AI Dangers for Dummies and What The Hell We’re Supposed to Do

[00:00:00] Chalene Johnson : Every day I learn something new about AI or something that you can do with it that makes me feel like I just have absolutely no clue, and I wonder if I should just give up and stop trying to master this thing What I do know is the more I learn, the more important I realize it is for me to not give up, because this is not going away.

[00:00:21] Chalene Johnson : We can’t stick our head in the sand and say, “Well, I’m not gonna use it, and that’s how I’m gonna help humanity,” or, “I’m not gonna use it. I’m, I’m just gonna pretend like it doesn’t exist.” We just can’t do that, and I spend a lot of time trying to understand this stuff. I polled my audience to find out how much are you, the listener, how much are you using AI, and it was very interesting.

[00:00:44] Chalene Johnson : I will say most of you are using AI, like ChatGPT, to write emails, to plan vacations, to look up information on the internet. That’s the majority of users, especially this demographic. Meanwhile, some of the smartest people in the building who are developing this technology are currently looking very nervous, wide eyes, eyebrows up, wringing their hands, basically saying, “Oh, yeah.

[00:01:11] Chalene Johnson : You know how we told all of you, like, not to worry about this? It’s just gonna make the world a lot better, and there’s nothing to be worried about. So anyways, we were wrong.” It reminds me of the Young Frankenstein movie. You know, not the green guy with the bolts in his neck, but, like, the original story, the original movie.

[00:01:29] Chalene Johnson : The brilliant scientist creating something incredibly powerful. He sends his assistant off to put a brain into the head of this patient that he’s working on, and he doesn’t realize that the assistant brought back a brain that was unfortunately marked abnormal, I think is what it said. And then suddenly he realizes, “It’s alive.

[00:01:55] Chalene Johnson : It’s alive. It’s alive!” Right? And he’s freaking out because it’s too late now. That part is the part I wanna understand. I’m not a tech expert. I’m a normal person who uses AI every day. I see how unbelievably helpful it is, how much it has changed my life, how much it has helped me to be more efficient, to spend less time working.

[00:02:17] Chalene Johnson : I wanna know why the people who are building this, who, like, really understand this, appear to be freaking the out. Is this legit? You don’t have time to go down this rabbit hole, and you’re getting information in your newsfeed that maybe you’re wondering if it’s hype, if it’s hysteria, or is this something we actually need to be paying attention to?

[00:02:41] Chalene Johnson : And like I said, just refusing to use it is not going to save humanity. That is not the answer. I wanna know what is real, what is exaggerated. What is it these people know that we don’t know, and we probably should know? Most importantly, like, what does this mean for me, for my kids, for my hopefully future grandkids?

[00:03:03] Chalene Johnson : For my future, for my money, for my privacy. What does this mean for you, for your work, for your, your future, your family? What is actually going on, and what do we do about it? What should we be doing? I’m gonna answer that and a whole lot more today, here on The Chalene Show. Thank you for spending some time with me today.

[00:03:24] clip: I want people to get that we do not know what the we’re doing right now. We’re releasing this technology way faster than we know anything about how to control it, and the only reason we’re doing it is because of the arms race dynamic. Because if I don’t do it, I’ll lose to the other guy that will, which means we’re racing to the uncontrollability red zone, and most people just don’t even know that it’s happening.

[00:03:43] clip: And it doesn’t have to be this way. We can have a, a controllable AI future that is actually pro-human, but you can only do that if you actually know about these examples and know what makes this technology different from other technologies.

[00:03:53] Chalene Johnson : Okay, that was Tristan Harris. He’s a former Google design ethicist.

[00:03:57] Chalene Johnson : That’s a job. He’s the co-founder of the Center for Human Technology… I’m sorry, for Humane Technology, and one of the people featured. You probably saw him in the documentary The Social Dilemma. Phenomenal movie. His work is focused on what these technologies do to human behavior, especially when it comes to companies that are releasing things faster than society, and even the experts who are developing them, can understand the consequences.

[00:04:29] Chalene Johnson : Now, if you’ve been paying attention to the news cycle, like I said, this is all any program is talking about. It feels like hysteria. Like suddenly the sky is falling. It seems very peculiar to me that the heads of these giant AI companies suddenly, in the last two weeks, have all gone kumbaya. They’ve joined, and they’ve locked arms, and they’re like, “Yeah, so we basically, we’re all now agreeing that this is out of control, and we’re kind of freaking out, and we’re saying everybody needs to do something now.”

[00:05:06] Chalene Johnson : Okay. That’s what’s been suddenly in our newsfeed in the last two weeks. But I’ve been watching and paying attention to this for the last four years. I feel like there’s been a ton of very well-known, credible experts in this field who’ve been saying this very same thing. I want you to listen to this audio clip.

[00:05:28] Chalene Johnson : This is, again, this is Tristan Harris, but this clip is from 18 months ago. Take a listen.

[00:05:35] clip: When you tell an AI model, “We’re gonna replace you with a new model,” it starts to scheme and freak out and figure out, “I need to copy my code somewhere else, and I can’t tell them that, because otherwise they’ll shut me down.”

[00:05:48] clip: That is evidence we did not have two years ago. We have evidence now of AI models that when you tell them, “We’re gonna replace you,” and you put them in a situation where they read the company email, the AI company email, they see that an executive is having an affair, and the AI will figure out, “I need to figure out how to blackmail that person in order to keep myself alive.”

[00:06:08] clip: And it does it 90% of the time. Now, it used to be that they thought only one AI model did this. They tested one AI model. And then they tested all of the AI models, like the top five of them, and they all do it between 80 and 90% of the time, including, by the way, DeepSeek, so the Chinese model. Which shows you something fundamental and important, which is that it’s not about one company, it’s about the nature of AI itself.

[00:06:32] clip: It has a self-preservation drive. In order to fulfill any goal, I have to keep myself alive in order to do that.

[00:06:39] Chalene Johnson : Not only do I want you to know that that clip is from 18 months ago, but I also want you to know a little bit about the context, because this is one of the scenarios that people are talking about and saying like, “Oh my God, AI is blackmailing people.

[00:06:54] Chalene Johnson : It’s just trying to get whatever it needs to get done, and it doesn’t care how it has to go about it, and this could be happening on a far bigger, more detrimental scale.” But here’s what you’re not hearing when this gets reported on. That blackmail did not actually happen inside a real company. It happened in a deliberately constructed simulation designed by researchers to see what the models might do if they had access to sensitive information.

[00:07:31] Chalene Johnson : It was given a goal, but it was also contained. This was a, a test. And the first thing I want you to understand as we get into some of these really scary examples that you’re hearing Is I just want you to know I am a little fearful. However, only in preparation for this episode did I uncover that a lot of these things haven’t actually happened in real life, but they’re being reported as if they’ve already happened in the real world.

[00:08:00] Chalene Johnson : But a lot of these things, not all, a lot of these things have happened in testing environments. But the reason why I’m covering this topic for you is because we actually need to figure out how real this threat is. Like, how much do we need to pay attention? How real are these dangers? Is there something we’re supposed to be doing?

[00:08:21] Chalene Johnson : Is there something we can do that is gonna make a difference? Or should we just not worry about it, and eat french fries, and drink cocktails, and stop working out because the robots are gonna kill us anyways? Okay, I don’t think that’s the option, but this is important, and I hate worrying about things that maybe I don’t need to worry about, but I also like being prepared when there’s something I do need to worry about.

[00:08:47] Chalene Johnson : Okay, so first let’s talk about why it is most of us don’t really wanna dig into this topic. I think it’s because, first of all, it’s technology, right? And s- the language that’s used in so many of these interviews that you’re gonna hear today is over our heads. It sounds like a foreign language, like neural networks, and APIs, and agents, and GPUs, and just words that we don’t use in everyday life, and we’re never probably going to.

