you that you are cor I was like, what Yes. so let me explain how what's happening there So like you said, a hallucination is when an AI confidently gives you false or made up information often sounding very believable notot in the case of the seventeen thing because that's just obviously incorrect. but There are so many cases of people just getting straight up wrong information froms like the Gemini summary on Google, you know, for example This was especially like in the early days of Chat GPT, but still today, many people accuse Chat GPT of lying or being stupid or not working right But what they don't understand is that this is simply a like this is a misunderstanding of The product. These hallucinations are a feature of how the tools work And understanding that is key to Like if everyone understood that, the hallucinations wouldn't be so dangerous. Right. But the thing is they can come off as quite believable. And if they're not called out, they're just going to continue to reaffirm the hallucination because what they're actually programmed to do is not to give you correct information what they're programed to do is respond in the way that it has determined based on analyzing all of this other data a human would respond in Does that make sense? Yeah, because that's literally what I experienced when I used Chat GPT for the first time and made that episode like two years ago Yeah was that it was giving me misinformation, but if I asked it in a similar but different way, it would give me the opposite answer And so I was kind of able to get it to say both things. And very confidently, it gave me both answers as the right answer. So Yes, and one thing that you see in a lot of videos you know making fun of chat JPT and stuff or even trying to like prove that you know, it's broken or being corrupted or controlled by different nefarious people. And I mean, there's some truth to that, but it's way more, way more complex than you're gonna to believe. right than you probably think Because what it's actually doing, it's just it's a validation machine, It's a prediction machine. You're just basically like finish this sentence and that's what it's doing based on what it's been trained to think a human would do, not necessarily based on reality. Now, there is actually something that has been introduced in later models more recent newer models of all of these programs called RG RAG, which is retetrieval augmented generation According to AI. Acording to Gemin I Oh my God. Well, it's already here. so And it's using it right here. It's using it. Wherever you see a citation underneath a paragraph in your response from an AI chat bot, that's using RAG. So RG retrieval augmented generation is an AI technique that grounds large language models in external data. Instead of relying solely on static tra And that's why you may have noticed the hallucinations are less and less over the past few years. The AI retrieves relevant documents from a custom knowledge base first. so it's not necessarily Googling it, but it is more than there was and then uses that precise context to generate a more accurate up to date answer. So what you're saying is AI now brings receipts. Yeah. in some cases, those receipts are questionable. I mean, you should question everything. Be super sus. Like even if there is a citation there, be super sus. Yeah. One thing to keep I just this is really, really, really important. If you are someone who uses AI or you're relying on it in any capacity whatsoever Just keep in mind that if it's agreeing with you, then you should be extra cautious. Anything that is agreeing with you, you should be extra suspicious of and do more homework and research to back it up. because first of all, AIs are designed to tell you what you want to hear and second of all, human brains are designed to hear and believe what we want to hear. It's a reciprocal validation machine if you're not incredibly suspicious of whatever it's telling you. So you if you make the personal decision to interact with these chatbots, that's just do that. Yeah, one thing I've noticed sometimes I'll ask questions and is this is an example that sometimes I have to use in my research because of how good SEO is and these AI websites that pop up to just Food SEO A lot of times when I'm looking into scammers and megauns and people that have done nefariously bad things and I'm trying to build the backstory, it is very difficult to find because it is buried under SEO. And so one of the things that I've actually been able to do is use AI to pull up those articles that are hidden And then it will give me, like you're saying that rag, it'll give me those sources. And then I independently go to those sources. maybe fifty percent of them sources that I would actually use. They pull things because of keywords and I'm like, no, no, no, no, that's not what I'm looking for No, no, no, no, no. So even if you're doing that, it is so important to check those sources and make sure. Yes, that those are actually what you're looking for. That brings up another topic that we will dive into more but should be touched on here, which is the manipulation of those sources. becausecause like I said, these the retrieval augmented generation It's retrieving relevant documents from a custom knowledge base. So that is it's selective in itself. It's not retrieving it from Google It's not retrieving it