1 00:00:00,000 --> 00:00:10,520 Welcome to the future, technologies and innovations that sculpt our industry. 2 00:00:10,520 --> 00:00:23,040 Learn more about the legal implications of generative AI at BloombergLaw.com. 3 00:00:23,040 --> 00:00:25,640 All right everyone, thank you for joining us once again. 4 00:00:25,640 --> 00:00:27,600 We're hanging out with Donald Thomas. 5 00:00:27,600 --> 00:00:29,280 How are you today, sir? 6 00:00:29,280 --> 00:00:31,080 I'm doing great. 7 00:00:31,080 --> 00:00:36,160 So we're going to spend a little bit of time talking about chat GPT and how it has evolved 8 00:00:36,160 --> 00:00:42,760 and potential uses for it and ways that we can utilize it in our own HVACR industry. 9 00:00:42,760 --> 00:00:49,320 So tell us a little bit about your involvement in the industry and what chat GPT has been 10 00:00:49,320 --> 00:00:51,320 able to do for you. 11 00:00:51,320 --> 00:00:54,440 I'll give you a little bit of background about myself. 12 00:00:54,440 --> 00:00:57,080 I by trade am a building inspector. 13 00:00:57,080 --> 00:01:00,880 I do a lot of inspections on HVAC equipment. 14 00:01:00,880 --> 00:01:03,360 Not just any building inspector though, not just any equipment. 15 00:01:03,360 --> 00:01:05,760 No, just not just any equipment. 16 00:01:05,760 --> 00:01:10,320 But you know, I do a lot of commercial building inspection, do some residential as well. 17 00:01:10,320 --> 00:01:15,760 Now building on that, I've been a college professor in the community college level for 18 00:01:15,760 --> 00:01:16,760 over 20 years. 19 00:01:16,760 --> 00:01:21,600 I teach HVAC, drafting design, you know, a variety of other topics. 20 00:01:21,600 --> 00:01:29,160 And one of the things that I've been successful at is trying to get young folks more involved 21 00:01:29,160 --> 00:01:35,680 in the trades and having, you know, showing them that being a tradesman is actually a 22 00:01:35,680 --> 00:01:36,680 rewarding career. 23 00:01:36,680 --> 00:01:38,680 That's an amazing one, isn't it? 24 00:01:38,680 --> 00:01:39,680 Yeah. 25 00:01:39,680 --> 00:01:42,480 So that's a little bit about myself. 26 00:01:42,480 --> 00:01:47,840 I guess with chat GPT, about in September, one of my students at the end of class came 27 00:01:47,840 --> 00:01:51,320 to me and they showed me this app that they, well, they had it on their phone. 28 00:01:51,320 --> 00:01:53,760 They said, look, Mr. Thomas, look what this thing can do. 29 00:01:53,760 --> 00:01:58,280 He kind of, he gave me my first touch of chat GPT and I said, well, gosh, this is, that's 30 00:01:58,280 --> 00:01:59,280 interesting. 31 00:01:59,280 --> 00:02:00,280 This is pretty interesting. 32 00:02:00,280 --> 00:02:05,800 So I, you know, when I went online, I learned as much as I can about it and I found it to 33 00:02:05,800 --> 00:02:08,040 be an amazing tool. 34 00:02:08,040 --> 00:02:14,880 And I think it has a lot of potential to help us in HVAC industry, but like anything else, 35 00:02:14,880 --> 00:02:15,880 it's still evolving. 36 00:02:15,880 --> 00:02:16,880 Exactly. 37 00:02:16,880 --> 00:02:18,360 There's, and there's a lot to learn about it. 38 00:02:18,360 --> 00:02:19,760 You know, what are its capabilities? 39 00:02:19,760 --> 00:02:27,000 I would say that for the folks interested in chat GPT, it's, it's what they call it 40 00:02:27,000 --> 00:02:29,000 is a genitive language model. 41 00:02:29,000 --> 00:02:34,640 So basically what it does is it reads a variety of texts that was written, you know, it could 42 00:02:34,640 --> 00:02:39,360 have been written 500 years ago, even longer, but if it's published on the internet, it 43 00:02:39,360 --> 00:02:41,080 has access to it. 44 00:02:41,080 --> 00:02:47,520 So what it does is it collects all this information and based on how it was programmed, you ask 45 00:02:47,520 --> 00:02:54,400 it a question and it tries to assemble it in a way that they, it, the program would 46 00:02:54,400 --> 00:02:58,600 probably feel it would be useful to you and it shares that with you. 47 00:02:58,600 --> 00:03:03,200 So if you have some questions on chat GPT, you can go ahead and fire Matt and I'll do 48 00:03:03,200 --> 00:03:05,200 the best I can. 49 00:03:05,200 --> 00:03:06,200 Sure. 50 00:03:06,200 --> 00:03:10,000 So where was chat GPT created? 