00:00.00 archpodnet All right? This is the start of the second half of segment 2 for the editing go ahead Bill. 00:05.45 Bill White Yeah, we're you know we're talking about things as they are. You know they're going to be in the future and everything and how they are right now but I still think that there's a pathway for a company or a series of companies or state historic preservation office to seriously. Find some kind of a way to use the database of things that they have to automate and really crank things out and you know one thing that I'm really thinking here is the state of Washington has their ah all their site files and everything online and if you've registered your. You know, background and everything you pass the test. You're a professional archeologist and you have access to that website. You can you can put the polygon from ah your gis polygons into that and show the project area and then have it project where all the sites and all the previous projects are within you know one mile or whatever. 00:57.81 archpodnet Um, you know. 01:00.34 Bill White And many of those records, especially if they've been written in the last you know 30 years or so there are already pdfs or digital files that you can then click on the reports, their relevant reports and get all those. So I mean I think everybody can see where I'm going with this if you had a long pipeline or you had some. Building in downtown Tacoma or something like that and there was all these historic properties 1 super easy thing to do would be to just aggregate all those reports and everything together and then have some kind of chat gbt software scrape those reports that are within that boundaries like we keep talking about the the writing thing. Chat gp is just out in the open internet ocean like trying to create things and you know like Doug was commenting it. It makes its own citations up and we don't actually really know all the places that it's getting all the knowledge that it's got and so it's a good place for students or other folks to start off. At that point and use that as the the knowledge. But what if it did just go ahead and tell you what reports it was getting it from where it's getting this culture history where it's getting that summary of all the you know previous projects where it's getting that summary of you know relevant sites even creating tables and breaking them down by you know year. That's the kind of stuff a lot of times that we have ah field text or early career. Um, you know ma graduates or folks that are really starting off. They're the ones who read all those reports and they create all those tables and make those historical contexts and backgrounds and all those different things. 02:29.27 Bill White Ah, you know I'm I'm concerned about that being an absolute possibility. Especially since I was at the essay and I saw that there's one company that has just grown to cover like all of the west and the midwest and they're the biggest company in Crm that we've ever seen you know and it's a company that's always thinking about tech. Specifically how can they reduce labor with tech. They've they've been at the front of recording things on tablets and finding different kinds of ways to synchronize the recordation of things in the field with folks back at home I can see how this would just be yet another tool that would then create a situation where there's like. Perpetual texts who never learn how to do those tables in that entry level writing perpetual crew chiefs who never learn how to do the writing and never learn how to move up to pi and then pi is like what are they good for when you can just have chat gpt write the next scope of work and contract and all that stuff. 03:22.89 Doug Yeah, so one of my ah my one of my favorite sort of like fun statistics is that there are more so in the Uk ten years ago. There were more automatic car washes than there are today as in, we're going backwards and hiring more people to do car washing car washers people are taking the jobs and machines at the moment. Um. And it's true. It's it's like so everyone's talking about all these amazing things that could happen but then like when you break it down. Um ai is not or machine learning is not cheap and like there's this idea that like somehow it might eventually become cheap. But. Honestly, like what's probably going to happen is it's going to hurt people's wages but we're still going to have people do things because it'll still be cheaper to have a person do something than to spend you know the millions it requires to um, go through and. 04:35.98 Doug Ah, it's probably going to have like a minimal effect. So like we're talking like there are already um websites so what you're discussing bill about like uploading all of the field reports and then getting every field report back? um. There's already a site that does that for like grant writing for um, you know if you're a nonprofit or I guess if anyone who wants to do any sort of grant writing like you you upload a bunch of its grants and then it it'll it'll spew back a bunch of boilerplate for you? Um, but like the problem is is like. It's only as good as the data you're putting in there and so like you're going to have a limited and number of reports and like things change. So like you know if you put in all the reports you're going to get things talking about like ah yes, the savages of like whatever you know the the Navajo savages and like ah. I wish I was joking or kidding but like there's reports from the 1950 s like archeological and anthropology reports that like I'm just like oh my god it's like 1950 S Disney racist level of like stuff. Um and like and we're constantly changing and this is like the thing is like. 05:30.28 Bill White Um, yeah. 05:40.10 Bill White Yeah, yeah. 05:49.17 Doug Ah so sorry guys this is could be like a little sci-fi but like you know, depending on like what time traveling sci-fi tv show or or movie you're watching. There's like a theory that basically um, you can't go back further than the furthest you've gone back? Um, with time travel because then you start to c create. Paradoxes and potentially we might like all this machine learning might end at 2022 like as people start to use it to like populate websites and stuff because you can you can look up and see what they used and basically. 