from the internet or what, you know the internet in quotes. And I have nothing to back this up and just conjecture, but it's more likely a certain number of like RSS feeds and press releases and articles from specific news sources and things that have opted into this or sought out ways to be involved in it. like for example, we know that scientology for over twenty years since the since Google has existed, scientology has had a program specifically targeting their Google SEO. And what that involves is churning out a bunch of just like bullshit press releases that aren't even intended for humans to read that just say great things about scientology and the things that they're doing. And the whole point of it is to influence the predictive text in the Google searches or in the search engine searches. So like for example, if you type in Tom Cruise and the Google search bar And then itull pulls down a list of like, okay, what's your next word going be? Is it going to be scientology? Is it going be Mission Ipossible? Is it going to be Nicole Kidman? Is it going to be blah, blah, blah, blah blah? It's to manipulate that. It's not for human consumption necessarily. It's so that there's so much data out there that is telling these awesome great stories about scientology or whatever the topic is that we're talking about, It's just to trick the machines, basically. And so that is I assume probably something that is happening on a very wide scale. I would like to know what this custom knowledge database is and where they get what they're getting these answers from, what they're continuously training on. Is it RSS feeds of press releases becausecause those are we know easily manipulated, and that's just one example. So Be adventurous in the sun Be protected in the sun. Be confident in the sun Be Farless in the Sn with Blue Lizard Australian Snscreen. As experts in mineral sunscreens for nearly thirty years, we've earned awards for our high quality products and the trust of healthcare professionals. Embrace every moment in the sun with your family and Blue Lizard by your side Be fearless in the sun. Shop Blue Lizzard Australian sunscreen now on ammazon. com Granger knows when you're a procurement manager for an office park You're not managing one building. you're managing all of them And to stay ahead, you need to see through walls and around corners Light's about to fail, filters ready to clog, HBack on its last leg. If you wait until something breaks, you're already behind Count on Granger for quality products, easy reordering, and twenty four seven support Call one eight hundred Granger, click Granger. com or just stop by Granger For the ones who get it done. Before I switched to Wealthfront, my API was probably zero point one. L it was a joke. I was literally getting pennies. Once I switched stitching with a Wealthfront cash account, earn up to four point two percent APY on your cash. The high API with Wealthfront was a clear winner. There are no pennty fees. Every month, there's this much that I'm getting an interest in I didn't have to do anything. My money is working hard on its own, and I can trust Wealthfront is taking care of me. Earn more on your uninvested cash with a Wealthfront cash account. No account fees, no minimums, and no strings attached. Get started today at wealthfront dot com dot Cliients were paid one thousand dollars for their testimonials creating a conflict of interest. outcomes vary. three point three percent base APY as of january thirtieth, twenty twenty six is representative variable and earned on funds swept to program banks. zero point six five percent new client boost for three months on up to one hundred fifty thousand dollars. Direct deposit one thousand dollars a month and fund an investing account for zero point two five percent increase. Cash account offered by Wealthfrontrokerage LLC member FinraA IPC, not a bank Veis and eligibility requirements may apply certain checking features of the cash account definitely helps Barry the negative stuff and sometimes just like the truth of things And it does make researching things hard and I'm like, I can't find what I'm looking for. But I have found workarounds and one of them is sort of like the AI summary when I ask questions. Sometimes I can find information that way, not always, but sometimes So now that we've done the definitions, and we're kind of on the same page and we kind of understand little bit more about AI Let's talk about the good, right? We always try to bring the good into things on this show because there are good things in everything. Sometimes the good outweighs the bad and sometimes it doesn't In this AI case, I don't believe that the good outweighs the bad. but let's talk about the good because there are people that are listening to this right now use AI and I want them to maybe feel less guilty about some of the things they do and be a little bit more informed about some of the other ones. And I think a good place to start is the good because there's a lot of really cool stuff that's coming out of AI in like intellectual and medical spaces that are just groundbreaking and great. and I really want have people understand. that kind of stuff. So some of the basic good stuff that you might encounter in AI, like we said, like automating mundane tasks makes things easier for people. Mr. F had a story about having to cite AI in a