51 00:03:10,000 --> 00:03:17,480 Well, chat GPT was actually created by a company in California called open AI. 52 00:03:17,480 --> 00:03:25,020 Now open AI has been kind of dealing, you know, kind of building this program for about 53 00:03:25,020 --> 00:03:26,200 several years. 54 00:03:26,200 --> 00:03:34,480 I think about in May or June of 2022, they sort of felt that it was about ready to be 55 00:03:34,480 --> 00:03:35,480 launched. 56 00:03:35,480 --> 00:03:40,680 So what they did was to set up a meeting with Bill Gates of Microsoft and they said, well, 57 00:03:40,680 --> 00:03:44,040 Bill, we got this thing and we want you to look at it. 58 00:03:44,040 --> 00:03:49,560 And he came to Washington state and they ran it in front of Microsoft, ran it in front of 59 00:03:49,560 --> 00:03:51,480 Bill Gates and the rest of his executives. 60 00:03:51,480 --> 00:03:58,480 And at first Bill Gates was not sold on it because he has tried this AI bot stuff before 61 00:03:58,480 --> 00:04:02,760 and it ended up not being that successful for him. 62 00:04:02,760 --> 00:04:09,880 But then, you know, the open AI really blew his mind because he kind of saw the potential 63 00:04:09,880 --> 00:04:10,880 on it. 64 00:04:10,880 --> 00:04:17,000 So at that point, right there, right around that summer, the same time Bill Gates, so 65 00:04:17,000 --> 00:04:23,120 I'm all in and they got a stable version together to release around November of last year. 66 00:04:23,120 --> 00:04:28,400 And then they, that was chat GPT three or 3.5 or whatever. 67 00:04:28,400 --> 00:04:31,520 And then they, now they're up to chat GPT four. 68 00:04:31,520 --> 00:04:40,120 Now chat GPT, they stopped putting information in it as far as what you can research up until 69 00:04:40,120 --> 00:04:44,000 September of 2021. 70 00:04:44,000 --> 00:04:45,000 So anything interesting. 71 00:04:45,000 --> 00:04:47,180 I've actually had that. 72 00:04:47,180 --> 00:04:51,360 So I was doing a little research project and I was trying to find some information and 73 00:04:51,360 --> 00:04:57,040 it said that there was nothing available as of the print date of September, 2021. 74 00:04:57,040 --> 00:05:01,240 So that, okay, that makes things a bigger perspective. 75 00:05:01,240 --> 00:05:02,240 Okay. 76 00:05:02,240 --> 00:05:09,760 Now, now what you could do now, and I found this to kind of work now before I get into 77 00:05:09,760 --> 00:05:16,000 that, I guess people have talked about whether or not, you know, chat GPT is alive or Ascension 78 00:05:16,000 --> 00:05:18,080 in any way. 79 00:05:18,080 --> 00:05:23,120 And it would almost seem that way because it's trained to talk just like the way we're 80 00:05:23,120 --> 00:05:25,560 talking in this podcast. 81 00:05:25,560 --> 00:05:29,460 And it's taught, is trained to teach it, talk in a conversational way. 82 00:05:29,460 --> 00:05:33,920 So, but it's like anything else. 83 00:05:33,920 --> 00:05:40,280 It can probably tell you the weather, you know, based on this program or the weather 84 00:05:40,280 --> 00:05:42,800 patterns of a certain area. 85 00:05:42,800 --> 00:05:50,600 And even if, even if it were programmed up until today, it wouldn't understand that say, 86 00:05:50,600 --> 00:05:56,080 for instance, the weather is this in Virginia where I am now, these types of flowers should 87 00:05:56,080 --> 00:06:00,760 be blooming and these types of plants should be growing and all the other things that, 88 00:06:00,760 --> 00:06:04,440 that make the human more well-rounded to its surroundings. 89 00:06:04,440 --> 00:06:06,320 It doesn't know what your surroundings are. 90 00:06:06,320 --> 00:06:08,320 It's just putting in information. 