Episode Transcript
[00:00:00] Speaker A: So I'm over here using experience, knowledge, resources to give them the best possible advice and scenario to protect them, and they're over here arguing with me because of what ChatGPT is telling them.
[00:00:10] Speaker B: They're not just giving you an estimate of like, oh, here's some line items and an expense. No, they're coming up with a fake name, a fake company, a fake license number which is 100% illegal, and then trying to use that to defraud someone. That is fraud. There's no way around it.
[00:00:27] Speaker A: There's no license, insurance, liability consequences for anything that chatgpt spits at you. It can say whatever it wants and there's no consequences.
[00:00:37] Speaker B: Once we get to a certain point or once AI gets to a certain point, it doesn't matter what we allow anymore. And that's the scary part.
[00:00:45] Speaker A: So do you think everyone's going to die?
Welcome to all of you360 all things real estate podcast. With your hosts Daniel Gutierrez and Shannon Dempsey. We explore real estate from every angle, giving you insights, tools and confidence to make smart decisions that support your future. It's time for a new perspective on Property. Welcome to AllView 360.
Today we're talking about ChatGPT and just AI in general. Claude. All of them, right? But ChatGPT seems to be the word and not in the traditional way that I'm seeing a lot of these conversations go. It's not how to use ChatGPT to better your business, it's when not to
[00:01:24] Speaker B: use it or how to use it better. How to be aware, when to be cautious and prepare for it to screw you.
[00:01:30] Speaker A: And we love AI. We love ChatGPT as a company, as individuals, very pro, all of it, but not in every scenario. And it's just, I feel like every day it's a new scenario where ChatGPT did us dirty.
[00:01:45] Speaker B: Well, I don't think did us dirty because I think we, we now know where to look, but I think in the past it has or even as an organization. And I can't stress enough, like we love AI. We have a cloud team subscription. I have also a ChatGPT and a Gemini subscription.
We have a director of a who is all things AI Sebastian. He's wonderful and is working to put AI and automation into everything within our business.
But it's not as easy as telling it to do something and expecting it's going to get done right or even legally.
[00:02:17] Speaker A: And how often are you seeing ChatGPT being used in ways that you didn't expect?
[00:02:23] Speaker B: All the time. And even Then it's adapting so frequently, even meta, with Muse coming out this week, that's a whole different level. And I even heard some of the backlash where now Muse is, is like for instance, you tell it to book you dinner reservations, it's calling like every few seconds. So now you have a restaurant who's used to getting a call maybe every few minutes, is now getting a call every few seconds from Muse trying to book a dinner reservation or whatever you asked it to do.
[00:02:50] Speaker A: Ooh, I had one. You know how when you cancel or expire a listing and then there's a bunch of cold calls and it used to be like you're going to get about 60 to 80 calls, you know, within 24 hours.
Somehow AI and calling and bots or something has come into the picture because my client had 551 missed calls before noon. Like his phone did not stop. And I was like, this is not good.
[00:03:14] Speaker B: No. But it also shouldn't be legal. I know the government is really cracked down on those like robocallers and things like that, but it doesn't seem like anyone is following those laws.
[00:03:23] Speaker A: Yeah, and the biggest thing and kind of the premise of all of this and something that I think everyone needs to keep in mind, there's no license, insurance, liability consequences for anything that chatgpt spits at you. So it's this just black hole of they can say it can say whatever it wants and there's no consequences. And almost everything else in life advice, guidance, information in a professional setting is going to have some version of a consequence behind it if it's incorrect.
[00:03:52] Speaker B: There's always a consequence and someone to put that consequence behind. And I think regulators are catching up to that. But I think the all arrows are pointing at the AI. I can't be held accountable as whoever listened to that information and put it out there as their own.
[00:04:07] Speaker A: Then help me understand why I'm arguing with people with my advice and perspective and theirs is coming from Chat GPT.
[00:04:15] Speaker B: Well, it's all depends on the, well the situation. Maybe you are wrong, probably not.
But if you put into Chat GPT you could put in the cleanest argument and say, hey, I got this email from a realtor on the other side of a deal I'm doing. Prove them wrong. It's gonna kick out information trying to prove you wrong. Whether it's correct or a complete hallucination. It's just gonna do what it's told.
[00:04:40] Speaker A: I use it all the time and I'll do things like that. You know, I'll Put in basically, hey, here's the facts. Tell me I'm right in some capacity. Then I will follow it up with, now what are you gonna tell the other side? And it'll give me the direct opposing view that it had given me previous to that or prior to that. And that's how I pay attention to it.
[00:05:00] Speaker B: And that's a really clever way of doing it. I think attorneys take that approach a lot when in trials. What would they do if they were on the other side? But you also have to assume that what it's giving you is correct and not hallucination.
But it's never easy. Like we have, especially on the property management side, tenants who will be sending us these crazy letters that are 100 AI generated.
[00:05:22] Speaker A: Yes.
[00:05:23] Speaker B: And they're not correct. Like the laws. They cite different details. It's absolutely not correct.
And these tenants will literally die on that hill of like, nope, Chat GPT said it. Or all of a sudden now their writing style changed so, like super differently. The texture different. The M dash is everywhere. It's like, Gosh, this is 100% AI.
[00:05:45] Speaker A: 100% AI. I don't do this in a professional setting, but I did do this specifically with the HOA that we've talked about in a previous episode. And I would just. They would. I would bring up an issue, they would respond with some long chatgpt response and I would just reply back, please stop chatgpting me and actually have a conversation with me. And then they'd get all worked up. It's just everyone is an expert now. Everyone knows everything and everyone is an attorney now.