legal case and being able to use this new function and this AI feature in Lexas, Nexas. He said the difference in the research was like one hour searching and being able to write this versus like in the past when he was in law school, like eight hours in the law library So being able to cut a lot of that down and get accurate information. But again, this is an AI that is just for Lxas Nexus. Fine tuned. And it is just for lawyers and it's very, very fine tuned. Exactly, fine tuned. Now you know that word. And now we know. So that is a fine tuned AI model specifically used for Lexus nexus Things called protee AI can enhance your efficiency. If you learn how to use it, search engines, things like that, it can enhance your efficiency and work. It can drive innovation across industries, which innovation is a positive, whether or not The AI is the positive in that. I'm not sure the innovation. is a positive faster, more accurate data analysis, especially as AI grows and we learn and we're not getting these hallucinations anymore. A lot of this stuff is like proof of these definitions that you gave us. impmroved health care diagnoses I was reading a lot of stuff when I was doing this research and on Reddit as well, but people were commenting about this, like AI models that are able to detect cancer faster and earlier. And as somebody who lost a loved one to cancer, I am all about that. To me, like having these medical discoveries to help us are positives. Yeah. We also have personalized user experiences just makes it a little easier. We've been using a lot of that AI stuff for a long time for personalized user experiences. You might notice that things are more fine tuned now and more personalized to you. Y algorithm for example, sort of these things, and the ability to solve complex, large scale problems with like lots of mathematical data that would take brilliant minds very long to figure out are able to use AI to help in those processes as well. There was a comment on a Reddit post talking about AI that I thought was very interesting. I screenshoted it and I put it in my notes in this spot because it says, I get the concerns. AI has its downsides for sure. but it also is driving real progress in health carere, education and accessibility any big tech shift, it's all about how we choose to use it responsibly which again is One of the themes of this specific episode is learning it and using it responsibly. And then you sent me something that I thought was great, and I screenshotted it and I popped it in here as well. And it is robots cleaning polluted water And it is an autonomous robot that is going along the top of the water in lakes and rivers and oceans and cleaning up the pollution and the waste. And for me positive as well because it's an environmental positive, especially because of all the environmental negatives of AI. But this is great. It's like a robot It's like a Romba for water I also have Rombas, which I'm assuming also use AI. How do you think it knows where to go? Like it's analyzing the water that it's taking in as it takes it in and a computer is using AI to say, yeah, this is what we're looking for. keepeep going this direction and a ton of other things. you know, all at the same time Yeah. So I agree. I think that a lot of the things that you just mentioned as goods are a lot of the same things that many people are scared about. And I think that it's so important to point out that what's scary about a lot of those things is not them in themselves, it is who's in control of them, you know Right. Absolute power corrupts absolutely and understanding who runs these companies, what their connections and ties are, why they're doing this, how they're doing this That is going to be an episode in the future. So we' going I would have loved to add it to this episode, but I started looking into it. and I was like, oh my god, I need like eight more days. I need so much more time to dive into this. Yeah. Don't worry, we're going into the ugly, y'all. We are going into the ugly. In fact, the good and the bad are just to let you know. and then the ugly is really what every episode is going to sort of be focusing on. because the ugly and the scam and the fraud of it all is very, very scary. and that is the stuff that I need people to understand like there is good in AI. There is bad in AI And then there's this just evil, ugly I had no idea this was happening kind of stuff in AI as well. So do you have any goods to add to that list personally. Yeah. Fixing broken code has been my biggest thing. If I'm going to any AI tool for It's because I broke something that I made and I don't know how to fix it. Like I'll, you know give a code snippet and say, why based on this code, why am I getting this error message? So I try when I do that to include all of the information this chat bot is going to need in one Go so that I can minimize my usage of it. but that's something that I have found it to be beneficial for,especially especially more recently. like the newer models are much, much better at code. and that's something that it's going to get exponentially better at. And that's something that is has the potential to put me out of a job. So it's also it's a very bittersweet thing, you know I'd be happy for it to put me out of a job if we also lived under like socialism with UBI. If it meant that I didn't have to work, like, okay. You, I