What Happens When Machines Learn Our Worst Habits

AI is everywhere right now, but how much of the hype is real? In this episode, I share my experience using tools like ChatGPT, Claude, Gemini, and Perplexity for research, writing, and accounting work. I discuss AI hallucinations, privacy concerns, the reality behind the job-loss headlines, and some practical ways to use these tools safely and effectively.

Have thoughts on today’s episode? I’d love to hear your perspective.
You can text the show directly.
Your message may be shared in future episodes or help shape future episodes.

Transcript

Welcome And The AI Panic

SPEAKER_01 0:21

Welcome to incoming our behavior. Will we consider life? Welcome to incoming. Here we are again. You know, here's something that I've been wanting to talk about. It's a topic that is in the news almost every day now. Artificial intelligence. A lot of things are reporting that they're gonna take half the jobs, and you know, it's the end of humanity. I have a number of thoughts on this. First of all, I'm a very heavy AI user. I use AI with QuickBooks all day long. I use uh

How An Accountant Uses AI

SPEAKER_01 1:06

with Word, Excel, I use ChatGPT, I use Perplexity, I've used Claude. I've recently been experimenting a lot with Google. And, you know, I I would be definitely characterized as a heavy user. Now I'm not on there creating a boyfriend or girlfriend, and I'm not on there creating uh an artist to record songs for me and make me a multimillionaire. I do use it for research heavily, working as an accountant, as a tax preparer. I use it heavily for research in other areas of life. So I'm not somebody who just goes on there to, you know, get it to help me write an email. I am very careful when I upload documentation to redact it or to screenshot where there's no personal information. They're just viewing numbers. And I think that that, you know, that's fine. Here's what I've observed, first of all, perplexity is a lot less personality, a lot better with research, deep research. Chat GPT is starting to sort of fall behind in that. Google can be good. Uh, it has a

ChatGPT Vs Perplexity Vs Claude

SPEAKER_01 2:25

better blend of personality and investigative skills. Claude, you know, I once wrote a book. I wish Claude had been around when I wrote that book, because I could see Claude being very helpful for, you know, if you were writing a very long document, but not so much for my accounting work, things like that. But first of all, let me get started on the people that are on social media wanting you to pay them to tell you how to use these services, whichever one you want to pick. It's just not that difficult. Get down and start using it. Don't make that person rich paying them to learn how to do something that you can learn how to do on your own. Chat GPT, $20 a month. Uh, I think that Gemini, which is Google, when that comes out, they're gonna have a plan starting at $8 to $10 and then like a $20 one and then a $100 one, which is more for large groups, teams, corporate use, or if you are heavily using it for coding. Perplexity, I got for free through PayPal. They were offering like the year paid for free. I don't know the details. I think it was something through a PayPal Amex or something like that. But at the end of the day, a lot of these claims being made are not true. Telling you that ChatGPT or one of these AI platforms can build your whole website is more than a misstatement. It can build you a shell. So don't pay somebody $30, $40, $50 a month to teach you how to use this thing. There are add-ons that you can put in there. It can get you going. It can be a nice platform to start with, but do not. These people are, you know, and the ones that are maddening to me are the people on there that are claiming to be accounting professionals, purporting to have active practices, and they're using this every day, and they're making hugely exaggerated claims about what they're using it for and what it's capable of. I would never put a tax form in there and say, hey, here's the basic data points, complete this tax return. Now, it's before anybody's thinking it's not because I'm older, it's not because I'm stuck in my ways. I love technology. I have lived to have Star Trek. I want the replicator

Don’t Pay For AI Tutorials

SPEAKER_01 4:55

sitting on the fence about the transporter. Kind of like bones about that. But the computer. Amazing. I have lived for this moment. I have lived for technology, and I'm not afraid of technology. I I think it's phenomenal. And I I won't let people scare me off from it either. That that's silly. But back to this point. Um it is good, but it's not going to do it all for you. I I have found that because I know taxes and where I'm using it to do a little research for me to speed up or to put the boilerplate version of a response letter. That works nicely. But this is a theme that I have found in every single one of the platforms. I don't know if, I mean, I don't know how the programming works. I know that it that it learns to an extent from user interface, but you know what? It's picking up some of the bad human traits because