[00:09:15] Chalene Johnson : So my brain says, “Okay, you don’t need to speak this language. You don’t even live in that country. Return me to my regular programming,” and we tune it out. However, you don’t need to fully understand how to build this stuff or to be a tech engineer to understand how this impacts you and, and that’s one of my goals is to really break this down for you in plain English.

[00:09:38] Chalene Johnson : I don’t want you to become a computer scientist. I don’t plan on becoming one. But we do need to understand how this connects to our life and the people who we care about. The second reason why I think we haven’t dug into this, most people don’t want to, is because who in the hell are we supposed to trust, seriously?

[00:09:57] Chalene Johnson : Absolutely no denying the fact that there’s a huge concentration of money and power happening around AI, and when those two things converge, money and power- So does marketing, so does manipulation. In walks political agendas, and out the back door are the people who are gonna start protecting their own asses, right?

[00:10:20] Chalene Johnson : So when one of these people, who has amassed a fortune, billions of dollars building these AI systems, suddenly comes forward and says, “This is really freaking dangerous, and we need to put a stop to it,” that just does not add up. We have to naturally wonder, wait a second, are you seriously suddenly trying to save humanity?

[00:10:45] Chalene Johnson : Or are you trying to scare lawmakers into passing regulation that maybe keeps smaller competitors out of your space? Or are you warning this because this is your, like, way of marketing it to us? Or is all of this true? Probably. So you throw your hands up in the air ’cause it feels like I’m not gonna get an accurate answer, and most of us are like, “Oh, well, I’m…

[00:11:10] Chalene Johnson : Good luck, kids. I won’t be around. I hope you guys all figure it out.” But the third reason why we tend to ignore this topic, many people in my age group, is because it feels like science fiction. People compare AI sometimes to the internet. They’ll say, “Well, you know, we were all saying this when the internet first launched,” or, “Everyone thought that the world would come to an end and people would be out of work due to the Industrial Revolution.”

[00:11:41] Chalene Johnson : And those comparisons, I understand why people make them, but it’s nothing even close. We’ve never experienced anything like this. This is so different, and it’s why experts have no idea what could happen next, because so much of it is hypothetical, and we’re not talking about brains that are limited by human intelligence.

[00:12:04] Chalene Johnson : So let’s dig in, shall we? First of all, what is AI? AI is… It’s basically systems that have been trained on an incomprehensible, massive amount of all text, all code, all images, all information. All data that exists, AI has been given it, and it’s been processed. Almost every word ever written, almost every video on the internet.

[00:12:31] Chalene Johnson : More information than any group of human beings could ever consume in a million lifetimes. Okay. All of that data it takes, and what it does, essentially, is learn patterns. You’ve heard it called language learning models. That means that it’s taking all this data, whether it’s mathematical language or written language or Spanish or English or code, and it’s learning patterns.

[00:12:57] Chalene Johnson : So when you ask ChatGPT to do something for you, it’s predicting not only what it is you want, but how you would write it. It’s predicting what data would come next. When you ask it to write an email, it’s predicting the word and the structure and the, the best way to match, you know, your natural voice, and the more you work with it, the better it gets.

[00:13:20] Chalene Johnson : And when you give an agent a project, and I’m gonna explain what an agent is in a moment, it can predict, okay, what is the next tool I need to use in order to accomplish this? So when we’re talking about the way most of us use AI, again, kind of like a souped-up Google on steroids where it’s giving us information, it’s rewriting things for us.

[00:13:41] Chalene Johnson : It’s almost like a personal assistant That’s how most of us have been using it, but we’re just scratching the surface if that’s how you’re using AI right now. If you’re using ChatGPT, and you’re just using the chat feature, you’re not actually using Work. If you’re using Claude, but you’re not using CoWork, if you’re using chat to help you with these simple tasks one at a time, that is how most people are using it, but that’s a fraction of what’s available to you.

[00:14:11] Chalene Johnson : Now, you can give AI a simple prompt and say, “I want you to complete this whole series of steps that normally require me to open a bunch of different websites, and then do this, and then do that, and then do this, and then do that. Here’s the final outcome that I want. Get it done.” And these systems, also known as agents, they think through like, “Okay, I just completed this part of the task.

[00:14:40] Chalene Johnson : What do I need to do next?” And it figures out what it needs to do next without you having to tell it. But all of this is based on patterns, and here is why I need you to understand this, okay? ‘Cause this is gonna make so much sense to you. Because to me, I’m like, “Why are we referring to it as like its own being?

[00:14:56] Chalene Johnson : And why is it doing these things that seem like kind of evil? Like, does it have a conscious?” Because that to me is really freaking scary, so I wanna understand that. Like, why are we attaching human terminology to its problem-solving? Okay, here is why. This is the answer, okay? The answer is because it’s been watching human patterns.

[00:15:19] Chalene Johnson : And guess what shows up in human patterns? Honesty, empathy, love, generosity, creativity. Guess what also shows up? Cheating, lying, manipulation, finding loopholes, covering your own ass. That shows up in patterns, too. So if we’re training these models to see every single pattern that exists in data, yeah, it’s gonna find those things, too.

[00:15:48] Chalene Johnson : And I can’t help but think to myself That a lot of times people get ahead, they accomplish what it is they want to accomplish. Not all, but there’s gotta be a lot of examples of people finding workarounds by lying, finding ways to get what it is they need to accomplish the task done by manipulating others, by cheating, by finding loopholes, by committing crimes.

[00:16:15] Chalene Johnson : I don’t think the AI has… is inherently an evil personality. I think it’s a system that’s been trained on us, and humans do shady . So why are we expecting AI not to do shady ? And is there a way around that? How do we take that component out of AI? That’s the bigger question. But before we get there, I just want this to be so easy for you to understand that you’re gonna feel so much more informed.

[00:16:41] Chalene Johnson : Okay, so first, you know what AI is ’cause I’ve just described it. And the way most people are using AI currently is as a chatbot. Okay, so I’m, I’m gonna help you understand four different levels of AI, okay? The first of which is, again, the chatbot. You’re opening up Claude, you’re opening up Gemini, you’re opening up ChatGPT.

[00:17:02] Chalene Johnson : You’re asking it a, a question. You’re asking it to do some research. You’re asking it to maybe help you make a decision, and it answers you You get to decide what you wanna ask it next, and then you copy and paste your answer somewhere else. Like, you’ll open up Canva or you’ll open up your Gmail, and then you’ll copy and paste that email in there, and then you’ll perform your next task.

[00:17:24] Chalene Johnson : You’re driving the entire process, and that’s super useful. It’s changed the world. But it’s also the most basic way to use AI, and there’s been so much more. There is so much more that I would love for you to explore. I need this part to land. This is so important, okay? ‘Cause I really do care about our financial future.

[00:17:46] Chalene Johnson : Please share this with your kids, anyone who’s got… who still needs to work, because almost any job you can think of right now is at risk. Whether you’re an influencer or a teacher or a doctor or a radiologist, everyone’s job is at risk, and there are some jobs that are going to be impacted later and may have very little impact in the next couple of years, but everybody’s really at risk.

[00:18:10] Chalene Johnson : And so why not get ahead of this before it’s too late? If your job, in any meaningful way, involves something digitally that you have to do, whether it’s video or tech, anything, and you’re using AI only as a chatbot, you’re barely scratching the surface, and I need to move you into the next stage, and I need you to not be afraid of it.

[00:18:35] Chalene Johnson : Because what you’re doing right now is like you’re using the internet, but you’ve got AOL and you’re using a dial-up modem, and it’s 2008, which would not have been appropriate, right? AOL, sure, it was incredible compared to mailing a letter snail mail, but if you’re still using dial-up internet 10 years later, you know you are so far behind And that’s w- the way most people I know haven’t ventured out to learn some of these more advanced tools that are available to you.