91 00:06:08,320 --> 00:06:15,800 I would say, will it ever get to the point where Ascension, well, I hope not, but 92 00:06:15,800 --> 00:06:16,800 And we would be outdated. 93 00:06:16,800 --> 00:06:21,000 I'd look at it as a tool. 94 00:06:21,000 --> 00:06:24,320 Just like a hammer or drill anything like that. 95 00:06:24,320 --> 00:06:25,320 Exactly right. 96 00:06:25,320 --> 00:06:28,760 And based on how you use it, you can be very productive with it. 97 00:06:28,760 --> 00:06:32,200 Well, let's talk about some best case uses that you have found since you have this much 98 00:06:32,200 --> 00:06:33,200 experience with it. 99 00:06:33,200 --> 00:06:34,980 I too use ChatGPT. 100 00:06:34,980 --> 00:06:40,000 So for those of you that follow our teaching tips newsletter that comes out of HVAC excellence 101 00:06:40,000 --> 00:06:43,720 on Tuesdays, those summaries are written through ChatGPT. 102 00:06:43,720 --> 00:06:50,000 So I'll typically create a topic and then ask for it to extend that into a longer version. 103 00:06:50,000 --> 00:06:54,920 Or if I am promoting a product, like say our friends over at FieldPeace, they'll provide 104 00:06:54,920 --> 00:06:57,720 me with a video of a procedure. 105 00:06:57,720 --> 00:07:00,360 And then I have a long description of that video. 106 00:07:00,360 --> 00:07:05,840 I may take that long description, put it into ChatGPT, ask for a summary, and then have 107 00:07:05,840 --> 00:07:09,300 a shorter condensed version of the same content. 108 00:07:09,300 --> 00:07:13,640 So as long as I'm using it in the correct context, I tend to like using ChatGPT, but 109 00:07:13,640 --> 00:07:17,200 I've not experimented with too many of the different ways that you can utilize that. 110 00:07:17,200 --> 00:07:20,200 So let's talk about some that have worked well for you. 111 00:07:20,200 --> 00:07:29,120 Well, I found it to do well with things such as business letters, resume writing, I guess 112 00:07:29,120 --> 00:07:32,320 more of a general business application. 113 00:07:32,320 --> 00:07:33,320 Sure. 114 00:07:33,320 --> 00:07:34,320 Professional delivering. 115 00:07:34,320 --> 00:07:35,320 Yeah. 116 00:07:35,320 --> 00:07:37,720 So I found it to work pretty well with that. 117 00:07:37,720 --> 00:07:43,440 And I also found it to be very successful with maybe providing you insight on something 118 00:07:43,440 --> 00:07:49,320 that you might be interested in as far as something in your career field. 119 00:07:49,320 --> 00:07:57,080 Now what I have done with ChatGPT was that if I'm being a building inspector, sometimes 120 00:07:57,080 --> 00:08:05,560 I'm kind of in situations to perform an inspection and I need a way to frame my approach. 121 00:08:05,560 --> 00:08:14,400 Now I would put this in the ChatGPT and what some people might not realize about ChatGPT, 122 00:08:14,400 --> 00:08:20,440 they think you can just answer, ask it just a straight up question and you'll get this 123 00:08:20,440 --> 00:08:24,400 answer and you say, well, gosh, it doesn't know what it's talking about. 124 00:08:24,400 --> 00:08:28,840 Because this is not, but what you have to do is you have to kind of that particular 125 00:08:28,840 --> 00:08:32,480 prompt and I guess that's what ChatGPT calls it. 126 00:08:32,480 --> 00:08:37,720 When you engage in dialogue in that prompt, you're actually training that prompt based 127 00:08:37,720 --> 00:08:39,960 on the questions you're asking it. 128 00:08:39,960 --> 00:08:45,180 So what you're trying to do at that point is actually train it to frame your question. 129 00:08:45,180 --> 00:08:53,840 So say for instance, if I was a television salesman, so to speak, or a salesman in a 130 00:08:53,840 --> 00:09:01,160 department store and then I wanted to promote a product to customers coming in the store 131 00:09:01,160 --> 00:09:05,600 explaining them this new feature, new type of television that we have and all these different 132 00:09:05,600 --> 00:09:06,600 features. 