335 - AI Overdose:The Hidden Cost Of Convenience

0:00AI overdose, the hidden cost of convenience. Now make no mistake, the AI tools are fantastic tools, but there's a lot to learn about how to use them, can they make mistakes, etc. etc. So I want to spend some time on that because it can be a tremendously powerful tool, but it can also mislead you with information. And I want to give you some examples of that. I was talking to a colleague of mine who's a lawyer in the healthcare space, and we were talking about data and where to get data and reliable data. And he said, one of the interesting things you might consider is if you have a paid subscription to ChatGPT, which I do and probably a lot of you do as well.

0:46You have access to its new task feature. I've set up tasks for it to search the web each day and find specific healthcare-related news, but there's no reason why you can't give it instructions to look for anything else in particular. I have two of those tasks running at the current time, and I have each one report to me early in the morning as to what it's found. Okay, it's a good use. It keeps you current. This is not only good for finding specific types of news. It's not a substitute for general news such as the Wall Street Journal, etc. But once you're chat GPT proficient, that part of it's important, you can have it automatically create short blurbs for you to then slightly rework into blog posts, scripts, etc.

1:37I wouldn't trust it to do completely because you get a lot of stilted language, although I've trained it in my style and it's pretty good but not great at copying it. And you have to be extremely careful of hallucinations, quote unquote, which is basically bullshit that AI tools like ChatGPT adds to its responses, some of which can be incredibly incorrect or just out of left field. I can give you examples if that would make you cry. Well, I don't need examples because I actually created my own example by mistake. I wanted to give ChatGPT six PDF reports on the economy, and they were 40, 50 pages each, and I didn't want to read each one if I didn't need to read each one.

2:28So I ended up asking it to and uploaded those PDFs to ChatGPT and then to go through each of those and summarize important points. And it did that incredibly fast. But it also created a graph, a couple of graphs, one of which was on Fed rates and the other one was on labor. And so when I looked at it, it just didn't seem right. And so I said, can you locate the specific source from the reports I sent to you? And it wrote back, I couldn't locate a direct reference to the M &A structures pie chart in your uploaded reports. However, I can still recreate a pie chart. So basically, it didn't take it from the data.

3:18Now, the other one was a little more concerning. When we were talking about federal rate actions, it said that the basis point reduction, interest reduction in 2024 was 75 points, not 100 points as was referenced. And so it gave me, it said, here's a breakdown. It said there was a 50 % basis point in September 18th, and then in December there was another 25 for the 75. And then I said to it, well, that's not what I understand. And then it wrote back, apologies for any confusion caused by earlier information. Let's clarify the current federal rates and recent adjustments. And then it said that basically the rates were 50 basis points in September, but then 25 basis point reduction in each October and December.

4:18so there was a 100-point basis point, not 75. So that's a great example of how the information can be wrong, just like if people did that the information could be wrong. So one of the things that I wanted to point out to you is AI is certainly a great tool, absolutely, but you have to not get lazy. You have to not say, okay, just do this for me or tell me this, and as a result of that, then I've got all the information I need. I don't need to verify it anywhere, et cetera, et cetera. And particularly, I think that one of the things that I've seen is where it has to do calculations. You have to be very careful with that.

5:10So remember that if you would. Remember it's a great tool. Remember that it can certainly do things for you and save you tons of time, but you have to understand how to use it, how to give it additional prompts, how to refine it, how to ask it to check something so that you just don't put in a very simple, lazy question and leave it at that. You have to use that. So again, it's a great tool, but it's not a tool that you can just sit back, rely on, and assume that everything you get is correct. And there's plenty of examples where that's not been the case. So I hope that you understand how powerful it is.

5:55I hope that you understand its limitations. And I also hope that you understand how to use it. Now, my marketing director and I created a course, How to Use AI to Supercharge Your Business Acquisition. That is still current in terms of the lessons you'll learn on how to use AI. I would highly recommend that you go over to brucewhipple.com and take a look at that and decide if it would help you or not. The other thing, they are continually, whoever the they is in chat GPT, adding new functions. You can do role plays. You can ask it to be a seller. You can ask it to be a buyer. And you can go back and forth with it.

6:37And it's incredibly powerful to be able to do that. Ask it to be a seller that's skeptical. Ask it to be a buyer and you play the seller. You can do all those things. So go over to BruceWimple.com. Take a look at that AI course. I think you'll find it helpful the feedback on it was great but use ai tools but understand how to use them and the limitations of them and the benefits super important