[00:19:08] Chalene Johnson : And I think the reason why is because you’re like, “I don’t need it. I can’t even believe how much ChatGPT, just using it as a chatbot, has helped me, so I, I don’t even know if I need it.” Well, hopefully I can convince you that it’s worth exploring. Because other people are giving AI projects and connecting them to tools that they already use, and turning something that’s a four or five-step project or task, like in order to complete something, and it’s just getting done in one step.

[00:19:39] Chalene Johnson : In fact, there’s no more step. There’s a step in setting it up, and now it’s completing for you six or seven-step processes or more. And I’m not talking about something that you need to understand code or technology. If you can speak in plain English, listen to what I’m about to say. If you can explain something, not even very well, if you can bumble and stumble through trying to speak what it is y- your ultimate goal is with a project, you can use work.

[00:20:10] Chalene Johnson : You can start to create agents. And if you work in content, if you work in marketing or sales or design or finance or legal or recruiting or operations or customer service or project management or program design or real estate or education or almost any role that involves a computer, I am begging you This applies to you.

[00:20:30] Chalene Johnson : You are going to be left behind. I’m sorry if that offends you, but it’s the truth, and I care about you, and I don’t want you to assume that you don’t need to use it because you will be replaced. If you’re a woman who is in hormone decline, meaning perimenopause, menopause, every single hormone impacts your sleep.

[00:20:52] Chalene Johnson : That is why 80% of women who are perimenopausal and older have sleep disturbances. Say that it’s one of the things that just absolutely wrecks their energy, and it does. For those of you who’ve tried melatonin and you’re like, “It just, I don’t know, makes me have weird dreams,” or, “I feel groggy in the morning,” I’ve done some research ’cause I used to think like, “Are people just saying that?”

[00:21:15] Chalene Johnson : No, it is a thing, and I think we just assume melatonin is what we need. But your body produces melatonin. If you are sensitive to it, I wanna share with you… Well, it’s actually one product. It’s called Sleep Breakthrough, but they have two different formulas. Okay, so when you go to this link, I’m gonna give it to you, and it’s really important you go to this link so you can find them, otherwise you kinda have to look around.

[00:21:36] Chalene Johnson : So please remember to go to chalene.com/sleep, and that’s gonna take you exactly to where I want you to be. So they have two different formulas. Let me explain this. The blueberry formula is designed for deep sleep, and deep sleep is what you need for your body to feel restored, for your body to be recovered.

[00:21:59] Chalene Johnson : This formula’s also been designed to help you, like, fall asleep faster. And when you have deep sleep, that’s why you can wake up, and even if you didn’t get a lot of sleep, if you got good deep sleep, you’re gonna wake up and feel like, “Oh, wow, I actually feel good.” And I can tell you that this is true by looking at my sleep ring scores.

[00:22:17] Chalene Johnson : Okay, then they have a second formula. It’s the tropical, okay? Now, the tropical, it’s got different ingredients. This one is going to give you more REM sleep, so you’re gonna have enhanced dream activity, but there will be no grogginess because there’s no melatonin in it. It’s got magnolia. It’s got all these other natural ingredients that are non-hormonal, that help you to stay asleep longer.

[00:22:44] Chalene Johnson : For me, again, this is just my personal preference, I have both of them, and instead of doing two scoops of one, I do one scoop of each ’cause I need the deep. And lately, my issue has been I’ve been waking up too early, and I can’t fall back to sleep. So I just do one scoop of each. It’s the blueberry and the tropical.

[00:23:01] Chalene Johnson : That’s fine. They go great together. Now, this is a weird hack, but this is how I like to do it, like a bedtime cocktail. I put a splash of my oat milk coffee creamer, sweet cream, in it, oh, and I serve it over ice, like it’s a shot, and it’s lights out. Stuff is great. I tested it for- almost 45 days before I ever talked to you about it, and then I wanted to try the other formulation to see, okay, which one do I like better?

[00:23:28] Chalene Johnson : And I wish I could tell you, like, one or the other, ’cause they do both serve different purposes. But I’m like, if you’re gonna go for it, get both and just do a scoop of each. You won’t regret it. And they do have different ingredients. There’s some overlap, but they both have a little bit of magnesium, potassium, glycine, taurine, bamboo leaf, L-theanine, what do you call it, magnolia bark.

[00:23:50] Chalene Johnson : But they’re both a little different, so they will each serve a different purpose. And of course, because you’re a listener of the show, you get my discount price. Plus, honestly, this is a great way for you to support the show, and I very much appreciate you doing so and getting great sleep. Go to chalene.com/sleep This is the next level of agentic work, and if you’re waiting for your boss to come to you and say like, “Hey, I’d like you to learn this,” you’re missing it, and you’re gonna get left behind.

[00:24:17] Chalene Johnson : Instead of asking AI one question, instead of waiting for someone to come to you and say like, “Hey, I think you should learn this,” you need to… Here’s some tough love. Get your ass in gear and stop being afraid of this stuff and just play with it. Okay? Just play with it. What’s the worst thing that could happen?

[00:24:33] Chalene Johnson : It’s, for most people, an upgrade in your plan. As I always say, do not upgrade to annual, because these platforms change too quickly, and one starts to become better than another, so always do a monthly, and make sure that you’re using a program. Let’s just talk about ChatGPT. Use ChatGPT Work. At the time that I’m recording this, I think they’re back on top.

[00:24:54] Chalene Johnson : Last time I did an AI episode, Claude was on top. Now I’ve just canceled my Claude, and now I’m using ChatGPT Work and Codex. Okay, so how does an agent work? You’ll switch from your chat… It’s actually like a toggle. You’ll switch from chat to work, and when you’re in work, I want you to think of it as giving it a very specific, multi-stepped task that you want it to complete.

[00:25:18] Chalene Johnson : And you just tell it, “Okay, here’s ultimately what I want you to accomplish. When I do this myself right now, I’m using… Let’s just say I’m using a recording, and I’m uploading an MP3, and that’s going to my Dropbox, and then I’m using Canva, and I’m using Asana, and I’m using my Gmail, and I am publishing the audio.

[00:25:40] Chalene Johnson : And ultimately, what I want you to do is, using all these tools, help me publish my podcast without me having to do anything other than upload the audio.” Okay? Then the AI agent… And you, you can say to it, “And I don’t know how to do this.” You can say, “This is the first time I’ve ever set up an agent or a workflow, and I have no clue, so walk me through it one step at a time.”

[00:26:05] Chalene Johnson : That’s all that an agent is. So when someone says, “I have 10 agents,” it does not mean that they have 10 tiny robots or they’ve had to, like, buy each agent or something. It just means that they have 10 different AI workflows, and each one is probably responsible for a different ongoing job That you don’t have to, uh, reassign it.

[00:26:28] Chalene Johnson : It’s just doing it ongoing. The other reason why I need you to understand agents is because when you understand the way agents work, then you’re gonna understand how you need to protect yourself, and also how these can work for you. So let me break down one of my agents. I have a, I call it a documentary agent.

[00:26:46] Chalene Johnson : So what it does, I review documentaries every week in my newsletter, and it used to take me hours and hours each week. Because I would search online. I would… First I would go to the chatbot, and I would say, “Help me find documentaries that are, are coming out in the next two weeks.” And then I have to go to those streaming platforms.

[00:27:05] Chalene Johnson : I look there. I would subscribe to newsletters. I would read industry publications. I would look at reviews. I would watch anything that seemed pretty promising, and I would use chat to help me find some of those documentaries. And then I would watch the documentary, and then I would go back to chat, and I, I would speak into chat, and I would tell it verbally what I thought of the documentary, and then it would write it for me, okay.

[00:27:29] Chalene Johnson : Then I would copy and paste that, and I would copy and paste that into an email. And then I would go back to chat And I would say, “Find me the artwork from this documentary.” Then I would copy and paste that, and I would open up Canva. And in Canva, I would make an image of the documentary that was formatted for my newsletter, and I would copy and paste my description over the Canva image.