90 so ninety eighty some percent of it is the internet like it was the top links on Reddit and then a search engine that crawled a bunch of the web is most everything Wikipedia is like 2% and then there's like 2 databases of books. Um out there like at 8% or something but almost all of it's the internet. But as soon as people start using this and putting you out there. It's going to become circular and it's just basically going to destroy everything. So like honestly as it's currently put together the machine learning that. Everyone's really like in awe about um one is I would call it more of a parlor trick than anything else, but 2 like it might be broken like it may be able to give you an answer as long as that answer is before 2022 because after that the internet's going to be flooded with all these ah machine learning and it is. It's going to be like people are going to plug it in to mass produce web websites and then they're just not going to know. 07:23.71 Doug What is right? or what is just random good sounding things but completely factually incorrect. Um I think what we're talking about is like and this is not to sneeze at but like when we're talking about like improving work. We're talking like. 07:28.77 archpodnet Are. 07:39.16 Doug So a report we we we think a report is a lot of work. But actually if you do the budget like the actual write up of the report the part that like you could put into chat you know wherever or or the Bart or you know whoever's Ai you have that part of actually doing the report is minuscule. It's actually like. Going out doing the excavation doing the survey doing um you know, actually putting that into a data that the machine could then read off like that's where all the work comes in that like we're talking like such a minor bit of our work that might be slightly improved. Um. I think it's it's going to be a huge deal but in the end I suspect possibly wages might not go up that much because it'll be held down but that we're still going to have about the same amount of people doing the work. Um, and we'll probably be taking some jobs from machines soon. Okay. 08:26.37 archpodnet Okay. 08:32.31 archpodnet All right doug we we need to have other chances to speak here. You're going off on the little monologue. Let me just respond to that real quick and then we'll take a break I do too. It's good stuff. But I think man I really just don't believe any of this that you're saying Doug I think you're completely way off base and I'll tell you why. 08:40.24 Bill White I Love it though. So. 08:52.19 archpodnet Because history tells us technology engineering everything goes through growing pains right? There's always you know missteps and there's always you know paths we went down that maybe we shouldn't have gone down and then people realized that or they don't. And then another path goes down There's enough people doing this There's enough companies trying to figure these things out that you know somebody's going to get it what we'll call right for now which may not be right and and we're going to go on down the line but I'm trying to think way further down the line like who cares if they have some sort of. Machine learning algorithm is writing your site reports right now I'm envisioning a future where we don't need stupid site reports. All we care about is the data and the analysis of the data and the trends and what that tells us about history site reports are stupid and nobody reads them. We've said that so many times over and over again. There are. Millions of site reports sitting in in offices of state historic preservation societies and blm um offices and forest service offices and all over the place that literally nobody's ever read and nobody's ever going to read right? No but you don't you don't even read the whole thing you you read the executive summary or you read the the conclusions like everybody else does. 09:52.16 Bill White I read them I read them. That's my bedtime reading. 10:01.80 Bill White So why look at the pictures to the map. The pictures I do that to the artifacts. 10:01.87 archpodnet And nobody gets in and reads. Yeah, exactly exactly exactly I know that's my point though I know that's my point though is that it's just like do we even need that I'm looking towards a future where where humans are allowed to do the. 10:05.22 Andrew And. 10:05.57 Doug Ah, yeah, feel like the picture good picture book. 10:17.97 archpodnet The high-level thinking and the and the really the the big picture kind of stuff that is really going to be difficult for for machine learning I'm not going to say impossible but it's going to be difficult for computer programs to be able to do that right for for a really long time again I think they're going to get there and then really what are we for, but that's ah, that's a question for hopefully a long time from now. But anyway, let's take a last break and then we'll we'll come back and argue about this on the other side. Ah, okay. 10:38.43 Andrew no no I have no no I have 1 thing to say and that's and that's this that there is no future Chris according to Doug Skynet achieves consciousness in 2022 wake up. 10:50.90 archpodnet Right? All right with that. We'll see on the other side.