Claude Mythos: The AI Model Too Dangerous to Release

Are you aware of the Claude has this new model called mythos? You heard the news about this. So, do you think that the LLMs are a tool or or more than a tool? >> I me here on the East Coast, I think we're a little bit more, you know, electronic sign go type mentality. But there are a whole generation that kind of the baby boomers, they they want to take it a little bit slower. >> I mean, maybe you can unravel where the future is going in your industry. Where do you think it's going? What's going to happen in the next few years? AI. Is it hype or help? Good day and welcome to the show. I'm your host Pete Cooper and our topic is AI hype or help. Today's guest is Philip Coats.

Philip is a financial advisor. He helps people plan their retirement so they have options when they get there. We're going to talk to him about how he uses it in his work and his personal life. So, let's get into it. >> Good day, Philip. Welcome to the show. Thanks for joining us. Um we know each other through a networking group and you got introduced uh through a common friend of ours. Um but tell us and the audience what you do as a brief introduction. >> Absolutely. So my name is Philip Coats. I am an adviser and my team and I advise on retirement planning, investments and insurance solutions with New York Life. We are based out of the DC area. Um we do work across the 50 states.

>> Right. So anyone who's a US citizen would be a client. >> Absolutely. US citizen or here working um really based in US assets. >> Right. Right. So why would somebody want to do retirement planning? I know it's an obvious question, but I'd like to ask the pain point, you know, and see where that goes. >> Yeah. So the way we really phrase it is, you know, retirement when you get there, it's exactly what you do on a Saturday and Sunday, but it's every day of the week. So, what do you want to do with your free time or your time where it's just you or your family? That's what you're going to be able to do in retirement. How are we going to fund that?

How what level of lifestyle do you really want to live? And then start making plans and start asking questions around that, >> right? And so, what happens if someone doesn't do that? Do they end up in a bad situation or is there a bit of FOMO there or >> Yeah, I mean, it definitely depends on what your goals are. Um it means someone can not plan for retirement and kind of just leave it up to the wind. Um especially in the US we have you know social security we have some federally and locally funded programs. Um but it really limits your options. So the main idea and this is something my dad always said you have money you have options.

Um so planning is an integral part of that and at least leaving certain doors open um is a good way of looking at it. >> Yeah. So, an option might be uh I can keep my home. Uh I can travel on a big holiday every year. Um I don't have to downsize necessarily. Those would all be options, right? >> Absolutely. >> Yeah. Yeah. Well, cool. Cool. Um, I don't know much about financial planning, but a little bit of background in finance cuz I run my own business and a lot of the stuff that I do is I guess retrospective or short-term plans like but not necessarily, you know, long-term uh retirement planning and hopefully I'm still a a few years away from that.

But um, do you entertain people of all ages? Do you say people should join early or late or in the >> Yeah, so nothing's like the present to start is is kind of the answer there. Um, now it changes what type of goal and what type of plan we're going to maybe present to you and give you options for the younger you are and the more you know avenues and maybe the higher risk you can take. Um, we do work with everyone in any age or stage. I mean I have clients who are 80 years old. I have clients who are 18 years old.

Um, so it can be a really wide variety, >> but it's a very different outlook whenever you're talking to, you know, a young family rather than a 70-year-old who's talking more about social security. >> So, it's what is your priority? Um, what is really applicable to you right now and then starting getting something started is is the biggest portion. >> Yeah. Yeah. Philip, how long have you been doing this? >> Um, so I've been in the finance world for about six years now.

I graduated from the University of Maryland um with a bachelor's in finance >> and then worked a little bit in more debt management and contract management with maybe Federal Credit Union >> and then I'm going a couple of years now here with my practice with New York Life. >> Cool. Is it your own practice or are you part of a franchise? How are you structured? >> So I'm structured underneath the New York Life umbrella. It is my practice. So clients that come to me are my clients. Um, >> but at the end of the day, I do own my book of business. >> Okay, great. Well, thanks for the introduction. So, um, what's happening in tech?