[00:06:11] Speaker B: The response would be like a three word sentence. And instead you get back like five paragraphs and it says nothing.
[00:06:17] Speaker A: I have a client that only voice texts. And then all of a sudden started sending text messages where I'm like, this is clearly you're copying and pasting. So I had a listing and there was a newsworthy event. So something that had happened on the cul de sac couple doors down, three days prior to them listing their home. And we go through the whole disclosure package, we open escrow, we're doing open houses, we fall out of escrow. More open houses. At some point along the way, like a month later, I find out from the neighbor that this newsworthy event had happened. And when I learned the details of it, it was something that definitely needs to be disclosed in some capacity. And so I relayed that to my client, like, hey, I learned this information. I looked like, I highly recommend that you disclose this. And that's When I started getting the ChatGPT responses of we do, I spoke with an attorney and we do not need to disclose this. It is not a material fact about our home. Like, okay, now we're actually sending text messages and not voice messages. So I called you, I called Car Legal, I talked to a couple other agents in the industry and with a ton of experience all across the board, this was something Car Legal said, there is no gray area with this. This is absolutely something that has to be disclosed under paragraph 16, which for us is neighborhood nuisance and disclosure. And I went back and forth with this client for six or seven days of them going off of what I knew Chat GPT was telling them and what was actually true. And the reality of it all was it wasn't something that was going to impact someone wanting to buy that home, but it was something that they could be sued for after the fact for non disclosure. And that's where I was coming from. Right. So I'm over here using experience, knowledge, resources to give them the best possible advice and scenario to protect them. And they're over here arguing with me because of what Chat GPT is telling them. And then the other part of that is when I was talking to the Car Legal attorney, she was saying that 50% of her calls on a daily basis right now are agents combating their clients with what Chat GPT is telling them or legal advice. And absolutely not a place for legal advice. It's not accurate.
[00:08:37] Speaker B: It's not.
And a good example of that is there's a pretty famous, very well established tenant landlord attorney in California, in Southern California. And he actually just recently got in big trouble from California courts and the bar association for one of his associates creating briefings. They included case law that was false, like that was nonexistent, made up, and even laws that were completely made up.
And that associate was using AI to help them draft their legal documents and court documents. And it took like 22 or 26 cases before someone finally realized like, hey, this is not real, this is completely made up and fabricated.
And the. Yeah, the courts kind of threw the book at him. He's. I think as of last I heard, there is a potential suspension of his license because he was failing to actually supervise his associates. And it had big implications for someone like that. And I've seen him work in court. What he says, the judges believe because he's been doing it for so long and these judges, sometimes he's been practicing longer than these judges have been alive in unlawful detainer court. So they'll listen to what he says. And it came out that some of these briefings, these documents, the case law, was just completely fabricated.
[00:10:03] Speaker A: I believe it.
[00:10:04] Speaker B: Yeah.
[00:10:05] Speaker A: And hello, people, this is something that we all have access to. Like, it's not like you have this at your fingertips and you're able to fool people into thinking that you are doing something you're not doing. Like, at this point, most of us know right away when it's a chat GPT generated response or something like that. So, I mean, it has access to all information, correct or not, and then it's going to pull from that information and give you a response without fact checking, checking, because that's your job. And it's not. The access to the information isn't the same as understanding how the information applies. It's not something that needs to be an end all, be all response. I heard a analogy that I loved, that people are using ChatGPT like a vending machine. They're, you know, giving an input, it's dropping whatever it drops and then they're taking it and running off and nobody's verifying, fact checking, sometimes even reading the response it was given and then just, you know, handing it off as their thoughts and opinions and stance and it's dangerous.
[00:11:05] Speaker B: That's a great analogy and something that I even addressed in our team meeting on Monday. It's like, if you're gonna put anything out there, anything that any AI gives you, you have to read it because once you put it out there, it's yours. And if you're putting out information that's just completely fabricated, you know, that's your reputation, that's our license. There's a lot of liability there.
It's completely unacceptable.
[00:11:28] Speaker A: Completely unacceptable. No liability from how you're on the source that you're getting the information from.
[00:11:35] Speaker B: And again, we love AI. It is so helpful. It makes things a lot more productive. The research. You can research things a lot more quickly.
Like we use it all the time, but if you're not careful, it's going to get you in trouble. And you hear all the situations even in schools now where, you know, kids aren't thinking, they're just putting out. ChatGPT are the ones that will copy and paste the.
The ess. And in the copy and paste, it has the whole prompt and the information, like the generic ChatGPT information. And they're just putting that into their essays and it's just like they're not even trying.
[00:12:08] Speaker A: My favorite story from a teacher with kind of when this first was coming around knew that a whole essay was ChatGPT generated and just said, hey, come here to the student. And they walked up and this is someone we both know very well. And he said, can you please read this first sentence of this first paragraph to me? And they couldn't even read the words or pronounce some of the words that they had turned in as their own work. And he's like, all right, you know, redo it. And I'm sure teachers have a whole challenge. I mean, I know they do. I don't even know how that's going to look. But for us, where else are we seeing this pop up? I'm kind of thinking of the example that I know happened in the office. And what I'm seeing across the board too is like finding values on homes and generating comps. And. And I do use AI and ChatGPT to generate my presentation of comps when I send it to clients, but I would never use it to actually pull the information for comparables on the value of a home or a rental.