133 00:09:06,600 --> 00:09:14,200 So what you would do is tell ChatGPT to place itself in the role of something, something 134 00:09:14,200 --> 00:09:21,840 like a store salesman or a college professor or a technology expert or CEO of a business. 135 00:09:21,840 --> 00:09:28,800 So when you do that, what it does is it kind of frames the question and gives you something 136 00:09:28,800 --> 00:09:32,220 more what you might want. 137 00:09:32,220 --> 00:09:40,360 So what I found to work best is that instead of just asking one question, you might ask 138 00:09:40,360 --> 00:09:45,040 a series of questions and if you don't get the answer you want, you go back in the ChatGPT 139 00:09:45,040 --> 00:09:49,800 and say, well, focusing on this area, say for instance, if it was a building inspection, 140 00:09:49,800 --> 00:09:52,320 I'm actually focusing on ventilation. 141 00:09:52,320 --> 00:09:57,520 I would say, well, I'm actually focusing on ventilation, more specifically exhaust ventilation 142 00:09:57,520 --> 00:10:01,440 because I'm working in a room with gas appliances. 143 00:10:01,440 --> 00:10:07,400 So when I do that, it kind of channels its insight more in that than something that might 144 00:10:07,400 --> 00:10:09,160 not pertain to that. 145 00:10:09,160 --> 00:10:15,360 So I found it to be very beneficial in helping you frame an idea. 146 00:10:15,360 --> 00:10:20,720 It's a good way to put it or frame and I'll give you insight on how to approach a problem. 147 00:10:20,720 --> 00:10:21,720 Okay. 148 00:10:21,720 --> 00:10:25,560 So you're creating additional dialogue, just more input for the machine to be able to create 149 00:10:25,560 --> 00:10:26,560 some output. 150 00:10:26,560 --> 00:10:28,360 Yes, that's exactly right. 151 00:10:28,360 --> 00:10:33,000 I think that would be an excellent way to kind of say that. 152 00:10:33,000 --> 00:10:39,480 What I did here recently was that I had to do a roof inspection and my first sentence 153 00:10:39,480 --> 00:10:42,960 was how should I do a roof inspection? 154 00:10:42,960 --> 00:10:45,560 And I put it in the ChatGPT and it just gave me. 155 00:10:45,560 --> 00:10:46,880 Oh, that simple question. 156 00:10:46,880 --> 00:10:48,360 How do I do a roof inspection? 157 00:10:48,360 --> 00:10:49,360 Yeah, yes. 158 00:10:49,360 --> 00:10:54,600 And it gave me a whole bunch of information that probably it was trying to like maybe 159 00:10:54,600 --> 00:10:55,800 write the King James Bible. 160 00:10:55,800 --> 00:10:57,800 It wasn't defined enough. 161 00:10:57,800 --> 00:11:01,720 I've seen those responses when okay. 162 00:11:01,720 --> 00:11:03,600 Yeah, for me to do anything. 163 00:11:03,600 --> 00:11:07,840 But then I kind of frame it in a way I told them, you know, I mentioned that the construction 164 00:11:07,840 --> 00:11:13,200 type, the material used and I gave it some background information because this was actually 165 00:11:13,200 --> 00:11:14,680 a commercial roof. 166 00:11:14,680 --> 00:11:22,360 Now once I did that and I framed it more in that light, summarized the response a lot 167 00:11:22,360 --> 00:11:27,440 better and it was and it gave me something, something a lot easier to use. 168 00:11:27,440 --> 00:11:33,840 And then I said, OK, now when someone is working on the roof, can you tell me some safety protocols? 169 00:11:33,840 --> 00:11:40,360 And it kind of went over some things and I put in OSHA safety protocols and then it kind 170 00:11:40,360 --> 00:11:43,520 of nailed down a lot of the things I was looking for. 171 00:11:43,520 --> 00:11:48,380 Now I know that people say they chat GPT may hallucinate. 172 00:11:48,380 --> 00:11:54,840 So when I did that, I actually had working in elevated locations OSHA regulation right 173 00:11:54,840 --> 00:11:55,840 beside me. 174 00:11:55,840 --> 00:11:58,840 So what you want to do is if you're going to use this thing, you want to compare the 175 00:11:58,840 --> 00:12:04,880 answers you get to make sure that whatever information you have is actually accurate. 