We'll get to AI, but what's happening in tech for you over those those years? I mean, you you're not like me as a somebody who started before the internet, before computers even. And uh, I was given a a book to write things in. Um, a real real bookkeeping. My mother used to do the books for my father who had his own business and she did it literally on a Saturday morning every Saturday morning with pen and paper. That's how she did this big ledger. Um, of course, uh, it seems so archaic now. But what changes have you seen? [snorts] >> Yeah, I mean I actually have I think next to me in my filing cabinet I have my spiral notebooks from college. So I have a little bit of that.

Um, but I see I've also seen the other side of it.

My second year in college I went everyone said hey you're not coming to class anymore you stay home during COVID when I really finished up my college so seeing both sides um it is paper pen it's kind of a thing of the past nowadays what you see in the finance world um >> even to the point of like balancing a checkbook that's kind of a foreign or a lost skill >> um but I've seen with New York life specifically in the past couple of years it's been a very slow progression of onboarding new technology new processes um I know it's it's different for every company out there but you have to think about our track record and our history we have policies that are potentially 50 60 70 years old we

have to integrate that um that's the kind of the positives and downsides of being in in business for over 180 years you just have so much backlog and back work to do to integrate new technology >> on top of there's regulatory processes, there's checks and balances, there's security um keeping divisions in information whether it's an investment advisor firm or or you know keeping those so there's no insider trading um >> concern. So a lot of the background to look at um yeah and just keeping all that lines of communication open. Yeah, I I got a friend of mine in Australia who's fairly senior with one of the big banks over there. In Australia, there's five big banks.

He's he's right up there with one of them handling massive massive programs and development. Um, and he says he's so tired of being in meetings all about regulations and compliance. He says that's what he spends all of his time doing. >> It is. It can be it can bog you down. Um, yeah. >> Yeah. I mean in this very fast pace of tech change right now in a way it's a comfort to me to know that the banks are doing that and the finance world is doing that because you know there's so many [snorts] things that uh could go wrong when you're dealing with money and there's vulnerabilities open up and people become vulnerable to scams and things like that.

You like to know that your bank is has got your back. You know that if you put the money in the bank it's going to be there when you need it at the end of the day in simple terms. And if it's like the wild west, you know, you go to the bank and the bank's been robbed. Uh, and you lost your money, too, right? >> Back in the gold standard days. >> Absolutely. >> Yeah. Yeah. So, I mean, onto our topic about AI. So, maybe just tell us if you're using AI at all just in your world um and how it's affecting your your business. >> Yeah, absolutely. So, I do use AI almost every day, if not every day. Um, in either personal or business, mainly in business.

Um, I find that the perks of it and the the streamlined use is nice for being a solo entrepreneur essentially. >> U, I can do a lot of the day-to-day tasks I need done in minutes or seconds. Um everything from either running through scenarios and talking through kind of numbers with me or writing emails, writing responses to correspondence. Um now that kind of goes back to our last question, compliance. A lot of stuff has to go then go through compliance and we actually I submitted something yesterday and >> it as my compliance system asked me is this AI generated? So, just a little key of I guess a lot of agents are using it because they added it to our our questionnaire.

Um, yeah, I don't know an agent that's really not using it and an adviser that's not using it right now. >> So, when you use AI, are you using ChatGpt or Claude or Gemini and those those that how you use it? >> Yes, we are given actually subscriptions to chat GBT for subsidized for those. Yeah. >> Yeah. Well, that's good. Okay. Well, I use it the same way. I made a commitment to double down on it this year and it's paying off for me the more I get into it. Uh, you know, we're we're not an AI native company, but we've been in a would you believe since 2016, but on the vision system side of things. Um, and that's about model development and then camera development.