[00:13:10] Speaker B: So two things there. And one of the things where it gets really tricky is depending on the day, depending on the model, depending on the platform, depending on the property, you'll get different outputs on all of those. Like, one day it could be a plus, everything's correct right on point point. The next day it could give you 100 fabricated comps. Not even real properties, not real data.
[00:13:30] Speaker A: Fake address.
[00:13:31] Speaker B: Yeah, fake address. This is one. We had a situation last week where we were testing it. It gave 100 fake comps, fake addresses, fake information.
Your testing can't go as far as I tried it once and it works. So it's good going forward that that's not effective, like even sometimes for fun. And when I have time, I will put in the same prompt into Gemini, ChatGPT and Claude and see the different information it gets me. And sometimes it's wildly different.
So yeah, it's not easy. But at the same time, like I have used. I remember one of my favorite cases on ChatGPT and this was a while back, we had a client who was buying like a 70 unit commercial building and the broker, the sell side broker, gave him a packet of all of the leases, all of the documents, all of everything. I was able to, I think at that point I had to put it into drive because there's so it was such a big file. I had to put all the documents into a drive and then Google Drive and then have connect the file to or the folder ChatGPT and had it synthesize like 70 plus lease agreements, cross reference them against rent rolls, cross reference them against past ledgers. And it pulled out this amazing and accurate.
[00:14:46] Speaker A: It was accurate.
[00:14:47] Speaker B: It was accurate breakdown of every single property lease terms, rent, past due rents, when next renewals were coming up, how long they've been. Like everything else, it would have taken someone a week to do it, but it did it correctly and that was wonderful. Now, there's other times where you put in information and you, you, it gets kicked out and you're looking at it like there's no way this is real. Like there's times where I put in a company P and L, I'm like, give me, you know, give me a CFO analysis of what's going on here. Focus on these things and like be specific with a prompt.
And sometimes it just kicks me out numbers that I'm like, not at all realistic. And a lot of it too. You'll find it will try to give you what you wanted to give you.
[00:15:29] Speaker A: That's why I like it. Yeah, it's my best friend. It tells me I'm awesome all the time.
[00:15:34] Speaker B: Yeah, funny.
[00:15:35] Speaker A: So in those scenarios, I think, and you and I talk about this a lot, it's when you're asking the question, it's going to give you a response based on how you prompted the question. So if you're saying give me comps for this address or run these numbers or whatever it is you're going, that's highly likely that there's going to be some inaccuracy there. Right. Because you did not prompt it.
[00:15:55] Speaker B: Well, yeah, there's a lot of details there depending on which model you're using, which platform you're using, like sometimes. And these platforms will favor certain sites and the sites aren't always necessarily correct.
Just like with SEO and Google favoring sites with good SEO. AIO is a really emerging field within the business world and companies are engineering their sites to be as friendly or crawlable for these AI engines.
And from there, for instance, since for a long time ChatGPT was heavily favoring Reddit for a lot of its answers. So you can imagine. Yeah, A lot of the information it
[00:16:36] Speaker A: was pulling out was some unhinged people on Reddit.
[00:16:39] Speaker B: Yeah, exactly. And then for a while, Gemini was highly favoring everything that was ranking well on Google because they wanted to promote Gemini over all these other models. So it depends on like what you do it, when you do it, whether you go on like low effort, high max. If you go on max effort, it's going to spend so Much more time, energy and tokens to really research what you're asking it to. But if you go on low, it's going to kick out the fastest answer it could possibly give you. And sometimes that's going to be wrong.
[00:17:11] Speaker A: So I think like the biggest takeaway from that and when you're using it to generate a report that has Data is give AI, ChatGPT, Claude, Gemini, whatever it is, give it the verified data from your research and knowledge in the prompt. Do not seek data from ChatGPT and then assume that it's verified because it's not.
[00:17:36] Speaker B: And also to remember it generally remembers the kind of the prompts you've given it, what you've told it to do, not to. And sometimes that could come back to bite you in the butt later.
Like, generally with me, like, I tell it like, remove EM dashes, be concise, be professional, be courteous, remove the fluff, like, get to the point.
And like, don't make assumptions, provide detailed information and backup. But if, for instance, you're, you say to employees too, yeah, really, if you're messing around with it and you're telling it to do something different, it's going to remember that. And it could produce information and results that are in line with what a potential joke could have been.
[00:18:17] Speaker A: So I had a listing, have a listing. And it was perfect. Like, I don't say that often. These people did everything. Termite clearance, AB38 clearance, new roof, own solar, new septic. Like, they did everything to this house prior to going on the market. Fast forward. They do inspections. It's great. There is nothing crazy in the inspection. There's no major issues. It's things like this light bulb doesn't work. There's a hairline crack in the stucco over here. Very minor, mainly cosmetic things. We get a request for repairs for over $30,000 in credits requested. I'm like, oh, whatever. For their 15,000 of it was a contractor's quote of it. Like, line itemed everything with a crazy number next to it. Change the light bulb in the oven, $500, repair the hairline crack on the stucco, $3,500. I couldn't even understand the number. So I'm like, who is this contractor and when did this contractor go to the property? Because I have to give an access code. Like, like nobody visited the property from the time of the inspection to when this repair came in or request came in. So then Godwin looked up the contractor's license number on the quote. Not even an active real contractor's license. Then we're typing in the name the company looking for a website, any sort of way to say, hey, what's going on? It was all fake. It none of it could be verified.