176 00:12:04,880 --> 00:12:05,880 Yes. 177 00:12:05,880 --> 00:12:08,920 And I found it to be accurate. 178 00:12:08,920 --> 00:12:16,960 So I went on from there and then kept drilling down and I ended up with a quality inspection 179 00:12:16,960 --> 00:12:17,960 guide. 180 00:12:17,960 --> 00:12:19,560 So that actually used. 181 00:12:19,560 --> 00:12:25,480 So it does have its its usefulness. 182 00:12:25,480 --> 00:12:30,720 One of the things I did find that as a limitation to chat GPT was that the documents that it 183 00:12:30,720 --> 00:12:33,480 pulled from have to be publicly available. 184 00:12:33,480 --> 00:12:39,040 If it was something that was copyrighted or restricted in any way, it might give you some 185 00:12:39,040 --> 00:12:45,120 bullet plate information on those publications, but it won't give the exact information because 186 00:12:45,120 --> 00:12:50,920 I think it's programming kind of restricts it from doing too much of that. 187 00:12:50,920 --> 00:12:53,880 And that's been one of the things that I have been very curious with. 188 00:12:53,880 --> 00:12:57,680 So I've actually been experimenting with that a little bit because we always talk about, 189 00:12:57,680 --> 00:13:00,880 well, what is the potential for plagiarism? 190 00:13:00,880 --> 00:13:09,480 So if we take an original copyright material and we copy the entire thing, say we have 191 00:13:09,480 --> 00:13:15,800 a small book, say we have a follow through book, a small descriptive book, and we copy 192 00:13:15,800 --> 00:13:21,520 everything and then we paste it back into chat GPT and ask for it to summarize or ask 193 00:13:21,520 --> 00:13:27,860 for it to rewrite, it will spit out a completely different version of the material that was 194 00:13:27,860 --> 00:13:28,860 inputted. 195 00:13:28,860 --> 00:13:35,980 So if we're looking for material and we find something that appears very similar to an 196 00:13:35,980 --> 00:13:45,320 original copy, how can we validate if it was actually a chat GPT content that has been 197 00:13:45,320 --> 00:13:46,720 from an original work? 198 00:13:46,720 --> 00:13:49,240 Repurposed in a way you might say. 199 00:13:49,240 --> 00:13:50,240 Exactly. 200 00:13:50,240 --> 00:13:59,680 Well, I would say at this point, unless you have access to the original document, it would 201 00:13:59,680 --> 00:14:05,720 be virtually impossible from my perspective because now you have programs, I think that 202 00:14:05,720 --> 00:14:11,560 are available now that could actually tell if something appears to be AI written. 203 00:14:11,560 --> 00:14:12,560 Exactly. 204 00:14:12,560 --> 00:14:21,560 But I don't know how much faith I have in that simply because say AI did write this 205 00:14:21,560 --> 00:14:22,560 thing. 206 00:14:22,560 --> 00:14:31,200 You could pull this document up in Word and you can make some similar changes and do some 207 00:14:31,200 --> 00:14:33,600 reformatting. 208 00:14:33,600 --> 00:14:35,680 Now this is your own stuff. 209 00:14:35,680 --> 00:14:36,680 Exactly. 210 00:14:36,680 --> 00:14:41,080 At that point, does that program know that AI wrote this? 211 00:14:41,080 --> 00:14:47,360 I'm not exactly sure because at that point you've altered the AI's writing. 212 00:14:47,360 --> 00:14:52,560 So I think even though you have software applications, they can kind of figure it out. 213 00:14:52,560 --> 00:14:57,880 I sort of think humans have a way of getting around things. 214 00:14:57,880 --> 00:14:58,880 Absolutely. 215 00:14:58,880 --> 00:15:05,520 Well, we've actually done a little bit of a test on ourselves to see if it actually 216 00:15:05,520 --> 00:15:07,060 worked here at ESCO. 217 00:15:07,060 --> 00:15:10,120 So we had a material that we were curious. 218 00:15:10,120 --> 00:15:15,340 It was submitted to us and we were curious, is this an original writing or is this a writing 219 00:15:15,340 --> 00:15:18,680 that was generated through ChatGPT? 