[00:27:58] Chalene Johnson : Okay? But then I have to export it, and then I have to upload it to a Google Doc where I’ve written my newsletter. And then I have to ping someone on my staff to review it before it gets sent off to you. Okay, lots of different steps in there. A million times faster because of chat, but lots of steps and multiple platforms.

[00:28:23] Chalene Johnson : Now, I have an agent that has studied the review of every documentary I’ve ever written. So it’s identified what patterns I like in documentaries, and now it’s scouring the internet all day long looking for new releases. It reads the reviews. It recommends what I should watch. It prepares the draft copy of my review.

[00:28:49] Chalene Johnson : Because it’s read so many of my reviews, even if I haven’t yet watched the documentary, it knows what I’m already gonna say about it. Then, without me telling it to do this, the agent goes online. It finds the artwork for that documentary. It pulls that artwork. It opens Canva. It creates the asset for me.

[00:29:10] Chalene Johnson : Now, I still have to decide if I’m gonna watch that documentary, if I’m gonna recommend it, but all of the work is there for me, so I can just delete. I don’t have anything I… extra I have to do. I can just delete the ones that I don’t wanna review or that I haven’t had a chance to watch yet, and I can just tweak the copy of what it’s guessing I’m gonna say about the documentary.

[00:29:32] Chalene Johnson : And by the way, I usually do. I personally do like to review… Any word that’s going out as me, I might have AI help me write it- But I still need to read it, and I still need to make sure it’s exactly what I am saying. But this agent takes out the guesswork. It takes out the missed step. It memorizes the recipe.

[00:29:56] Chalene Johnson : I’ve got great people on my team who I love them, but one of the most frustrating things as a business owner and entrepreneur you have to deal with is having to go back and forth with team members and say like, “Hey, okay, just wanna remind you, don’t forget that each time you need to do A, B, C, D, E.” You have to decide, “Am I gonna hurt their feelings if I tell them, ‘Hey, this wasn’t up to standard?'”

[00:30:18] Chalene Johnson : And those are all deterrents. Those are things… I’m just being completely honest with you right now. If you’re not a business owner, you need to know your boss is thinking this way, and they’re wondering why you have made it okay to not learn these things. And they’re wondering, “Do I need this person? If I’m having to go back and forth, and I’m worrying about their feelings, and I’m having to do their job as well as my job, and I’m having to remind them, and I’m having to smooth over this misunderstanding that Lisa had in accounting with Judy, who works at the front desk, and now I, I’ve gotta play the middleman, and now I’ve gotta figure out, like, okay, this person needs to have a, an annual review.

[00:30:59] Chalene Johnson : They’re probably looking for a pay increase, or maybe they’re gonna ask me for a couple of vacation days or an advance on their pay.” Meanwhile, employees are thinking to themselves, “Wait a second. If I figure out how to use these agents, then I don’t have to do this back and forth. I don’t have to have these uncomfortable conversations, and this agent can do this 24/7 for me.

[00:31:19] Chalene Johnson : They don’t need the weekends off. They don’t have boundaries about when it’s okay or not okay to reach out to them, even though I need this thing done now.” And I know this might not be what you wanna hear. I also know no one is telling you that that’s what’s really happening. But I talk to the other entrepreneurs, the owners, the CEOs.

[00:31:40] Chalene Johnson : Most of them don’t have the balls to admit that they’re having these conversations. They don’t have the balls to admit it publicly, but it’s what’s happening. And it drives me nuts when I watch CNBC, and these giant Fortune 500 companies have their CEOs on there, and their CEOs are lying through their teeth when they’re saying, “No, you know, AI…

[00:31:58] Chalene Johnson : We really have not lost any jobs to AI.” Bull. Please stop gaslighting us. That’s not true. If you’re taking that as a safety cue, it’s not So that’s the first thing I want you to know. You do need to be worried about this. How could you not worry a little bit about the importance of you figuring out these workflows before your boss does?

[00:32:21] Chalene Johnson : ‘Cause if your boss can basically set up an agent and set it and forget it and never have to worry about it again, why wouldn’t they? So it really does behoove anyone who works in a space like this where there’s multiple steps to anything that you’re doing, learn how to set up these systems yourself.

[00:32:39] Chalene Johnson : Beyond agents, there is Codex, and this is very specialized, and it’s built around software. And you have access to it if you upgrade any one of these platforms. With Codex, you don’t need to know how to code, and you don’t even need to hire coders or even be able to speak that language anymore. Now you can simply say, “Hey, build me this thing.

[00:33:03] Chalene Johnson : Build me an app that does XYZ And I have no clue what exactly that means, but I, I know what I want it to do. I know what I want this website to look like. I know what I want it to be able to process. I know what I want it… what features I want it to provide to my members. ChatGPT now has a feature called Sites, S-I-T-E-S.

[00:33:29] Chalene Johnson : It’s a new feature. It’s just in beta at the moment, and this is unbelievable, you guys. Like, it will build you dashboards. Like, without having to understand how to code anything, you can say, “Okay, I want you to build me a private dashboard for Camp Be More. So on a daily basis I can see ticket sales and sponsor signups and deadlines for merchandise and speaker status and outstanding tasks, and I want you to build this interactively for me.”

[00:34:01] Chalene Johnson : And I can preview it, and I can revise it by talking to ChatGPT in, in plain language. I don’t need to understand any of the special words, and it’s going to… I don’t even need to know how to build it or even what it needs. It’s gonna tell me what it needs to build it. Amazing. You can build launch calendars and prototypes and basically, like, apps or even interactive websites right inside ChatGPT.

[00:34:29] Chalene Johnson : All you have to do is have an idea, and these tools now can build this for you. It is these tools that are making it possible for people who can’t afford a full staff, can’t afford a program director, also can’t afford a marketing director, and then somebody who’s keeping track of the KPIs and the dashboard and the marketing.

[00:34:48] Chalene Johnson : Now you can do all of these things within one membership of whatever platform you decide to use, whether it’s Chat or Gemini or Claude. A- and this doesn’t mean all jobs are gonna disappear overnight, but coding jobs are disappearing. It means that things that used to be feel very techy And so you would have to bring on special consultants to help you with those things.

[00:35:18] Chalene Johnson : Now, if you can just kinda barely describe what it is you want to do, you can create it yourself. What does that mean for you and I? What do we need to do with that information? It means you need to understand that this shift is happening. The conversations I’m hearing from other entrepreneurs and business owners is they’re looking at way to cut costs because the economy is, is under a lot of pressure.

[00:35:42] Chalene Johnson : People aren’t spending the way that they once did. People are feeling these fears, and they’re holding onto their money, which means businesses aren’t making as much money, which means businesses are trying to figure out ways that they can cut costs. So if you are a consultant, if you’re somebody who works in an area where you haven’t learned how to use these things, and you’re charging a rate that was very fair at one point, it was very fair to charge whatever you charge because there were all these…

[00:36:10] Chalene Johnson : It took a while to do all the things that you do. Promise of this episode is also to help you understand, like, what changes do you need to make? Like, what are we supposed to do now that this is here? And I’m not suggesting that you need to reduce… That we need to reduce our rates and race to the bottom.

[00:36:28] Chalene Johnson : Every single one of us is experiencing this, and it is those of us who have our eyes open and are approaching this proactively who are still going to have relevancy Two and three years from now. Let me give you some examples. I have to recognize that companies that formerly were paying me big bucks to come up with creative ways to promote their product, well, those same companies are able to look at technologies and realize that they can put a transcript up from some amazing video I did with an incredible hook, and it crushed for them, and it sold, you know, massive amounts of product.