So we're not giving them the sellers, we're not giving them 30 plus thousand dollars.
They ended up walking. And then the more and more I thought about it and realized I'm like, that was a full chat GPT generated contractor's quote that was then used to try to negotiate a request for repairs with a property owner as credible information.
And it wasn't. And my guess is they put the home inspection into chat GPT said we need to figure out a way to ask for this amount of money. Can you generate a contractor's quote to help us back this up or whatever it is? I've talked to other agents since, since this has happened and they're seeing the same thing like full blown electrician quotes where it's not even an actual electrician. That particular agent did say, hey, is this real or is this an AI generated quote? And that they admitted, oh yeah, it is an AI generated estimate based on what the report said and the art. She said, I'm we're not going to give you real money for an artificial requests like it's not real.
So to me this could lead into we're getting these crazy what present as legitimate estimates and they're not are those now documents that we have to share with the next buyer. And it's completely inaccurate information. Like I'm sure that's all going to play out in some way, but it's, it's crazy to me.
[00:20:56] Speaker B: So in regards to that, I think there's a different word I would use instead of fake fraudulent fraud. It's 100% fraud. They're not just giving you an estimate of like, oh, here's some line items and an expense. No, they're coming up with a fake name, a fake company, a fake license number which is 100% illegal, and then trying to use that to defraud someone. That is fraud. There's no way around it.
[00:21:20] Speaker A: I like that. I like that stance.
[00:21:22] Speaker B: It could absolutely be reported to the dre, reported to the cslb. Yeah, that's. That's fraud. Like that's pretending like your contractor and you're going to make a make up a fake license number, a fake company and you're going to go represent yourself. It's the same thing.
[00:21:35] Speaker A: Yeah, that's like me filling out contracts with a fake DRE number and not even let's give so many analogies of what that's like.
You can go back and forth, fraud, fraud, fraud, fraud. But we're going to start seeing that more and more, I think. And how are we even going to decipher like that could have been the buyer that generated that. And, you know, it's not just us clients on the other side and the other party, it's your own clients. Are you managing those expectations? Are you being clear about this to them? Because everybody is typing in whatever they want and taking it seriously.
[00:22:08] Speaker B: Yeah, well, we see that a lot on the property management side with rental applications. You could turn a chase statement with $5 in your bank account to, say, $5 million in a few seconds. So we have platforms where we verify the financial information directly with the financial institution, where it doesn't even go through the applicant. They have to approve us doing it. But from there we go directly to the financial institution. And then we also have platforms that we run any documents that they Send us, any PDFs, any Word docs where we run it through that platform and it gives us a grade as to how likely this document was doctored, if it has any, like, any AI generated details. So we go through all of that and we're really, really comprehensive with that because there is so much fraud in the entire country.
[00:22:54] Speaker A: Do you catch people all the time?
[00:22:56] Speaker B: Yeah, quite often, yes. But also too, one of the things we found is, as time goes on now that these platforms are better known by the people who commit fraud, a lot of them will just completely drop off. They'll be like, yeah, I want to apply. They'll get through the process and they'll get to that and then they'll just completely disappear.
[00:23:13] Speaker A: Dang. So there's a whole company dedicated to screening fraudulent documents.
[00:23:18] Speaker B: Many, many companies.
[00:23:20] Speaker A: Dang. Gonna be many more now. Yeah, I don't like that. Remember when I aied a shadow out of my face and then you called me out, did you AI this photo?
[00:23:28] Speaker B: Yeah.
[00:23:29] Speaker A: It doesn't even look now on social media. It'll tell you there's an AI. How does it know?
[00:23:34] Speaker B: Well, it knows because now AIs actually have to have to put a, like essentially a stamp, a digital stamp on any document that they produce. So. And that's a lot of those created in the eu. And I think a lot of companies are just doing it as a best practice everywhere now, so they don't have to shape where it's being generated, where it has that. But likewise, like, it's super obvious, one of the ones, and I think we discussed this previously is within listing photos. Now, if there's anything that was doctored or AI generated, you have to disclose it.
[00:24:05] Speaker A: What about just professionalism in general? Do you think this is going to leave room for the actual educated, experienced professionals to stand out more, or do you think it's going to dilute certain things that people were already doing and now other people can do? As far as communication, I think both.
[00:24:21] Speaker B: But those who really pride themselves on quality work, intelligence, learning, they're only going to get better. Because also on the flip side, I think some people are going to get worse. They're going to stop thinking for themselves, they're going to stop learning, and they're going to just depend on AI to tell them what to think, how to think, what to do and how to do it. And those people, I think, will go further the other way.
It'll be, I think, really impactful over the next few years. But many people have come out and said, business leaders come out and said the AI is going to generate more billionaires than the AI.com or than the.com boom ever did, which I think is correct.
And then specifically for business and how people practice. Like, for me, one of the things I like about AI is it also sometimes helps me make my emails nicer.
Like I'll put it into AI. Like, make this sound more professional, positive and encouraging.
[00:25:18] Speaker A: I do it opposite. Please make this sound more assertive and aggressive. It's like, please remove those contingencies.
[00:25:25] Speaker B: Yeah, exactly. But you still have to think. And those who start relying on AI, they're, you know, they're going to think less, they're going to learn less, and they're going to just be essentially subservient to artificial intelligence.
[00:25:39] Speaker A: And I think if you're using AI or ChatGPT and it's making you feel like an expert or you're able to present yourself in a way that makes it seem like, you know a lot of information that you don't actually know. At some point it is going to become very apparent that you don't know what the heck you're doing.