But then we were blindsided, I guess, by the whole emergence of LLMs, and I'm still still trying to figure out, you know, how best to use it. Um, and everybody I who comes on to this podcast is using those um, chat bot type tools and universally um, thinking that they're they're helpful, right? So, absolutely. >> In what ways is it helpful for you, Philip? >> Yeah. So, I'm a little bit older in the business now, but when I first started, we had some of the programs that were rolling out. They were a little orphaned at the beginning. Um, but now we have interactive conversations. So, a lot of what we do is client facing. >> Mhm.

>> Having that those right things to say, you know, knowing how to pivot if there's an objection. We had interactive AI conversations that we could go on and talk back and forth without being client facing, you know, without possibly. Um, >> that's a good idea. So, like uh your sales training but doing it with the bot. >> Yes. Um Yep. And it would give you different responses.

Is you can kind of pick and tether what conversation you want to have if you want to have it on a product um just a general conversation and then you can also pick the level of difficulty so the more advanced adviserss weren't getting bored of oh this is a you know 5-second conversation that's not real concern right >> so that was nice that was good for for all all levels of agents to really have that resource >> what I like about what you just said is that you're not using the the chatbot without in isolation and expecting that to be everything because you get to to take that and then go in front of a real client and see if it actually worked, >> right?

>> And I think that's that's a really neat way to use it. It's just making you do your job better, better prepared, uh making you better as but not disconnecting from the real world. And I think that's where uh there's a danger in its use is that we take everything that's said as as being the truth. And it can be very overly optimistic in some areas and completely wrong and but be very very confident, >> right? >> And you just take it and then you could be setting yourself up for a big failure, but that's a small failure.

If if you're testing it and you you go down the wrong path, you can recover the conversation and take it, you know, another way pretty easily because it's dynamic and it's human. Nobody expects perfection. >> Yeah, that's great. I mean, um, anywhere else that you're using it? >> So, we do use it and I advise all the newer agents I either mentor or talk to to look at the product information. So, we have fact sheets for all of our products or our investments. A perspectus for a new investment would be a great thing to look at. Upload either PDF, upload the terms, and then just ask it questions or have it kind of break it down and and then have it ask you questions.

For us, we need to know the terms of the riders of the investments. Um, and just going that factual information again more toward geared towards the newer agents, the newer advisors who may not be as seasoned in the terms.

Um, but then you can tie you can kind of put those hand inand right you can say hey you share the two LMS together you can say hey this conversation we're having take some details from it give me either objections or ask me questions based off of the facts I gave you >> so then chat GBT knows this is the product we're talking about these are the terms >> yeah yeah >> then it gets into again client f being able to have that experience without experiencing in front of a client. >> Anyone who's been in front of a client before knows it's very easy to get tongue tied, very easy to lose your place, um, and then just kind of get all off track. So, >> yeah, >> a great resource for that.

So um what I noticed with something like uh a contract and I had a podcast guest on who was talking who was in the legal side and you were saying well look there's a history to how contracts and documents are created and certain words have quite a lot of inference right and if you try to take those words and put them and have the LLM describe it in its own words it may change the meaning slightly because if you go in front of a judge he's going to say Well, those words mean this, but those words actually don't mean the same thing.

And so I think it I think where you're going with it, maybe you agree with me, I don't know, but it's really good to help you understand and unravel the contract, but but be need to be careful how far you can take it. You know, I think you you're getting into danger, at least he was saying, if you say, "Well, just write me a contract and make it simple that I can understand and expect that to stand up and you're going to be in that world, right, where the contract really means something. " >> Absolutely. Yeah. And I've seen that before where they I've had clients ask me, you know, what does this term mean?

Because I I looked it up on Google or something like that and it's like be very careful with that because every everyone nowadays the first response is AI telling you what it is, >> right? Google the Google search engine. So, >> it it can be a slippery slope and something I'm worried about a little bit, especially when I talk to younger clients. You don't really see it in the older realm, but >> yeah, >> I've definitely had that.