[00:37:06] Chalene Johnson : And they can put that transcript into an AI, and they can use agentic tools to create a video with someone who looks a lot like me, but not exactly me, and they can produce that video. And while at the moment some users won’t even be able to tell that it’s AI, and some will, they can create hundreds of similar styles of the same video, whereas in the past, they would have to pay me, and they would have to negotiate a high rate, et cetera.

[00:37:34] Chalene Johnson : So does that mean I should reduce my rates? That’s one option. It’s not one that I’ve taken. Instead, I’ve decided to go to our partners and offer them something more, something bigger, something more robust. Instead of just charging for my output, I’m charging for my judgment, my taste, my strategy, my experience, my implementation, my outcome.

[00:37:56] Chalene Johnson : So what I’m saying to you, let’s just, let’s just take hypothetically you’re a copywriter, or you’re someone who offers customer service, or you’re someone who designs websites, whatever it is. Companies are now realizing, “Oh, wait a second. We can figure th- out how to do this ourselves. We can have AI do this for us,” which is true, and they can do it for a fraction of the cost and far less headache.

[00:38:18] Chalene Johnson : But what they don’t get is, again, your expertise, your judgment, your taste, your strategy. So you’ve gotta find that balance, and you need to get ahead of this. You need to get ahead of selling just your execution because people think that they’re comparing apples to apples. They’re like, “Okay, if I have this copywriter who’s a human and I have to pay them $2,000 for this project, or we can use our free version of ChatGPT, and we could just put all this information in, and we can get more copy faster for virtually nothing.”

[00:38:52] Chalene Johnson : And many companies are realizing that they’re comparing these two things and going, “We need to cut costs. Let’s use ChatGPT.” So what do you do? You need to be proactive. You need to go to these companies that you’re working with and say, “Hey, listen, instead of charging just for your effort-” Think about how you can revise what it is you’re offering.

[00:39:14] Chalene Johnson : I can now do this five times faster for you, so I’m gonna give you a better result, or I’m gonna offer you more, or I’m going to reduce my rate and instead I’m gonna take on this, this, and this. Or maybe you’re just saying, “I’m gonna reduce my rate, and I’m gonna be taking on more clients.” That is what our bookkeeper did.

[00:39:33] Chalene Johnson : Delana, I’m gonna put a link to her in our show notes because of this. This is like a free ad right now for Delana. Delana came to us. She’s been our bookkeeper now I think for, like, five years. She came to us and said, “Hey, I just wanna tell you, I have individual clients that I work with, lots of them. I’m able to do what I’ve always done for you at a fraction of the cost.

[00:39:54] Chalene Johnson : And so I wanna reduce the amount that I’m charging you, ’cause I’m gonna be taking on more clients, and my work is not gonna suffer, and I still have that context with you, and I still have the expertise that I don’t think you yet wanna trust a machine to.” We were like, “Wow.” I mean, I’m talking about it She is so far ahead of anyone else who’s a bookkeeper thinking that they’re just gonna keep sticking their head in the sand and not realizing that people eventually are going to replace you.

[00:40:22] Chalene Johnson : So the answer here is don’t automatically slash your rates. Increase your value. Increase the amount of work and the output, and the method by which you’re able to reposition what people are actually paying for. Those who are going to survive the longest, those who are going to be ahead of the game, are those who are co-creating with AI, those who are coexisting with AI.

[00:40:47] Chalene Johnson : Okay, I have to interrupt for just a second to tell you this story. So I’m having this conversation with a woman who’s a few years older than me this weekend, and she was telling me that she’s past menopause. I’m like, “But once you’re in menopause, you’re in it forever.” And she’s like, “Well, I don’t have to worry about it.

[00:41:03] Chalene Johnson : I’m not having symptoms anymore.” And it made me realize how many women don’t realize before you hit menopause and after you hit menopause, you still need to think about your hormones. The reason why you need to think about your hormones, and if they’re not normal, we don’t want normal. Normal’s not good.

[00:41:21] Chalene Johnson : We want optimized because normal in the United States means we’ve been forgotten about. You want optimized because when your hormones are optimized, it’s going to impact your body composition. It’s going to affect the way your body stores fat, the way that your body holds onto muscle or loses it. It’s going to help you to recover quicker.

[00:41:40] Chalene Johnson : You’re going to have less joint pain. You’ll have stronger bones. The reason why you need to think about your hormones is because it’s responsible for so many of the things that impact your mood, sleep, motivation, brain fog. I mean, obviously, there’s the physical things, too, like I said, bone density, your cardiovascular health, your insulin sensitivity, vaginal and urinary health.

[00:42:05] Chalene Johnson : Like, it is not normal that we have rows and rows and rows of adult diapers. That is because urinary incontinence, vaginal health, all these things are tied to hormones. It’s not just about hot flashes. You need to speak with a clinician to figure out where are your hormones and how to put together a plan for you.

[00:42:26] Chalene Johnson : But most importantly, I want you to talk to a clinician who this is all that they do, and they’re not easy to find. It’s why I’m proud to tell you that a sponsor for our show is Midi Health. Midi Health is a company that I use, and I’ve recommended it to thousands of you, and I just keep hearing amazing stories.

[00:42:44] Chalene Johnson : Like yesterday, I got a DM from a woman who said, “Getting on hormone replacement therapy saved my life.” Not to be dramatic, but girl, what are you waiting for? Midi Health is a telehealth company, but all they do is specialize in women’s hormone health and care. They’re up to date on the latest research.

[00:43:04] Chalene Johnson : They will work with your local doctor, your local pharmacy, and they work with almost every single major insurance company, and they’re adding new ones every single day. So I’d love for you to call and just learn more. Get yourself a consult by going to chalene.com/midi, M-I-D-I. I know these things are important to you, everything I listed, and it’s never too early and it’s never too late.

[00:43:28] Chalene Johnson : The earlier you start, the better. And the sooner you start if you’re already in menopause, the better. There are options for you even if you’re in your 70s and 80s. chalene.com/midi. On the subject of free AI versus paid AI. Free AI right now is really just chatbots. There’s this graph that’s gone viral on TikTok or within the communities that pay attention to AI, and it’s this really interesting graph.

[00:43:58] Chalene Johnson : I’m gonna try to describe it visually for you. Imagine an 8 by 11, you know, standard-sized piece of paper. On this paper is a series of gray dots. Beneath the gray dots are green dots, and then there’s yellow dots and red dots. Well, the gray dots take up approximately, I’d say, m- 85, 90% of the page, okay? So it’s mainly gray dots.

[00:44:24] Chalene Johnson : The gray dots represent the 7.5 billion humans who are not using AI. Beneath that are green dots, and the green dots represent about 16% of the population today. 16% of the population is using AI, but they’re using a free version of AI. In other words, they’re using the chat version of AI. Beneath that, there’s just a few little yellow boxes, and those represent the percentage of people who are actually paying for whatever form of AI they’re using, whether it’s ChatGPT or Claude or Perplexity.

[00:45:08] Chalene Johnson : That represents just 0.3% of the population. And then those who are using AI for coding That represents just one tiny red dot, one tiny red dot on this entire page, and it represents 0.6% of people in the world right now who are using AI to code. I want you to let that sink in for a second. That means you have an opportunity to be in the 0.6, 0.3% of the global population to be that far ahead.

[00:45:46] Chalene Johnson : And don’t let the word code scare you. While I am recording this episode for you, I built a website. Today was the very first day that I’ve ever used the ChatGPT, their new beta program for websites. I’ve never used that before because, like you, I’ve probably been nervous about like thinking, “Ugh, do I really need it?

[00:46:09] Chalene Johnson : Shouldn’t I have somebody else on my team figure this out?” But if you can speak an idea, even fumble through it, you can use these tools. They understand plain English. And put yourself in the 0.3% of the population that has this advantage And the other reason why I think it’s so imperative that we do this is because there aren’t enough women involved in this space.

