Human Perspective on an AI-Assisted Podcast Season

Welcome to Building Better Developers, the Developer Nord podcast, where we work on getting better step by step professionally and personally. Let's get started. Hello and welcome back. We are continuing and potentially wrapping up yet another season with Developer Nord Building Better Developers. I am Rob Rodhead, one of the founders of the pre-mentioned Developer Nord and Building Better Developers. Also, the founder of RB Consulting, where we are boutique consulting. We are a fractional CIO. Bottom line, we help businesses assess and simplify their technology. We figure out a way to walk through with you what you've got and how to leverage it so that there is a roadmap for success for moving forward for growth through your company. Bottom line, it really comes back down to we assess where we're at, help you build a roadmap, let's get you to a better, brighter future, essentially. We do that through simplification, integration, automation, innovation. There's a lot of different ways that we can utilize technology that you can utilize technology, but also that we can utilize it to help you do it better. That's what we're here for. We help you craft a specific recipe for you, custom to your business. We connect your business goals and where you're at with where you want to be and how to get you there in the best and brightest future that you have ahead of you. Good things and bad things. Oh, actually, I guess I would be amiss if I didn't mention rb-sns.com. You can check out our website, which has been, we were just talking in the pre-show about like it has been evolving. We've been working on it. We've been doing a lot of cool stuff. We like it. Give us feedback. We'd love to hear how you like it. Rob at rb-sns.com. Also check out our tool, matrix.rb-sns.com. It's really a cool way for you to get quickly to a roadmap of sorts. It's not quite as custom, but it's one of those that definitely, based on where you're at, can help you figure out what your next steps are. And it's free. You just go in there, sign up, answer a few questions. I think it takes about 10 minutes, something like that, depending on where you're at. It's a little bit technical, but it will help your business get to the brighter future or at least give you the steps to do so. Good thing, bad thing. Good thing is we are wrapping up yet another season. This is like, really, I can't tell you how much it is just heartwarming, how much it is like the endorphins and all that cool stuff. Getting through another season when you started way, way back, episode one, and you weren't sure you were going to get through any episodes. You weren't sure how far you were going to go. You weren't sure how far this was going to last. And the fact that we were just like marching into now what I think 26 is going to be our next season. That's a lot of seasons is way beyond what a lot of people do. Sometimes you have to mute your mic when you have a sneeze or two. Apologies for that. The bad thing is that we are through another season. It is always a little bit of a challenge to like, what are we going to do next? It's sort of funny that this has been a challenge that we have now faced multiple times. Where do we go next? What do we do? What's our next topic? One of the things we may discuss is actually going out to AI and saying, hey, where's an area that we haven't covered? We've done that. We actually got some pretty cool little information back. If you check out the YouTube channel, you'll be able to check out part of that bonus material. But first, I think I need my co-host to introduce himself. Go for it. Hey, everyone. My name is Michael Milash. I'm the co-founder and developer of Envision QA, where we help businesses take back control with custom software that builds around the needs, not the other way around. So our focus is simple. Great service, great solutions, and a rock solid quality. We build tools that replace frustrating systems, streamline operations, and are fully tested to work right the first time. At Envision QA, we combine development and quality assurance to give you software that you can trust and support you can count on. Check us out at EnvisionQA.com. Good thing, bad thing. So good thing, I guess it's mixed, good and bad. So I'm another year older, another year closer to the 50 mark. Not quite there yet, almost. And I'm having to reflect on a lot of things and update all my freaking policies and things like that. You know, as you get older, you have to worry about things like mortgages, health insurance, life insurance. And, well, I just got dinged by or pinged by my mortgage broker. It's like, oh, interest rates are going down again. You know, do you want to refinance? And it's like, do I? How long do I have to retirement? These are questions I don't want to ask yet. But anyway, so that's kind of the good and bad. Another year older, had a wonderful birthday. More crap to go with it. Mute. Here's to one year older and not realizing when you've actually turned your little mic off. Apologies to that. Yeah, I wish I was getting closer to 50, but I'm getting closer to 150. So that's just sort of how these things go. More importantly, we're wrapping up the season. We have spent another 30 plus episodes talking about getting better developers. This time we did it with AI. And what I want to do in this wrap up instead of, and we can, I mean, this is, we'll see where Michael wants to go, but I want to spend a little more time in a meta retrospective a little bit of like, looking at this is what we did when we came into this was said, hey, AI is a big thing right now. Everybody's talking AI, AI, AI, AI. People that can't spell AI are talking about AI. I had a conversation today and it's like with a guy that they're in this area where they really need AI and they like all these people are experts because they've done like three things. There are very few people that have spent more than I don't know, three to six months probably playing around with AI. Most of them wrote a chat bot. And it's really about like, OK, great. Nobody can spell it, we'll say, and then thinks they're an expert. But let's really talk about like, how do we use it? How do we utilize it? And during this season and a little bit before, but definitely during the season, it really has been a journey on what is AI provide us? Where can it help us? And where do we need