Yeah, I think that's uh you know I'm a little bit older than you, Philip, but I see that as a kind of flip things have flipped a little bit that uh people who've had such long life experience in my generation and older um they can spot those errors pretty quickly, right? Just oh, I know that doesn't make sense cuz I actually experienced that back over here and I know that's wrong. And someone that's I guess growing up now or leaving school now with AI under their belt, they just have they just assume everything's correct, right? >> Yeah, absolutely.

I mean, >> it's great for certain things, but if you really want to get factual evidence on it, I would get something in writing, you know, actually learn get to the bottom, get to the root. >> Yeah. I mean, maybe that's the next generation where we get to something more trustworthy.

I was at a uh a construction conference uh talking about what's happening in AI in construction and and a professor from Yale University working in AI said I don't he says I don't believe in AGI um artificial general intelligence I think that's a misnomer I don't think it's it's going to be real and and if it is real we're not on the right track for it because the tools that are developing now with these models are not that he said because there's a disconnection between like it's a pattern recognition engine and it's just just trained on lots of patterns and that doesn't correlate to the physics of the real world.

There needs to be something else uh which hasn't been developed yet which says this is what the real world is like then maybe the two things come together. [snorts] >> Yeah. No, absolutely. I mean I've seen that because I've seen it make up makeup stories or make up things that are not true. Like that's >> and I've even told it you need to reook at that. Type a new response and you >> check yourself. So, >> um >> I I've seen that with people coming out of college nowadays. I don't think it's I don't know if this is a topic we were going to get to, but I don't necessarily see it as a benefit for for the younger generation.

I think that information levels and knowledge levels have become shallower and more widespread >> rather than really deep understanding or a level of knowledge that we've maybe had in prior generations. I think that's one thing we've lost. >> Interesting observation. Right. So I've always uh I was taught at college that the the knowledge your knowledge is like a tea, right? Um it can be like a tea I suppose and that is that you have broad knowledge but then you also have a specialization where you can go deep into one area and I think what you've said is that kind of depth going down is almost just getting shorter and shorter. I don't know if I'm in a bubble, but from my observations, yes.

I mean, nowadays, kids don't really know anything. I say kids, I'm a kid myself, but people come out of college, they're very kind of a shallow depth of knowledge unless they've had some experiences. I mean, that changes whenever you can say, you know, I've done this or I've seen this. Um >> yeah, >> even even me having a background in finance and coming into this world coming to New York Life, I I had mindopening and eye opening conversations that I I had never thought about prior and >> even with six years already in the finance world, I was kind of having my eyes opened. >> Yeah. Yeah. Well, that's uh I think you're right.

I think um one of the experiences I have is uh I I just recently brought on some college grads. Great. They were great. But they would generate uh reports and documents for me that I knew straight away they just prompted. And I was encouraging that. I said, and I do still encourage that, but I said, you have to know every single word and you have to stand be behind everything that you've sent to me. I said, I I do not want to be reading something that you've AI generated, >> right? >> I'm not here for that. I'm here to talk to you as a person. You have to make sure that if you present it to me, it's yours. You cannot you cannot excuse yourself anything wrong in here.

That just just because you use a prompt engine. That's just your tool. You have to be like a hammer, you know? You can't blame the hammer for for hitting your thumb, right? You're the one wielding it. [gasps] >> Yeah. Yeah, absolutely. >> So, do you think that the LLMs are a tool or or more than a tool for you? Do you think it's >> how do you view it? >> I think at this current stage or this generation, it's just a tool. >> Yeah, >> I think that's I think this is kind of as far as you should go with it. >> I think it's a great tool. I mean, just look at like call trees, phone trees now.