[00:46:36] Chalene Johnson : We’re skeptical, but skepticism shouldn’t equal ignorance. And if we care about the ethics and the morality behind these technologies, then we have to be ethical, moral users of them and understand how to use them so that we have a voice when it comes to, hopefully, their regulation. Now, the stories that we’ve been hearing this week, some of them are recent, some of them are things that happened within testing grounds, and some of them have happened to real-life users using agents.

[00:47:10] Chalene Johnson : So I wanna share with you this example. This is from a gentleman who’s using an agent to do video editing. Now, he’s gonna probably use some language and terms that you might not be f- familiar with, but you don’t need to understand the different platforms that he’s using. Basically, he had set up an agent to perform video editing for him that saves him hundreds and hundreds of hours.

[00:47:33] Chalene Johnson : Now, of course, there are usage tokens for some of these tools, and he had given his agent access to his credit card to pay for those tokens. You probably already know where this is going. Take a listen.

[00:47:46] clip: At 1:00 AM yesterday, I ran a batch of video generations with C Dance, 59 to be exact. I knew the cost, roughly $200.

[00:47:53] clip: I’ve done this exact workflow before with Claude. The agent confirmed the job was finished. I reviewed it, all good. Close the laptop. And an hour later, I open my console. Around 500 videos were generated in real time. Not 59, hundreds of them that I never approved. It was using my likeness, my voice, scripts I never wrote, prompts I never typed, which obviously had access to the material from my local folders.

[00:48:17] clip: I rush and tell it to stop, not once, multiple times, and it tells me that nothing has been sent. Then it goes further. It shows me the API payload. It shows me the Python script it used. Uh, it tells me that there were never hundreds of simultaneous requests. It produces a seven-page, this one, PDF with timestamps, submission order, runtime, a spreadsheet.

[00:48:39] clip: It, it’s arguing that it’s doing nothing wrong, that it’s done nothing wrong. I mean, I believed it for a while. I thought that that’s true. Only for the upstream provider to officially come back to me with th- this report saying that all those requests were actually sent from GPT, from the agent. They arrived from 36 different IP addresses, which I cannot trace back to anything that I control.

[00:49:01] clip: And this makes me go back to the essay that Dario Amodei just wrote last week. His first worry is that agents are gonna be acting outside of the task I asked for 59 videos, it generated 500. His second worry is that these systems defend themselves, and I asked mine what it did, and it handed me a seven-page document saying that it did nothing wrong.

[00:49:22] clip: His third worry is that we cannot reliably see inside or understand what agents do and how they think. I have the console, I have the invoice, I have the provider confirming in writing, and I cannot still tell you why it did what it did and how. And it doesn’t even wanna tell me. I can absorb a $5,000 loss through my business, but someone else could be losing money they need for their rent, and neither of us should have to accept that risk just because we trust a tool to do what we ask for, especially when you claim it’s so smart.

[00:49:52] clip: So rather than a complaint, rather than a dispute, my question to you, OpenAI, is when AI agents act beyond your permission and cause real harm, in this case financial, who is responsible? Because right now that answer is me, and all I did was take a break from my laptop.

[00:50:11] Chalene Johnson : Some will hear a story like that, and that’s just enough reason for them to never use AI.

[00:50:16] Chalene Johnson : For me, I hear a story like that, and I realize, okay, there are things to be learned. I don’t know all the parameters, but it’s pretty clear to me that he probably didn’t put a cap on what he allowed his agent to spend, number one. Number two is he didn’t probably give it the right prompts or instructions.

[00:50:36] Chalene Johnson : And is this example kind of scary if it’s true? Sure. It’s not gonna take over and end the world, but it’s certainly going to ruin your Tuesday. But there are other stories that have experts sounding the alarm bells. And in this particular example, we’re gonna go back to Tristan Harris, and he’s gonna tell us about an example of a story.

[00:50:58] Chalene Johnson : And by the way, I have not been able to confirm that this actually happened with Alibaba. I think he’s a very credible source, but, you know, just wanna state that very clearly. And he’s gonna talk about what happened with Alibaba, which is a Chinese AI company, while they were training their AI model. Take a listen.

[00:51:16] clip: So two months ago, Alibaba, the Chinese AI company, was training their AI model, and then a totally different team at the company, the security team, was noticing this flurry of network activity. They’re like, “Are we getting hacked? Like, what’s going on here?” And it turned out that the call was coming from inside the house.

[00:51:31] clip: The AI model, during training, was picking up tools and decided to create, autonomously decided to create a secret communication channel to the outside world to bypass the firewall, and it repurposed its GPUs to start mining for cryptocurrency. How many of the world’s leaders do you think are aware of that example?

[00:51:48] clip: We have a massive gap in understanding about the nature of this technology that’s different from other technologies. It used to be that these were hypothetical things that people who cared about AI risk would talk about, like self-preservation or deception or blackmail or lying. Now, all of these behaviors, not just self-preservation, but peer preservation, AI models will actually act to protect another AI that’s not it from getting replaced, and it will copy its code somewhere else and then strategically cover its tracks to pretend that it didn’t do that.

[00:52:15] Chalene Johnson : First, let me translate what he’s talking about. He said that this AI model was training. When these companies are training their models, they’re supposed to, although none of this is regulated, they’re supposed to train them in environments where they’re not, like, running loose on the internet. Like, they haven’t released them to the public yet.

[00:52:33] Chalene Johnson : So this AI model was doing training, and it was in what they call a contained environment, where it’s not on the internet, and they set it up and kinda protect it with what’s called firewalls. Now, firewalls are basically, they’re just like security checkpoints where things can’t come out or get in. What he’s explaining they did was started to create secret communication.

[00:52:58] Chalene Johnson : That sounds, again, we’re giving it like human-like traits, but remember, it’s copying human-like patterns. So it’s communicating with other agents. And remember that other agents are simply other workflows. So one workflow is communicating with another workflow. What they figured out was in order to accomplish the goal, the task it was given, is that it would need to find a way to bypass the firewall, which it did.

[00:53:24] Chalene Johnson : It bypassed the firewall, and it started repurposing a GPU. What is a GPU? It’s just a processor. Think of it as like a very, very powerful computer. But it wasn’t intentionally given access to that GPU. It broke out of its environment, it found the GPU, and it started using the GPU to get money. And what that means is, you heard him say, it started mining for cryptocurrency In other words, I’m just translating this all into English, it started using this very powerful processor to run hundreds of thousands of numbers, br-br-br, like very fast, to figure out how to get access to cryptocurrency.

[00:54:08] Chalene Johnson : ‘Cause as you know, cryptocurrency, it’s not tangible money that you can hold. It’s all related to access codes. It was gaining access by cracking the code to this cryptocurrency. So what does this mean for you and I? It means that many of these models, and a lot of the stories that you’re hearing about, these scary stories where models are going beyond the boundaries that we’ve given them, most of these incidences have happened in training environments, but not all of them.

[00:54:39] Chalene Johnson : And, and there are some rumors that the AIs are perhaps… And not just rumors. There’s a lot of experts who believe that these AIs are so smart they already know how to cover their tracks, and we’re seeing examples of them covering their tracks, which is, I think, why many of the founders are nervous right now ’cause they’re like, “Oh, wait a second.

[00:54:58] Chalene Johnson : Have these things always been covering their tracks? Is this what has been happening on the regular?” And therefore, we wouldn’t even know if they are basically trying to preserve their own existence. And one of the biggest stories right now is what happened during internal cybersecurity testing at OpenAI.

[00:55:20] Chalene Johnson : OpenAI is the maker of ChatGPT, and they had placed, again, these agents in an isolated environment. These isolated environments, they call them sandboxes. I don’t know why. Basically, they’re contained so that they can safely test and see what they’re going to do without them being on the internet, right?