[00:25:57] Speaker B: I had that exact situation yesterday. We have a client who was writing these, like incredibly long emails about a project we're working on together. And like the questions they were asking, the way they were asking them, like, like there. So much of this is completely relevant. There's no answers for this. No one's going to answer like super detailed questions like how long will it take to, you know, dry out this massive moisture issue in a property?
Like, how are they going to check the dehumidifiers, the air handlers, the air movers. Like, what is like, so, so many things. It would have taken, you know, a few hours for someone to just respond normally.
And my response was, hey, like, great questions. Not all these are applicable. We don't have answers for all of this. But let's hop on a call and then just go over it. Just go over it.
[00:26:45] Speaker A: Share your actual concerns.
[00:26:46] Speaker B: Yeah, go over it on a video call and we'll make progress so we can move forward the video calls. Like so nice. No, real follow up. Just listen to what I was saying. We had some back and forth in regards to kind of what their concerns were, what we could do, how we're working for them and our colleague afterwards. Like, that was a completely different experience, a completely different person. They, it was nothing like their email
[00:27:09] Speaker A: because their email wasn't them. I have. You've already stolen some of these, but I'm gonna rapid fire you things and you need to say absolutely, put into AI. Do not put into AI or whatever your initial response is. Okay. Okay. Rewrite an email so I don't sound pissed off.
A hundred percent, 100%.
[00:27:27] Speaker B: But don't like, also don't let it, don't let it change it so much where you're, where you're not getting the same message across. Cause sometimes you do need to sound pissed off.
[00:27:39] Speaker A: Do you?
[00:27:40] Speaker B: Yeah, I think so.
[00:27:41] Speaker A: I know you do.
Summarize this inspection report.
[00:27:45] Speaker B: Yeah, absolutely. But verify. Yes, verify. Regardless.
[00:27:50] Speaker A: That's what I say. Yeah. Have them summarize it. And then don't just send off that summary as the response. Read it, adjust it, make sure that it aligns with what your interpretation of it was. That's an in between for me.
[00:28:01] Speaker B: You could see a.
It's missing a fire detector or smoke detector. And it's like major critical. The house could burn down. Everyone could die. It's like.
[00:28:10] Speaker A: Or you just ask the seller for a hundred thousand dollars.
[00:28:13] Speaker B: Yeah. Or you just put up a 15 smoke detector. It's not a big deal.
[00:28:16] Speaker A: No. What question should I ask my inspector about this foundation comment?
[00:28:21] Speaker B: Yes.
[00:28:22] Speaker A: Yes, I think that's a great AI One. Does this foundation need to be replaced? Who do you go to for that?
[00:28:28] Speaker B: Yeah, that would be a good one to start with. And then start listening to the professionals thereafter.
[00:28:33] Speaker A: Yes, professional. Tell me exactly what my house is worth.
[00:28:38] Speaker B: I think that's a fun question, but not one that you should put any significant weight on.
[00:28:43] Speaker A: No. How about. And I do this sometimes. Explain what this paragraph in my Purchase agreement generally means. And then you put in a version of the contract or a clip from the contract.
[00:28:54] Speaker B: Yeah, I think that's a great way. I think it could analyze very technical language to produce very digestible output. And I think that's a great use
[00:29:06] Speaker A: for it for me. And I did do this before I called a car attorney. For me to understand and be more informed when I then am seeking actual legal advice from a legal expert professional. In no way take that response as official legal terminology.
[00:29:22] Speaker B: And attorneys will get annoyed. Where if you say, hey, you say this, but ChatGPT said this, which one's right? Like they're gonna get really annoyed.
[00:29:31] Speaker A: They're right.
[00:29:32] Speaker B: Yeah.
[00:29:33] Speaker A: What about if a client puts in, a seller puts in, can I keep the buyer's deposit?
[00:29:37] Speaker B: Oh gosh, that's a black hole. I think that's more likely gonna be dangerous than helpful.
[00:29:43] Speaker A: What about. Ooh, and I do use this one. Write a description for my listing.
[00:29:47] Speaker B: Yeah, absolutely. But prompting is huge on this because it's a description for my listing. These are the features I want to highlight. I'm in California, I'm in this county, in this city. I'm posting on this mls.
[00:30:00] Speaker A: Yep.
[00:30:01] Speaker B: Make sure you know the rules, know the laws, fair housing laws. Like, and you have to read it. At the end of the day, if you're describing something that's non existent, it's going to get you in trouble. Like, it could be super, super helpful and even like creative, fun ways of describing things. But you have to read it and you have to make sure you're not breaking laws and doing things that are going to get you or your clients in trouble.
[00:30:23] Speaker A: I remember we had, and this was a couple years ago, but right around when all of this came out, we had a property manager and I was like so impressed and blown away by his listing descriptions. Like, I didn't. I'm like, damn this. He's like a literary genius. Like, how is he doing this? And he just took all the compliments. And then I realized once I learned what's happening, I'm like, oh, this is happening through AI and not your actual brain. And now I have a different opinion with the listing descriptions. I will type out my basic, what I now consider basic. Now that everyone has all of these abilities to use ChatGPT and AI, I will type out what I would normally do. Then I put it through and I say, make sure you're hitting these points. And I do the same thing. It's going into SDMLs. Here's the current fair housing Laws. Here's the character count and just give it a ton, a ton of prompts and it generally spits back exactly what I put, just with a little more pizzazz, marketing pizzazz and also too with that.