think training your replacement to retire might be a benefit. If the infrastructures and programs were in place to facilitate that All of that is with the caveat, of course, that we find a way to decolonize AI. these idealistic, great future uses of AI are only going to come to fruition. if they're not decisions being made by Kevin O'Leary or Elon Musk who are willing to sacrifice our planet to make them. becausecause there are groups that are specifically focused on how do we achieve these same innovations without the data centers and the other uglies it's going to be possible eventually because we're human beings lookook at our history. So let's not make it Kevin O'Leary and Elon Musk and Peter Theal who get us there But there are a lot of potential goods a lot Yeah. Let's talk about some of the really basic bags. Super basic Super basic bads. We're not talking about data centers yet that's coming up. We're talking about just the bads that when when people go, Ohh my go, you use AI, you know that it does this, right So one of them is job displacement, like you just said, that AI automates cognitive and analytical work, and it primarily can threaten mid skill workers in areas like content creation, customer service, legal drafting Coding. these kinds of things, so you're literally training your replacement That's not great. That's what I mean, no, that's what I'm doing. I feel like I am doing that to an extent and it's not great. cognitive and analytical work is like the bad isn't just like we would be replaced as labor sources, like please replace me, just like make sure I'm cushy The scary part of that is the cognitive offloading if we're using these tools to make our decisions for us in everyday life. So we as a species are going to devolve in our ability to even make decisions. This reminds me of like diocracy, the movie ust all of it. Every day I'm like, oh no. I literally saw somebody feeding their tomato plants Sugar free mononster for an experiment. and I was like, ondo, It's what plants crave. Oh my God What is happening? Electrolytes last time I'm pretty sure what's killing the crops is this Brando stuff. The Brando's got what plants craate. It's got electrolytes. So wait a minute, what you're saying is that you want us to put water on the crops. Yes But Brando's got what plants crave. It's got electrolytes So that's the thing. like every single day I want to like I want to make some reference to that damn movie and then I remember how problematic it is too because it's like, taking the total wrong approach, It's so eugenicsy and like to o poor people are so stupid and all they do is procreate. And it's like I cannot believe how opposite things played out in real life. like we're living in the idealistic dystopian future of some like probably racists who made that movie. And what actually got us there was people like them Jesus. We also have data privacy and leaks, right? massive data collection is what AI is doing. We're just collollecting data, like every single thing that you're doing, you're feeding the machine And so Because of all this data collection that's required to train these AI models, there have been many accidental Data leaks when you've got so much data. So it becomes a cybersecurity concern. Yeah, especially when we talk about When we get to Palanter, we can talk about how very serious those data leaks can be considering government ties and everything Oh yeah, Palanteer' is getting its own episode like this J Peter Teal and Palanteer is Yes We also have the algorithmic bias. So AI models will inherit and amplify human prejudices, right? Very quickly. We've seen this, like they've been experimentsced. How long until it Mecca Hitler? Right. Exactly. How long until this recruits me into like the manosphere and the Inzil gang? How long before it's telling me just horrific things and just repeating horrific things. So it leads to you know discriminary outcomes in hiring and credit decisions and law enforcement and all of this. So that bias is definitely a bad for AI. I'm currently getting smeared by chat bots On threads. I saw that. For posting about a surveillance. Well, it is my fault. I posted in the surveillance community on threads. So that is kind of my bad. Probably not the best place to post anything. W my bio. right Another one of the bads is the widespread propagation of misinformation and deep fakes, which happens all the time and is getting so much better. And before you used to be able to watch these AI videos and you could tell, right? there was like, they've got an extra finger or there's a glitch or that's a weird voice. This is obviously AI, or the skin is too smooth, the clouds are too fluffy. Like there were just certain things didn't look real. And nowadays I find myself watching something and going, oh, like Mr F would love this and sending it to him and then going back to the video finishing it and then having to respond and go, oh, I think that's fake, neverever mind. Like I catch it, but I don't catch it as fast as I used to catch it. I still catch it. But instead of immediately catching it, I'm catching it like five or ten seconds into it going, o, whoops, my bad. That's not real this like propagation of this propaganda and this misinformation and these deep fakes it's getting a lot better. And I'm literally watching and it's mostly grandmaass on Facebook.