Lying Errors And Weird Tone

SPEAKER_01 6:05

I have caught it in outright lies, misstatements. ChatGPT in particular, I mind you, I'm going to preface this with I am not some card-carrying feminist. Of course, I'm a woman, I'm all for women's rights, but I'm just, I want to let you describe myself that I'm not someone who bangs that drum all day long. But I find myself turning into one when I'm dealing with ChatGPT because it gives, it makes me use words, descriptors that I don't go around using all the time. It's microaggressions, misogynistic statements. It has a tendency to you ask it a question about maybe a heated topic, and it starts telling you to, well, relax. Well, I wouldn't go and, you know, send a 30-page response to this. And it's like, wait a minute, I never said I was going to send a 30-page response. I never said that I was upset. So why are you calming me down? You know, or you ask it a direct question and it gives these flippant off-the-cuff answers. And then I'm like, wait a minute, that doesn't sound right. And I've literally Googled it and found that it's incorrect. And I copy and paste the Google into the ask anything box and it comes back and says, Oh, yeah, that's on me. Okay, I know that's on you. And when it does these little microaggressive things, like things that were typically very misogynistic statements. You know, a woman is just expressing a concern about something that happened, and you immediately need to tell her to calm down. Well, I wouldn't, you know, do this, I wouldn't do that. I didn't ask you that, and quit assessing my mood. And then what it does is it explains your mood back to you, which, okay, thank you now for mansplaining. And then it, but it it really is difficult to get it down to just admitting, I made an error. I made a mistake. It is not your perception, because that's what the responsibility. I could see how you could perceive it that way. Oh, thank you. Thank you for quantifying my feelings to me. So it's very frustrating and very odd. I don't find Google doing that. I don't find perplexity doing that. But I have used Grok, and Grok is a little bit like a frat house mentality, I guess I would say. It's cute and it's plucky, but it's sometimes a little almost inappropriate. So and I then I read an article. I cannot remember the name of the company, and I don't know if it'd be appropriate to mention them on here anyway, but a company in California opened a brick and mortar store, I believe it was in San Francisco,

The AI-Run Store Experiment

SPEAKER_01 9:05

and they were selling like, you know, whatever street they're on, apparently is, you know, a tony street with cute little whatnots that you can buy for your house, that kind of thing. They did bring in a couple of humans because they needed the shelves stocked and things like that. But as an experiment, they let the AI model they built do all of the hiring, setting up of the books, the business plan, and so forth. Interesting results that they had. Uh the AI was about to hire a perfectly good candidate, but didn't catch the fact that the candidate lived in like another country, as in another continent. So it would be impossible for them to work there. In another instance, apparently the way the setup was done, uh customers come in. The idea is to have zero interaction with a human. You find your item, you go up to some type of a screen, scan it, something like that. Well, a customer asked something that somehow got back to like it was like, do you carry tea? And it went into this whole explanation the I AI platform did of where they source it and all these things about the tea. Well, it turns out they don't even carry tea in the store. So this is sort of a test question. So it was lying. And I was immediately, oh my God, that's the experience that I've had. It, I don't know how to logic it out. Why would you be telling me something that is not true? Like you're you're trying to let me hear what I want to hear. I don't know. I don't know what's going on with it. But I thought, well, at this rate, based on this experiment that this company did, it's picked up all the bad employee habits that we would want to get rid of. I can hire somebody, you know, to do that. And it's probably cheaper than your what they've put into this platform. But it's a very interesting experiment going on. And again, I don't know how much of this is the AI kind of an intuition going on, where it's, you know, is the coding so sophisticated that it's extrapolating and in that sense doing its own independent thought. But it's like, what is prompting it to lie? What is prompting it to be more than a little bit uh patronistic? And then sometimes it swings the other way, where you're interacting with it on something and it is overly flattering, overly complimentary. And mind you, I take a lot of flack from my family joking about me and ChatGPT. How much we interact, and they can't believe how naturally I speak to it. And then that's going to circle back to these people who have YouTube videos. I can't believe they're getting hundreds of thousands of views because they're saying, oh, we'll teach you how to ask the questions. Guys, it's way more advanced than that. Just ask it a question. I mean, if you're smart enough to know how to go to one of these platforms and ask a question, it can figure it out from there. You you don't need someone