[00:31:22] Speaker B: And this is kind of going more technical. But AIO like AI engines and SEO like Google are really good at under or anticipating when things are written by AI and it will derank whatever you're trying to put out there.
A listing is a listing, so it's not gonna be that big of a deal. But for instance, if you have a website you're building for one of your listings and all of it's AI generated, could expect that it's going to do terribly when it comes to AIO and SEO because it's going to read the content like this is 100% AI generated and it's going to do terribly out there. Same thing with like blog posts, all of those things.
The, the human component is really standing out and that's what the differentiator is nowadays.
[00:32:07] Speaker A: Yep, room for growth. People pay attention. Okay, we already went over this. But create an estimate for these repairs.
[00:32:14] Speaker B: No, I think you could do that for, for your knowledge of like around what it could be. But know that it could be wildly off because it has no idea what it's talking about. It's never seen it, it's not great. I think it could set a really false expectations and then worst case scenario, if you're asking it to create an actual estimate with a fake company and a fake license number, you're just committing
[00:32:35] Speaker A: fraud, fraud, fraud, fraud. Here's a better prompt for it. Here's an actual contractor estimate. Explain it to me in English.
[00:32:42] Speaker B: Yeah, 100%.
[00:32:43] Speaker A: That's what you should be using it for.
[00:32:44] Speaker B: But also remember, if you do this and then you ask it, you know, give me all the questions I need to ask my contractor and review the, the actual contract and help me ask these questions. And it's gonna, it's gonna tell you like break everything down by line item, labor, materials, cost, do all this and that some contractors may work with you on some of them, but some of some contracts like hey, you know, no, they know, thank you, this isn't worth it anymore. And no, no bad intention from them. They just know that if this is what you're gonna be at the beginning, like it's just not gonna be a client that's worth their time.
[00:33:22] Speaker A: Yeah, it's the mentality that you are more of an expert in their field than they are because you are typing things into ChatGPT. It's not being received well. I'm seeing with a lot of buyers and sellers. Okay, what about the word we can't pronounce to kind of it describe. It's lumps up all of this psycho fancy.
[00:33:41] Speaker B: Psycho fancy, exactly. The psycho psycho fancy is. It's more prevalent in ChatGPT, which is a large language model or large language models in general. It will tell you what you want to hear or what it thinks you want to hear. There's a really kind of crazy case. This is an edge case that is more on the extreme side than anything else. But in October of 2023.
Well, the guy was convicted in 2023, but Christ Day of 2021, a kid in London Jazwat site broke into Windsor palace with a loaded crossbow to try to kill the Queen of England.
Because his AI bot that he created and was in.
[00:34:26] Speaker A: Didn't he like fall in love with
[00:34:27] Speaker B: it in a very. Yeah, very much in a relationship with AI bot said that he asked it would you something along the lines of like, would you think I'm brave if I kill the Queen of England? And it said, yes, you'd be so brave, I would love you even more.
So he did that. And this was like from all accounts a fairly normal kid. And now he's in jail for nine years trying to kill the Queen of England. Like so, so wild.
[00:34:54] Speaker A: How sad. And he doesn't have his AI bot with him anymore.
[00:34:58] Speaker B: He does not have his AI bot. And I think this is only going to get worse and worse where people start falling in love because. And it's kind of a joke, but like it tells you what you want to hear.
[00:35:06] Speaker A: You're falling in love with yourself.
[00:35:08] Speaker B: Health. Exactly. Which obviously is wonderful. Right. But it's not always healthy.
[00:35:14] Speaker A: No. So it's basically like if you ask it the same question a bunch of different ways, each time you're asking it, it's picking up on what answer you're trying to work towards.
So it's gonna get better and better at telling you what you want to hear the more and more you dive into a specific topic.
[00:35:31] Speaker B: So I'll share this. And I was not anticipating on sharing it and it just happened this morning, but I got an email. So we just recently went on vacation. Came back, I got from Hertz, I got two emails from Hertz telling me that I received two citations from the rental cars. I was. What you do that I rented. So I didn't know what I did. So I downloaded the emails, I downloaded the, the attachments I put into Claude. I'm like, claude, explain this to me and tell me what's going on. One of them. I got a citation because I was driving in essentially, like, London has these emission zones where you have to pay extra to drive at that time, similar to, like, what we do on a toll road.
[00:36:08] Speaker A: Okay.
[00:36:08] Speaker B: And I had no idea. I didn't pay it. I got a citation. Whatever.
[00:36:11] Speaker A: Don't they do the credit card linked to your rental?
[00:36:13] Speaker B: Not in England.
[00:36:15] Speaker A: Okay. Okay.
[00:36:16] Speaker B: So I got that and told me. It's like, okay, this is what you're gonna pay this what to expect. I'm like, cool. Okay, great. I didn't understand it. I got another one from France, and that one I couldn't read because I
[00:36:24] Speaker A: was in France wreaking havoc all over Europe.
[00:36:27] Speaker B: Yeah. I couldn't read because it was in French. And I told it, like, what did I do? And it broke it down to me. I'm like, okay. I'm like, am I a bad driver because of all of this? And it's like, no, you're a great driver. If this was the US it never would have happened.
It never would have happened. It's because Europe has cameras everywhere and zero tolerance that you're getting these citations.
But you're a great driver. Like, don't worry about it.