220 00:15:18,680 --> 00:15:24,080 And interestingly enough, we were able to copy the material, put it into ChatGPT and 221 00:15:24,080 --> 00:15:29,040 ask the question, did ChatGPT create this content? 222 00:15:29,040 --> 00:15:34,240 And it was actually confirmed by ChatGPT that it had created the content. 223 00:15:34,240 --> 00:15:35,760 So it's very interesting. 224 00:15:35,760 --> 00:15:41,720 It's very intriguing to be able to look at an AI product that can recognize the difference 225 00:15:41,720 --> 00:15:46,880 between an AI generated product and a human generated product. 226 00:15:46,880 --> 00:15:47,880 Wow. 227 00:15:47,880 --> 00:15:49,680 That is something. 228 00:15:49,680 --> 00:15:56,400 But I think in a bigger sense of things, I think what you're going to see in our industry 229 00:15:56,400 --> 00:16:04,960 is that you're going to have companies that have products and what they will do is that 230 00:16:04,960 --> 00:16:09,960 instead of you reviewing a bunch of service manuals, building code specifications and 231 00:16:09,960 --> 00:16:16,520 a variety of as-re documents and a variety of other things to make a technical determination, 232 00:16:16,520 --> 00:16:19,920 you're just going to have a prompt and all this stuff is just going to be preloaded for 233 00:16:19,920 --> 00:16:22,680 that particular system. 234 00:16:22,680 --> 00:16:26,760 And you're going to ask it certain questions and it's going to just file doing a whole 235 00:16:26,760 --> 00:16:30,520 bunch of extra research just to tell you what you need to know. 236 00:16:30,520 --> 00:16:33,280 I think that's where this whole thing is going. 237 00:16:33,280 --> 00:16:40,640 And I think that's where it would make sense because one of the things that I've been involved 238 00:16:40,640 --> 00:16:43,920 with as a building inspector is plan review. 239 00:16:43,920 --> 00:16:50,240 Plan review is very tedious and it's very time consuming, but all the information is 240 00:16:50,240 --> 00:16:51,240 there. 241 00:16:51,240 --> 00:16:54,360 You just have to find it. 242 00:16:54,360 --> 00:16:56,100 You have to look at building codes. 243 00:16:56,100 --> 00:16:58,760 You have to look at international mechanical code. 244 00:16:58,760 --> 00:17:00,640 You have to look at the drawings. 245 00:17:00,640 --> 00:17:03,200 You have to look at as-re standards. 246 00:17:03,200 --> 00:17:11,720 But if all of that was kind of built in and you just asked a certain question about a 247 00:17:11,720 --> 00:17:16,760 mechanical requirement in a certain space and it gives you your options, that would 248 00:17:16,760 --> 00:17:21,360 save a lot of time and it would probably cut down on the human area of it too. 249 00:17:21,360 --> 00:17:27,160 And I think that's where ShatGPT or any type of AI program is going to make a difference 250 00:17:27,160 --> 00:17:29,920 in the HVAC industry. 251 00:17:29,920 --> 00:17:31,320 Oh, absolutely. 252 00:17:31,320 --> 00:17:37,000 One of the things that I anticipate seeing AI involved with is building automations. 253 00:17:37,000 --> 00:17:39,040 So I was a mobile engineer. 254 00:17:39,040 --> 00:17:43,960 I did building automations, refrigeration, HVAC, all of these combined inputs that are 255 00:17:43,960 --> 00:17:45,720 already existing in structures. 256 00:17:45,720 --> 00:17:52,760 Well, what if you have an AI generator that is sorting through these inputs and looking 257 00:17:52,760 --> 00:17:57,280 at the outputs compared to analytics that it already has? 258 00:17:57,280 --> 00:17:59,980 And I did a podcast a while back. 259 00:17:59,980 --> 00:18:02,360 The topic was Hello House. 260 00:18:02,360 --> 00:18:07,160 You approach a home or structure and we don't know how the house is responding. 261 00:18:07,160 --> 00:18:10,400 We have to use tools to analyze and come up with our own calculations. 262 00:18:10,400 --> 00:18:16,160 Well, what if a building was powered by AI and you walked up to the AI controller and 263 00:18:16,160 --> 00:18:21,520 said building, how is your system performing today? 