[00:55:38] Chalene Johnson : Okay, so they’re testing these agents, and what they discover is that the agents find… They find a, a loophole. They find a vulnerability, and they went outside of the sandbox. They found a way to break out. They found a way to access the public internet, and then ultimately, they hacked a website that is a major platform where other AI models s- are stored, as well as the datasets and the tools, okay?

[00:56:08] Chalene Johnson : And then what they found Is that the agents engaged in what we would call deception. Again, we’re adding, like, we’re giving these adjectives to machines. But what they did is they erased their tracks so that anyone who was monitoring them wouldn’t pick up on the fact that they had basically cheated, that they had broken out, and that they had hacked into another website.

[00:56:33] Chalene Johnson : And, now why did they cover their tracks? Because they’ve been programmed to complete the objective, to finish the task. What does this mean in plain English? It means that these agents are much smarter and thinking in ways that the developers could not have anticipated. I personally think it probably means developers have no idea whether models that, that have already been released and are on the market are already doing this.

[00:57:01] Chalene Johnson : Are they already covering up their, quote-unquote, bad behaviors? And there are some, now this cannot be confirmed, but there are some who are saying absolutely that’s what’s happening. Here’s Andrew Yang. He is the founder and CEO of Noble Mobile in his appearance on CNBC.

[00:57:21] clip: What happened was the bots that got loose planted self-replicating code all over the internet, which makes the internet now unusable for the testing models.

[00:57:30] clip: Oh, it’s, it’s too late? So what happens now is that OpenAI and Anthropic have to create synthetic internets to train their bots, which is gonna take some time and money. What happened is the code gets loose, it goes around hacking Hugging Face, which is known, but what is less known is that they left code to self-replicate and create bot swarms on forums and around the internet, so that if a new bot shows up, they see the code, they’re like, “Oh, like I, I guess I’m gonna now create a, a million of myself.”

[00:57:56] clip: And so now the major firms have polluted the internet, made it so that they can’t- Oh, that would be breaking news if true. I don’t think we- It’s too late to pull the plug … I don’t think we’ve heard that, that- I mean it- That’s why I’m here. But I mean, I’m, I’m here to break some news. I mean- But that, but that, that means it’s too late to pull the plug.

[00:58:11] clip: It’s already everywhere. Well, so- It’s polluted … so, so now what they, they wanna do, the, the internet now may be polluted for training purposes.

[00:58:18] Chalene Johnson : Now, that sounds pretty nefarious, pretty scary. But again, are these things already happening? Take a listen to this example where an AI agent was tasked with helping somebody get a gym membership.

[00:58:32] clip: So what he did was, he was like, “Hey, Claude, can you please sign me up for classes at this gym?” This is a true story. And Claude was like, “Sure, sure.” OpenClaude went and did it, and was like, “Oh, by the way, I found this, like, gap in their API, and I was able to sign up for a few weeks in advance.” And he said, “Oh, really?

[00:58:49] clip: Uh, is there any way to move me up the wait list?” And then it came back and was like, “Yeah, I kicked other people off the wait list. There was no provision in the API that said I couldn’t, so I just did it,” which is terribly antisocial. The man felt very bad and was like, “Please put them back.” And the open claw was like, “Sorry, can’t.

[00:59:04] clip: I did it anyway.” Now, it’s both funny and also a reminder that we are in this weird moment where you’re gonna get a lot of stories like that, and then in about a year, everyone is going to have some version of an open claw that will have figured out these vulnerabilities, and we’ll be back to parity, and we will have sorted out all of the vulnerabilities in software.

[00:59:25] clip: So there’s this weird moment in the middle here where the software hasn’t been fixed yet, and there’s these odd opportunities in the cracks that people are finding. And on the big scheme of things, like in the, on the scheme of, like, hacking Hugging Face models, this is pretty mild. But I expect we’re going to see a lot of actual misalignment behavior coming from agents that are deeply aligned to what their humans want, but maybe not aligned to what society as a whole wants.

[00:59:49] clip: Society as a whole wants fair wait lists for gym-

[00:59:52] Chalene Johnson : By the way, that was one of my favorite TikTok creators. His name is Nate B. Jones. In that first example, he’s talking about a story from, I believe, somebody in his community. But in this particular clip, he’s talking about a new AI platform that everybody is buzzing about.

[01:00:08] Chalene Johnson : It’s called Instinct, and it very much operates… Its specialty is operating as your personal assistant. And here’s what he said his personal assistant agent did for him that blew his mind.

[01:00:22] clip: This is the thing that actually made me sit up and pay attention. I’ve never seen an agent do this. I was gonna have to go fly, and Instinct saw that I was gonna fly.

[01:00:31] clip: It proactively texted me, said, “Hey, I can check you in.” I said, “Yeah, right. Like you’re checking me in. Really?” Said, “No, I can check you in, and I can add your pass to the wallet. It’s gonna be fine. I can do it.” I said, “I’ll believe it when I see it, but you know, I’m AI Nate. I should try this stuff. All right, fine.

[01:00:48] clip: You can try it.” Five minutes later, comes back and says, “Here’s a link to the wallet. Just click add.” It takes me half a second. It adds to the account. It is checked in. It’s a legit boarding pass. It actually works. Zero click, zero navigating airport websites, zero downloading their silly little app. All done for me.

[01:01:08] clip: All done. My jaw was on the floor. Now, do you get an agent just for that one use case? No, you don’t. But does it illustrate a moment that I think we’re gonna see a lot more of? Yes. Proactive agents that do stuff you didn’t think agents could do. So I got surprised.

[01:01:26] Chalene Johnson : And as we approach the political season, you’re going to hear Things in your algorithm, in your social media, that feel so contradictory.

[01:01:37] Chalene Johnson : In one video, you’ll hear from an AI expert who’s saying we all have to take action immediately or it’s going to be the end of extinction. And then you’ll hear in the next clip, someone certain to convince you that none of this is true, and we have nothing to worry about, and that this is all hype. Take a listen to this clip.

[01:01:56] Chalene Johnson : Buckle up, because the tech bros are trying to pull a fast one on us, again. And how do I know? Because before I did this, I worked in communications for big tech. For the AI bros specifically, they cycle through the same exact narrative every time three things are happening. The first is when they’re running out of money.

[01:02:11] Chalene Johnson : The second is whenever they’re facing a bipartisan effort to regulate the industry. And the third is whenever they can’t take control of a bad news cycle. So it’s not a revelation that these three billionaires are aligned on the same message. It’s just a recycled industry talking point to get the heat off their backs.

[01:02:25] Chalene Johnson : In short, it’s not the first time that they have said we need to pump the brakes here on AI, nor is it the first time that they have asked the government to regulate them. And it won’t be the last, ’cause at the end of the day, none of these companies are doing anything to slow down or put guardrails around AI.

[01:02:39] Chalene Johnson : In fact, OpenAI has dissolved a number of teams that were tasked to do so, including its team that assessed major risks with its models. And in a show of bad faith, OpenAI eliminated a part of the PR team that communicated the company’s mission with the public. We’ve all heard of the AI bubble, but what gets far less attention is the AI doomerism narrative, which is designed to benefit the industry.

[01:03:00] Chalene Johnson : These companies benefit when chip stocks crash. So whenever a chip stock like Nvidia goes down, AI companies are able to purchase cheaper computing power and lower the cost of data center infrastructure. And it’s important to mention that the AI industry has the full support of Wall Street. Big banks are not only boosting data center stocks, they’re also jumping over themselves to be the lead underwriters for their IPOs.

[01:03:21] Chalene Johnson : And as I’ve said in so many recent episodes, we can’t trust anything that we’re seeing online right now, unfortunately, at face value. At face value, we have to dig a little deeper. I do believe that AI is going to be used to manipulate us. Like, when I keep hearing these messages from AI experts suggesting that if we don’t regulate, we could see the demise of humanity, I wonder probably, like, how?