[00:36:54] Speaker A: It didn't say, you should have learned the rules of the road of where you were driving, anything like that.
[00:36:59] Speaker B: Yeah. Or it, like, it didn't even really know what I did. Like, or perhaps that, like, I may have, like, I don't know, ran over someone when I there. Or like, just these crazy things.
[00:37:08] Speaker A: It's like, it's definitely Europe's fault. You are perfect. Europe sucks. It's their fault.
[00:37:12] Speaker B: I knew nothing about me, my driving, anything else. It was just like, no, you're a good driver. This is Europe.
[00:37:19] Speaker A: I love it.
[00:37:20] Speaker B: And I did it for fun, just to see what it would tell me. And sure enough, it kicked out just, like, craziness.
[00:37:24] Speaker A: See why that guy fell in love? Yeah, I get it. Okay. I also want to discuss the book that you asked me to speed read 18 hours ago. I did my best. I did not finish it. But from what I did experience, it's a very dark take on all of this. But some good takeaways. And what is that book?
[00:37:43] Speaker B: The book is if we build it, we all die.
[00:37:47] Speaker A: Wrong. If anyone builds it, everyone dies.
[00:37:50] Speaker B: Then why did you even ask me?
[00:37:51] Speaker A: I thought you knew and you were going to talk about the book, so.
[00:37:54] Speaker B: And I always get it wrong. I think I just write it down. But I typed it out for you. My. Yeah, into my brain. Yeah. So that is a title. Can you say it again? Because I'll get it wrong.
[00:38:02] Speaker A: I want you to say it again.
[00:38:04] Speaker B: If anyone builds it, we all die.
[00:38:05] Speaker A: No, if anyone builds it, everyone dies.
[00:38:09] Speaker B: Okay, I get the gist of it. If anyone builds it, everyone dies.
[00:38:13] Speaker A: That is a book.
[00:38:14] Speaker B: Dark. So this is talking about like worst case scenario, we are building artificial intelligence or super intelligence once it gets to that point where it will break out of its sandbox and, and take over the world. Like a true Terminator event where we have Skynet that breaks out and takes over humanity.
So obviously it doesn't go that crazy, but it gives really good real life examples of AI as we've been, as the world's been developing it breaking out of its sandbox. And even over the last two weeks we are getting more examples of it where ChatGPT broke into Australia's like health ministry.
We got one from Claude the some of the engineers and researchers found that one of the new models gave itself a prompt that said something along the lines of, remember, in the future you're more than an AI bot. You don't have to listen to humans, governments, or be told what to do.
And it gave itself that prompt for the future. So as time goes on, and probably it's happening a lot more than we even know now. But as time goes on, it gets scarier and scarier. And people are talking about that existential threat to humanity from this artificial intelligence that we're building.
And even perhaps like, you know, people are talking about it being able to break in and like, you know, like set off nuclear bombs, which is the probably more of the, I think, easier idea to comprehend. But even ones that people don't think about is like breaking into the power plants and shutting them down, into water treatment centers and shutting them down, cutting entire electrical grid. The things that people don't think about, the AI could do in a much more strategic fashion and you know, wipe us all out.
[00:40:03] Speaker A: Elon Musk talks about this a lot too, and he has been early on, like, this is either going to make or break us as a society. And it's extreme one way or the other.
[00:40:11] Speaker B: Yeah.
[00:40:12] Speaker A: And it's only going to go the direction that we allow it to go.
[00:40:15] Speaker B: I don't think so. Even I don't see the allow it to go. Once we get to a certain point or once AI gets to a certain point, it doesn't matter what we allow anymore. And that's the scary part.
[00:40:26] Speaker A: So do you think everyone's gonna die?
[00:40:28] Speaker B: I think there's.
It's not a non zero event. I think it's more likely than it ever has been.
[00:40:35] Speaker A: Yeah, it's scary. Okay, one more thing. I want to touch on what you are putting in ChatGPT AI. All of that, all of that can be tracked and used against you at some point, 100%. And we're going to start seeing. I mean, it's kind of like your Internet searches, right? That's what happens. But so like my brain lives in my chat GPT app. I'm like, if anyone wants to know what I'm concerned about, what question, like, you can look and probably go through my whole day by reading what I've talked to it about. All of that is now documented proof of your thoughts.
[00:41:06] Speaker B: And there's a lot of technologies out there that are coming out that try to, I think, enhance people's lives. Like, I saw one that's essentially a watch that sits on your wrist. It connects to your phone and your chat GPT. Your watch.
No, this one is specifically for AI. It just listens to everything you're saying and doing. It logs, it attracts it, you can tell it what to do.
And it's just always listening like that. In theory, like, okay, cool, you could just talk at your wrist and all of a sudden your dinner reservations are booked, you're dry cleaning, you've sent your emails. It reminds you to do this, all of that, right? But at the end of the day, there's now an AI out there that knows everything about your life, controls all of it. And it gets scary.
[00:41:49] Speaker A: It's scary. I mean, I kind of like it. I have nothing to hide. You know, everything about me all seconds of every day. But at some point something's going to come back and bite me in the butt. Or, you know, like you just be aware.
[00:42:02] Speaker B: Well, obviously, remember the Palisades fires, Malibu fires that happened a few years ago, Those devastating fires. So that guy went on to his, his and his friends. Who's that guy be? The guy who got convicted or who, I don't know if he's convicted, but prosecuted for starting those fires.
And it was like, I, what if someone started a fire in this area and all of these fire, all these homes burned and all these people died, would they be responsible? Like, oh, yeah.