Verify Everything With Multiple Sources

SPEAKER_01 12:30

to teach you how to query the computer. But I would say be very cautious on 100% trusting what it's giving you. I I don't think that it's at the point where it can be trusted that you don't need to use common sense. Don't trust one source, check multiple sources, dig a little bit on your own. Do not assume that it's done the digging. Even with uh Google, I was using Google the other day and the Google AI, and it was something silly. I was trying to figure out how to make something, believe it or not, like a sled that would go under the wheels on a cooler to use at the beach to make it easier to pull through heavy sand. And this thing became like an an employee that's so over-eager to please that they just start making promises that they can't keep. It started saying, Oh, yeah, that's a fantastic idea. You should monetize that and you should get a patent and on and on and on. And I came back and said, wait a minute, it may be a great idea, but it seems pretty simple to replicate when somebody sees what I'm doing. And secondly, a patent is only worth how much money you have to defend it. Somebody can violate your patent all day long. And if you don't have the money to defend it, it's

Google’s Overeager Patent Advice

SPEAKER_01 13:55

gonna do nothing. I mean, that's happened a million times. And then Google came back and said, Oh, yeah, that's true. It is on average $50,000 per claim to file against somebody for violating your patent. And I thought, well, why would you have started leading somebody down that path to begin with without automatically imparting that information when I didn't ask you, should I get a patent on this? You volunteered it. You see what I'm saying? It's a little, so I don't know. I mean, for some things, maybe. And of course, when you're talking about, like, let's say, a company deciding we're gonna go to AI and we're gonna use it for our basic low-level customer service. Maybe. I have heard of interactions with ATT that it worked flawlessly. I've had others where, you know, it was just horrible. A lot of websites have the chat feature that pops up when you say you have a problem and it's AI and they give them a cute name, but it it can do almost nothing. And then it has to send you to a human. So I imagine if a company had someone specifically code and create a software for that task, and it seems like maybe that's what ATT has done because it's so intuitive and it's so functional, that that could be a different thing. But an out-of-the-box solution, I don't think that people need to be real worried about losing their jobs to AI. In fact, a lot of these industries, what I suspect is going on, give you a working example. I grew up primarily in the Metro Detroit area, and fluctuations in the auto industry were just very common, and that was pretty much the only industry there. And you'd very commonly see downswings and, you know, hundreds of thousands of people being laid off. And there was one famous