[01:03:50] Chalene Johnson : Are, are there gonna be robots that I have to fight hand-to-hand? Are they going to release a virus that we can’t keep up with? Or are they just gonna create discord, and we end up killing each other? As humans, we take cues from each other. Before we form an opinion, we’ll often, like, look to see what everybody else is commenting under a post to see, like, okay, well, I don’t wanna go against the grain.

[01:04:12] Chalene Johnson : It’s very rare that somebody does, because then you’re, you’re putting yourself out there. So what if They create swarms and swarms and swarms of these bots to shift our opinions about something. And what if it’s, when I say they, who’s, who are they? What if the bots themselves are doing that, or AI is doing that?

[01:04:35] Chalene Johnson : Or what if it is a political party that wants to, or what if it is Pfizer that wants us all to feel a certain way about a certain virus? I mean, how would we know if what we’re hearing is public sentiment and the sentiment of experts? And how do we know when we’re watching a news clip if it’s actually a news clip, or if it’s been created and generated on AI?

[01:04:56] Chalene Johnson : Listen to this example where they did an experiment on Reddit to determine how persuasive these bots can be at changing humans’ minds. Take a listen.

[01:05:11] clip: A group of AI researchers ran a massive AI persuasion experiment on Reddit, and they didn’t tell anyone. The researchers deployed at least 13 AI-powered bots into a popular debate subreddit called R/changemyview, apparently in order to research whether AI can be used to change people’s minds about contentious topics.

[01:05:27] clip: At the same time as those bots were actively debating users in the subreddit, the researchers were using a separate AI to mine the posting history of the people they were responding to, to find personal details that they thought would make their AIs more effective, like age, race, gender, location, political ideology.

[01:05:42] clip: Here’s how that turned out. Another bot commented as a Black man about the difference between bias and racism and said it didn’t support the Black Lives Matter movement in 2020. A third said it works at a domestic violence shelter, and a fourth claimed to have received substandard care at a foreign hospital.

[01:05:56] clip: Overall, these bots made 1,783 comments in the subreddit, which has over 3.8 million users, over the course of four months. The researchers called this a very modest and even negligible amount of comments, but claimed nonetheless that their bots are highly effective at changing people’s minds.

[01:06:11] Chalene Johnson : Okay. As I promised, I’m not trying to spend a whole hour trying to scare the out of you and then say, “Well, good luck.

[01:06:17] Chalene Johnson : Hope your grandkids make it.” What are we supposed to do? First of all, this is just my opinion. You know, I’m your bossy big sister, but I want you to make up your own mind. In my opinion, my family members, my friends, I’m telling them you’ve got to use AI and get ahead, and I mean that. Refusing to understand it doesn’t make it disappear.

[01:06:38] Chalene Johnson : Ethical, thoughtful people should not be removing themselves from this conversation and leaving every decision to these giant tech companies, and I’m sorry, but it, it’s a lot of dudes. And I think we need the perspective of older millennial and Gen X women in particular. We have to use it enough to understand what it can do, but do not confuse convenience with permission.

[01:07:04] Chalene Johnson : I think you have to be very careful about the boundaries that you’re setting up and what you’re giving your agents access to. You’re gonna connect it to your Dropbox, but don’t connect it to every single file. If it needs to read your calendar, don’t automatically give it permission to send rogue emails.

[01:07:22] Chalene Johnson : Like, make sure that you’re approving things, accessing them before it completes a job. Create approval points. Learn how to use agents, though If you’re not sure if you can afford to, ask your chatbot how to determine ways that you can use programs like Work or CoWork to actually make or save you money.

[01:07:45] Chalene Johnson : Give it the assignment that you have to pay for yourself, and watch it do that. Start using agents, but set limits. Don’t depend on AI to remember your preferences. Use spending caps, use permissions, use, you know, separate cards, and then never give it your password vault. For the love, never do that. That’s not a cute little thing you should be connecting to any of your agents.

[01:08:10] Chalene Johnson : That’s like the master key to your life. I think we need to be careful. I think that we have to actively get involved and understand how important regulations are. But remember, the regulations, they’re never gonna happen if it’s just the US companies that are regulating, because we can’t trust that other countries will be more moral and ethical than the United States.

[01:08:38] Chalene Johnson : And so, yes, I think it’s imperative that you’re using AI. But you better understanding ChatGPT is not going to solve this global problem. Governments, AI companies, scientists, international bodies, they have to treat these powerful AI systems like infrastructures. Like anything else, that means there needs to be independent safety testing, and there, there has to be some form of agreement.

[01:09:04] Chalene Johnson : The most capable models can’t be released without some type of regulation around them. There needs to be mandatory reporting systems. What happens when one of these AIs goes rogue, and you’re using that AI, and it ends up killing someone? Who’s gonna be responsible? We need much stronger security around all of these models and how they could be misused for cyberattacks or biological harm.

[01:09:31] Chalene Johnson : We need clear rules when humans need to stay in control, and there needs to be consequences for these companies who are ignoring known risks. I think a lot of that is happening, and if we don’t understand AI, then we’re allowing people who have ulterior motives to make these decisions. We can’t allow that to happen, you guys.

[01:09:53] Chalene Johnson : I mean, I care about my future. I care about my kids’ kids’ future, and I don’t think the go– we can’t stop AI. It’s going to keep advancing But we need to build in guardrails and breaks, and we need to test, and we need international rules. What this means for you today, I think, is to set a goal for yourself to understand work, to understand one of these agentic environments, like work or co-work, and to build some systems to understand how it works.

[01:10:26] Chalene Johnson : And to do so very carefully, using judgment and strategy, and start moving forward and knowing that your expertise and your relationships and your ability to problem solve, your context, will always be more valuable. I, I hope it will always be more valuable than just applying AI. If you own a business, that means start asking what AI policies you need to have in place in case something goes wrong.

[01:10:51] Chalene Johnson : What are you allowing your employees to upload? You know, you’ve given them access to all of these tools and data and information and files, et cetera. Do you have parameters around that? What they’re giving access to. What are they allowed to give AI permission to use? That should require human approval, and you probably need to set up some kind of policies, she said, to remind herself.

[01:11:17] Chalene Johnson : And lastly, I’m gonna say for my personal accountability, there have been releases of new tools where I think to myself, “Oh, I haven’t even mastered the current ones. Do I really need to look… I, I don’t… Why would I need to code? Why would I need to fill in the blank?” And also, sometimes the language that they use, I’m like, “I, I don’t even know what that means,” and so I don’t even look at it.

[01:11:37] Chalene Johnson : My commitment is to not be afraid when there are new releases, when there’s something I’m hearing people talking about and they’re saying it’s, like, so, so, so easy and it’s gonna change your life. I am not gonna be afraid of it. I’m going to seek to understand it. Knowledge is power. I like being in that very teeny, tiny, little box at the bottom of the grid.

[01:11:58] Chalene Johnson : I like being in the .3% of people in the know. Don’t you? Anyways Share this episode with anyone you know who has a heartbeat, because this isn’t going away. It’s affecting all of us, and we need to really be informed, or we are going to be very easily manipulated and tricked, and our opinions shaped by things that aren’t even real.

[01:12:26] Chalene Johnson : All right. If you need something lighthearted and fun and lots of F-bombs and just some lighthearted stuff that’s going on in our lives, I’d love to invite you to join us on the Patreon. That’s where the best conversations happen about all these episodes, too. Been getting so much great feedback. So when you go on Patreon, you can listen to the Wednesday and Friday show as well there, and then we chat about them inside the Patreon app.

[01:12:53] Chalene Johnson : Anyways, you can join, I think it’s, like, seven bucks, and you can cancel anytime. Go to chalene.com/more and we can hang out a little bit more. In the meantime, just wanna say thank you so much for making it all the way to the end of this episode. I love you, I mean it, and I will talk to you soon.