Whenever you call any type of business, it's >> gets you a better answer if you're again looking at that surface level. >> But >> when we start digging a little deeper, I think that the personal side of it, especially in the finance world, having that personal experience and knowing how to pivot and really understand where someone's coming from is much more valuable. >> So for for the audience, what's a call tree? >> Yeah. So, anytime you were to call, let's say like a bank, a big bank, and you have questions about your credit card, you have to go through five prompts.

Instead of having to go through the prompts, you talk to more of an AI, you know, generation, and it gives you the answers you need rather than select one, select five, >> that type of conversation. >> So, that we're going into that sort of online chat support chatbot type thing, right? Uh, so you mean when rather than phoning up and waiting on the call, you actually go online and you start going through the chatbot, >> right?

And in that circumstance, I think it's it's a a better tool than I see business owners all the time, they want to implement something to streamline the process and have their employees either work on other projects or dedicate more towards the more in-depth problems that clients are having. M so I had an experience just a couple of days ago with a chatbot and both me and my uh my senior engineer were getting quite frustrated because we [sighs] we had a problem and we were going back and forth and must have been half an hour and then it still had to call up an admin person and then they're asking a whole series of questions. It just went on and on.

Um I sometimes wonder in that sense you know here again it's the implementation right it's it's it can be really good but it can also be pretty lousy in that experience it was pretty lousy because it just was so time consuming all the burden of trying to unravel it all was on us I mean it didn't know where I was in the website the problems I was having it wasn't prepared to look at them didn't give me a channel to look at them I had to copy and paste everything over and then we just couldn't resolve it and we had to go around it a few times to convince it that we weren't getting anywhere and then it finally got to a human person and then we're waiting for that.

So it created I guess more friction than rather than less back in the day before um pre- internet I'd go to the bank every week and I would do my banking. I would put my money in and I would take my money out and I would go to the bank when I needed something and if I had a problem I'd talk to the teller. Of course, I had to physically go there, which was time consuming, park my car, you know, queue up, do my banking, and come out. Um, but at least when I was there with the person, I'm I can express clearly the problem and they can fix it fairly seamlessly because I'm right over the teller, right?

There is a I'm there's I guess we've traded off the physical the difficulty of the physical world of getting there but now we have to pay for it because of the the chat bot and all the the barriers that are there which to me kind of seem more frustrating. >> No, I get that and and something we haven't even really talked about is you probably had a relationship with that teller. >> You maybe knew the banker's name. You >> saw the same people when you came in. So, I've seen it kind of across the country be a little bit different.

Um, I me here on the East Coast, I think we're a little bit more, you know, electronic sign go um you know, can't wait for the elevator kind of get faster um >> type mentality. But there are a whole generation that kind of the baby boomers, they they want to take a little bit slower. They may want pen and paper. Um >> they want that banker to talk to. So again, pretty diverse group, pretty diverse answer depending on who you're talking to. >> So in your world, it sounds like there's still a human interaction needed. Is that is that diminishing? Are you sort of having to be bipodal?

You have you have to have the the personal touch with a say older generation and the younger generation just want to do it all through a chatbot. >> Absolutely. Um I found the younger generation kind of just want to click, you know, fill in the boxes. Um, and then the the older generation, they wanna they want to sit down. They want to talk to you. They want to come to the office. Um, >> for example, I mean this week I was talking to two younger people. We I never even saw them. We did it over the phone on a phone call like >> policies in place. Whereas my retiree clients, he wants to come into the office. We're going to look at the contract together. Wet signature >> check. Mhm.

>> It's a it's a you do definitely have to be able to pivot and you have to be able to do both at the same time. >> Nice. Okay. That's a really good tip for for anyone listening including me when we do our sales work. Um having that bipodal bipodal sit between the two things and just pivot depending on the the client. So is it implied then I mean maybe you can unravel where the future is going in your industry. Where do you think it's going? What's going to happen in the next few years? Yeah, I mean I think more and more people will look towards financial professionals. Um especially as we get into the a the age of kind of information is constantly changing and constantly running faster.