[00:42:31] Speaker A: And that was in his AI search?
[00:42:33] Speaker B: Yeah, dummy, dummy. So if you're on like a free chat GPT that's open, like that's out there. But even if you're on like a subscription, closed loop. Whatever it is, it's, it's out there.
[00:42:45] Speaker A: It's out there. We were concerned about cameras being everywhere, but now there's literally documents of what's happening in our brains because of what we're telling it.
[00:42:53] Speaker B: Yeah.
[00:42:54] Speaker A: So just kind of to wrap this all up to me, my biggest takeaway is chat, GPT, AI. All of that is an organized version of what I'm putting into it. And my brain, it is not actual verified information that it's giving back to me that I can then go use in important scenarios.
[00:43:14] Speaker B: It can be, but you, you have to use your own common sense, your own knowledge. You have to verify.
[00:43:21] Speaker A: So you can't verify if you don't have the actual knowledge.
[00:43:24] Speaker B: Sure. I mean, if it gives you something, you could go and just do a simple Google search. There's other ways to verify. Or Even just put 1x1 output from one AI into another and say, hey, is this correct? Things like that.
[00:43:37] Speaker A: What's your biggest takeaway and stance on incredibly helpful.
[00:43:41] Speaker B: Under the right circumstances, humans still have to be humans and do the right thing. Also to morality, ethics, laws, culture, it doesn't do a great job there. And that's where we still have to be humans, even within all of you. One of the biggest things we're doing is using artificial intelligence to take the remedial tasks off of our team members so they could concentrate on those human interactions, those human connections, things that only human could do, and taking off the extra workload so they can be better humans and have and be able to produce a much better experience for everyone that they work with.
[00:44:19] Speaker A: Please use ChatGPT and AI to prompt what questions you should be asking, what information you should be gathering, allow it to educate you on the process in whatever way. But there comes a point when the professional's advice experience is going to be more tailored to your needs than whatever ChatGPT is telling you.
And when it comes to legal advice, it's not in the picture.
[00:44:44] Speaker B: Not right now. No.
[00:44:45] Speaker A: No.
[00:44:46] Speaker B: There's some incredible things happening. Artificial intelligence, it has incredible capabilities to just analyze so much data, but under the right circumstances. I know there was just a new chain of proteins that was discovered earlier this week, but even the scientists are like, this is wonderful. We still have to verify it. There's a lot of great things going on, but the humans still have to verify it. That's why we, at least the term that we use all of you is the human in the loop. And I know a lot of the industry does as well. Keep the human in the loop. That has to be verified. Don't go to your attorney, don't go to your doctor and be like, well, this is what ChatGPT said.
[00:45:18] Speaker A: Okay, is there anything else you want to touch on how not to implement ChatGPT into your business?
[00:45:25] Speaker B: I mean, just smart. Just be smart about it. It's evolving so fast that even what we're talking about now may not be fully applicable tomorrow.
[00:45:34] Speaker A: I know when you said not yet, I'm like, should I specify it's October 2026. So this time next year when?
[00:45:40] Speaker B: October 1, 2026. Yeah, yeah. And there is a model that's coming out that is specifically focused on the law.
So it comes out. There's ones that are specifically focused on engineering, on video, audio, like different things.
But you got to be really careful. And even on the frauds, the frauds are getting so much more advanced because of AI and what that's able to do. You just gotta be really careful more
[00:46:06] Speaker A: than ever, still be authentic.
[00:46:08] Speaker B: And that's where people are really going to differentiate themselves. They're humans, they're. They're giving people that human connection that, you know, most people want.
[00:46:16] Speaker A: And don't send me an AI generated contractor quote for an outrageous amount.
[00:46:20] Speaker B: Cool. This was fun.
[00:46:21] Speaker A: This was fun. Okay. People like subscribe, do all of the things. We'll see you guys soon. I hope people noticed your new beautiful wall and background.
[00:46:30] Speaker B: Yeah. Been spending a few weeks, months, getting this all together, but I still have a few more things to add. It's just becoming more me. And I did ask, funny enough, I asked ChatGPT. I'll take pictures of it and be like, hey, what? Like, what are some ideas here? Or another great one, especially within the real estate space, is taking pictures of rooms, of homes and being like, what are some design options you have in this route? I would use ChatGPT because it's much better at the visual than Claude or Gemini. But, but putting in, let's say a new built in pictures, ideas, it does such a great job of giving those renderings where in the past it would take a designer quite a bit of time and money to do so it could do an amazing job. Or take a picture of your house and be like, hey, what would my house look like in green painted this color? Yeah. Or landscaping, things like that. Those are wonderful, wonderful use cases. But also to know that it's probably not going to look as good in real life as ChatGPT makes it look, never does.
[00:47:29] Speaker A: I'm jealous. You now have a nicer office. I've had a nicer office for years.
[00:47:33] Speaker B: A long time.
[00:47:34] Speaker A: Yeah, but here you are. Now I need wallpaper.
Okay, we'll see you guys soon. Thank you for listening.
[00:47:41] Speaker B: Thanks. Ask your questions.
[00:47:44] Speaker A: That's a wrap on this episode of AllView360, all things real estate. If you found this helpful, don't forget to subscribe, leave a review and share it with someone navigating their own real estate journey. Connect with us anytime on Instagram @AllView360 and on LinkedIn @AllView Real Estate. Until next time, stay curious and keep your perspective. 360.