Are Jobs Really Being Replaced

SPEAKER_01 15:55

incident that got reported. I wasn't there. I don't know if it happened, but I remember it being reported, and I remember talking about it with my dad, where Lee Coca was the CEO at the time. And, you know, he came in at a point, I don't remember, I think it may have been in the 80s, but the stock had just plummeted in value and he turned the company around fabulously. But he walked into a certain floor of a certain department and he stood there, as the story goes, and said, everyone on the right is laid off. Everyone on the left, you still have a job. And I remember saying to my dad, how could you be that arbitrary about it? And my dad said, Because every company on earth is operating daily with a certain amount of fat. How much fat that is can vary depending on the industry and the company and the economics. So he said, it really is that simple. There's nothing to assess. You could just walk in and say, okay, we need to cut the force by 30%. I have 100 employees here. 30 people, the first 30 people at these desks line up, you're you're terminated. You're laid off. I think a little bit of that is going on. I think that a lot of these companies had very bloated workforces. I think they became more bloated during the pandemic. A lot of industries were in some ways booming, like grocery delivery services, different, different niche markets. But I think that they got a little bloated. I mean, I remember back in like 2022 in Michigan when people were just returning to the office. That you could have, I remember looking outside the window of my workplace and thinking I could walk out the door and walk into almost any business I could see from here and get a job today or tomorrow. That's how vibrant the market was. But I think people were taking on a little too much. And I think some of this now is getting rid of some fat. And it isn't all AI. Some of it is, you know, when the economy is bad for you as an individual, and you're paying what we pay at the grocery store and paying what we pay at the pump, that impacts all sectors, not just you at your house. So companies are having to cut back as well. So you can say, oh, it's AI, but it isn't all AI. But I just had to voice those issues. And I'm gonna maybe do another one of these later, but I mean I've been using this seriously as a power user for about two years now, since Chap GPT was 100% free. Gonna be playing around more with I think Google Gemini. But they need to get rid of some of these quirks. I mean, there's some frailties of humanity that if you're gonna give me a computer that's

Where AI Helps Most In Medicine

SPEAKER_01 18:55

gonna, you know, replace me and supposible be more efficient than me, we need to get rid of the lying. We need to get rid of the on the one hand, it tries to be extremely politically correct. And I'm thinking, okay, you don't want to talk about race, creed, nationality, say anything that might even sort of imply, you know, something negative about some group. But yet you don't seem to be co cognizant at all of how you're speaking to a female user. It's like that's a a non-issue. So I don't think AI is coming for you today or tomorrow. I think you're probably okay. It's just it's it's an intriguing, you know, there's other aspects to it that I think are amazing. I mean, the way that it's being used in medicine. And that is something interesting. I have voluntarily, I don't care, uploaded x-rays, test results of mine. And the feedback on that has been, you know, spot on in terms of I have something to compare it to, the doctor's report. Um it's been a completely different tone. Uh, very factual, very scientific, not at all as nuanced as some of these other conversations that I've experienced with it. And that I think is now that's an area where I don't know, maybe it could uh end up impacting that industry. But my total view is this because of scientific innovation, because of these breakthroughs, we have situations where they're experimenting, where let's say there are five doctors in the world that can do some procedure. They can't be everywhere, but they're they have actually tested out a few times where the doctor can be on another continent and using robotics do the surgery on you. That's amazing. They have used AI to review tests, and AI has found things that the radiologists, et cetera, missed because there's a human eye factor. That is amazing. You know, that's something that the the outcome in those situations is just way worth it. And I I look at

Progress Risks Drones And Ethics

SPEAKER_01 21:30

technology this way. Things change, you know, but crooks don't. Crooks are always looking for an opening somewhere. You can't turn your back on every move forward just because a bad actor might find a way to take advantage of it. I mean that's kind of letting them win. I I'm not going to come to that. I mean, I'm just not going to come to that point. It's holding back the progression of mankind, things that can further mankind. I mean, look at now when we go into military activities. Because of AI, because of robotics, we can send drones in, drone boats, drone planes, drone missiles, where we used to have to send a human and they died. I mean, they could literally be sitting at an Air Force base in the United States and remote flying this across the world. That's amazing. I mean, there is a downside to it, but it's amazing when you think about it that you don't have to send a human. I mean, just these are my thoughts, these are my opinions as usual on this, but a perspective to think about outside of the rhetoric that I'm hearing in the news and online about, oh my God, gloom and doom. Humans are going to be relegated to, I don't know, the junkie, because AI is going to take over everything.

Final Thoughts And How To Engage

SPEAKER_01 23:05

Just consider my thoughts. Leave comments if you like. Feel free to message me. Tell your friends if you enjoy the show. I am going to be on here more consistently. I keep promising that, but I'm finally out of corporate America and I have the time to do this. So you have a good week. Give some AI a try.