264 00:18:21,520 --> 00:18:29,520 And it could analyze inputs and outputs versus original program content and actually respond 265 00:18:29,520 --> 00:18:30,520 based on inputs. 266 00:18:30,520 --> 00:18:32,560 Sort of like Star Trek. 267 00:18:32,560 --> 00:18:34,400 Isn't it crazy to think? 268 00:18:34,400 --> 00:18:35,400 Yeah. 269 00:18:35,400 --> 00:18:40,600 You would have never thought that it would get to that point, but we're only about maybe 270 00:18:40,600 --> 00:18:43,080 a year or two from that right now, I believe. 271 00:18:43,080 --> 00:18:44,080 I believe that as well. 272 00:18:44,080 --> 00:18:45,440 I believe we are approaching that. 273 00:18:45,440 --> 00:18:50,400 As far as the industry, I don't think it's anything that anyone needs to be afraid of, 274 00:18:50,400 --> 00:18:53,120 but it's like any other technology. 275 00:18:53,120 --> 00:19:00,760 One of the things that I stress during my time as a college professor, if you don't 276 00:19:00,760 --> 00:19:06,280 embrace technology, it's going to impair your ability to stay employed. 277 00:19:06,280 --> 00:19:10,960 And it was also going to impair your ability to progress in your trade. 278 00:19:10,960 --> 00:19:18,840 Give an example, when I first became a building inspector in the late 1980s, early 1990s, 279 00:19:18,840 --> 00:19:22,720 I had a Polaroid camera that would go on the job site. 280 00:19:22,720 --> 00:19:27,160 And I would take the pictures of what I saw and I had to lay all the pictures out so they 281 00:19:27,160 --> 00:19:28,160 wouldn't blow away. 282 00:19:28,160 --> 00:19:29,160 Sure. 283 00:19:29,160 --> 00:19:30,160 And they had to dry. 284 00:19:30,160 --> 00:19:36,120 And then I had a three ply copy of my inspection report and paper clipped all that together. 285 00:19:36,120 --> 00:19:38,240 And I tossed that in the supervisor's tray. 286 00:19:38,240 --> 00:19:40,560 And that was my building inspection. 287 00:19:40,560 --> 00:19:42,880 Now all that's gone now. 288 00:19:42,880 --> 00:19:46,480 You know, all that technology is done away with all that other thing. 289 00:19:46,480 --> 00:19:53,000 But people who remember or want to participate in that level of building inspection probably 290 00:19:53,000 --> 00:19:54,240 wouldn't be employed right now. 291 00:19:54,240 --> 00:19:56,680 Everything's on the tablet right now. 292 00:19:56,680 --> 00:19:59,000 You don't even have a lot of paper drawings at this point. 293 00:19:59,000 --> 00:20:03,480 A lot of the drawings are actually scanned into your tablet. 294 00:20:03,480 --> 00:20:09,520 So technology has completely changed my career. 295 00:20:09,520 --> 00:20:11,160 But I've always embraced it though. 296 00:20:11,160 --> 00:20:15,240 So I've always enjoyed using technology in my trade. 297 00:20:15,240 --> 00:20:17,600 So it was something I enjoyed doing. 298 00:20:17,600 --> 00:20:25,320 But I found people who didn't embrace technology typically either they stagnated their growth 299 00:20:25,320 --> 00:20:28,760 in their trade or they didn't stay employed in their trade. 300 00:20:28,760 --> 00:20:29,760 That's exactly right. 301 00:20:29,760 --> 00:20:35,000 Well, Donald Thomas, thank you so much for joining us today and very much appreciate 302 00:20:35,000 --> 00:20:37,800 the opportunity to learn more about ChatGPT. 303 00:20:37,800 --> 00:20:44,320 Hopefully we can do another broadcast when an application is developed that can really 304 00:20:44,320 --> 00:20:46,000 be a big benefit to our trade. 305 00:20:46,000 --> 00:20:51,400 We can kind of share it with everyone and everybody on board with it and basically try 306 00:20:51,400 --> 00:20:53,720 to move the industry forward into AI. 307 00:20:53,720 --> 00:20:54,720 I love it. 308 00:20:54,720 --> 00:20:57,600 I got my attention and I am all aboard. 309 00:20:57,600 --> 00:21:00,320 Well, thank you so much and have a wonderful day. 310 00:21:00,320 --> 00:21:01,320 You're welcome. 311 00:21:01,320 --> 00:21:14,840 Thank you.