That's ultimately what we've really seen with with you know. Com bubble. Um just the growth of the internet information has become more widespread. Uh but being able to pivot from what I've seen a lot of the AI models they're not able to understand the big picture um >> and that's something your advisor is going to either one know already because of prior inter interactions >> or be able to observe and connect dots and ask the right pivotal questions. >> Yeah. Well, that's that's that's a good way to to to frame it. Did are you aware of the I wanted to bring this one up. It just come to mind. The new Claude has this new uh model called mythos. Have you heard the news about this? >> I'm not.

>> Oh, it's it's threatening the world, if you want to take it that way. >> It is so powerful that uh they don't want to release it to the public. They're releasing it to the likes of Microsoft and Amazon so that they can use it to to tighten up their systems. And what they they found when they run this model on um software vulnerabilities that have been sitting in there for like Linux for 25 years and if that vulnerability is exposed through this new tool, you know, you can take down literally half the servers in the world with with this v with this virus.

You know, um I think banking systems are built maybe even more robust than that in different written in different languages, but modern servers aren't. They're written on Linux and they're written on, you know, stuff that's less than 20 years old and they have vulnerabilities that no one's even aware of. They can't release this to us, to the general public, because there'll be the bad actors out there will just get it and just go straight after those vulnerabilities. So they need to get ahead of it and that's going to trickle down.

I think you know anybody that's this is there's the tech world now JP Morgan's included in that group as well and then there'll be this next layer of things coming through. All the financial institutions are going to have to harden their stuff up. They're going to have access to it and then it'll roll down to the likes of you and me. Who would have thought that that was coming? >> I didn't. >> Yeah. No, that's that's not something I think we really realized was going to be an option whenever whenever it was released. Um, >> but yeah, I've seen that in the finance world, they've been hesitant because of that.

I mean, >> like I said at the beginning, it's we have hundreds of years of contracts in our database. It's just very difficult to integrate safely. Um. >> Yeah. Yeah. Well, Philip, we're coming to the end. Do you have any questions for me? >> Uh, no, not off the top of my head. >> Well, I have got one very important question for you. AI, is it hype or help? I >> think it's help under the right circumstances. >> Do you think you're Do you think you can see how you're going to adapt yourself to those right circumstances and what they might be? [snorts] >> Yes, absolutely. I mean, like I said in the cup in the middle, um, definitely use it as a tool, not necessarily as the end all be all.

Um, I think that the people who promise it as the savior are are definitely overhyping it, whereas it should be that hammer, that screwdriver in your tool belt that you can use it for the right aspects, use it for the right questions and then fill in the cracks and double check it with experiences and your own personal knowledge. >> Yeah. Yeah. Well, it's a good summary, Philip. Look, thanks for being on with us. Really enjoyed um you talking to us about AI and your finance world. Um thank you very much. >> I appreciate the opportunity. Thank you. >> All right. Thank you. Well, that was great talking to Philip about his world and what he does with AI.

One of the things we touched on was sales training using uh a bot like chatbt to riff off of go back and forth what happens when there's objections and how to handle and pivot a conversation to practice practice sales training. I thought that was a great tip. He uses it on a daily basis. Um, and it helps him with his work, you know, and some of the areas that he uses it to understand contracts, particularly not well, not so much him, but people younger than him coming in to try to navigate the contracts and understand them in ways that are simple and easy.

Now he did point out some warnings you know that to assume that everything that the LLM says is correct is a a falsehood and we should be very careful about that and he felt that you know he has enough industry experience to know which way is up but often the new generation do not have that and he pointed out that people coming out of college have what he calls a very shallow but but broad um experience uh in life um not a lot of depth um and I think the LLMs kind of do not help that in my opinion. I think we both agreed on that. Look, if you want to know more about Skillian, please follow the links in the description. Uh we'll also leave a description for our guest uh to for Phillip.

I hope that helps and thanks for watching.