to still like make some progress? And I would just like in this summary, this little retrospective, just look back. It's like it really has been, it's been nice as being like a third host, another co-host in it that it's really brought, although it has definitely touched on things that we touched on, which is probably for the same kind of reasons, because when you have a certain topic in a well-known field like this, there's, there's like, there's, it's a target rich environment. In a good sense, we get a good distribution of things too in or offline as well as on a regular basis. And then it's the ability to just sort of like divide and divide bitch over 好, mainly in terms of information, you can really swing this around. You, you really missed out on a ton of things that you can talk about. Where you're like, oh yeah, that makes sense and that makes sense. And that makes sense. And this is probably shifting the mission. You know, it's like when we started way, way back, there's things that we talk about the software development lifecycle. There's certain things you're always going to talk about with the software really good about that. But what we've seen is that it's really good about touching on additional areas. It really is. It's another set of eyes, another set of ears to say, well hey have you thought about that? And I think that has been a big boon going through this season is that there have been some areas where we didn't really connect it and sometimes where it misunderstood we'll say where it took it a little bit different direction but I think we got some great conversations out of that. So I'll pause there and get your like you know initial thoughts. Yeah so I think I don't remember which one was initially kicked off the idea of hey let's like revisit a past season with AI but it became an interesting journey. Initially we're like oh well let's just see what it can do and we'll just go with it and as we went with it it really did a decent job of analyzing what we talked about. In some cases it was completely different but in a good way where it's like oh we didn't even think about that and it gave us more ideas to kind of expand upon past episodes. One of the best things I can I guess we compare it to is it's like throwing ideas at a whiteboard seeing what will stick and AI was great for that based on the way we used it. Now would it be great to actually flush out the content for the full episodes? I don't think so. I think we'd get into that kind of false fake or not fake but just where it would just kind of give us an answer but not really be an answer for a lot of these solutions. So for something a little higher level like summaries and outlines and things like that I think it did a great job this season. I think so too I think that's a really good point is that it gives us yes we can get some very good details if we want to out of AI but for something like this it's much better at a high level at a summary kind of level it was sort of like hey here's an outline or something like that and let us take it from there. There are more than a couple of times that I think it came back with some stuff where it was just like huh that's an interesting way to go we don't want to go there. It didn't have the right context of the term or the topic or things like that so it did it could get lost but honestly so can we. There are times that we would put together a topic and we would go in very different directions because we were thinking different things so I don't think that's a I don't know that's necessarily a knock against AI but I think it is a caution against AI. No I agree I mean it gives you a good it's good at giving you content but it's not necessarily good at giving you the meat of what it is that you're trying to convey. Yeah I think that's this goes back to like some of the other things I've done with AI that are outside of the developer realm which actually we'll see some of this as we like part of things I'm gonna be doing when I'm revamping the website is basically a lot of it is like take this stuff take this bucket of stuff and reorganize it. It's like it's sometimes it's not much different than like just shuffle this around but sometimes the shuffling makes a big difference it's just like A-B testing where you're like well what if I put the button at the top versus the bottom versus the left or the right or things like that. That's some of what it is it's basically like say okay let's take this and let's take honestly in some cases you know like best practices or standards or things like that and let's like apply that to this thing so that we can get something done that maintains the meat of it but also dresses it up a little bit and it's it is sort of a gosh it's so it's really at the end of the day the things that work best are the ones that are process oriented are the things are like there's always a series of steps that you want to get through. I will share which is there's a lot more to go into it but I will share like one of the things we've done that I've found has been really good is using it as a an advisor of sorts like we we wanted to plan trips and would throw questions at it and it would be able to go like gather a lot of information and say well particularly in cases where you're trying to figure out like you're going into new areas of research it is a great way to very quickly figure out that like this is a good path or this isn't a good path. I've used it a lot since we've started into this I think since this season of things like saying like if I want to build an app from scratch in this realm and this is sort of the target audience and this is what I want it to look like what are the best technologies for it and it will give me stuff like well here's here's an example here's the the popular technology or here's the technology it's gonna take more time to learn but it's actually gonna be a better long term. Those things are you know it can get you there but it really does come down to it's people talk about like I had a chat with chat GPT it really is I think where you need to go is it has to be about a chat it has to be about you learning how to use AI as much as AI giving you that information. Yeah I mean for those of us that have been around for a while it goes back to like the old dictation software Dragon naturally speak you have to spend time teaching the tool that you're using how to interact with you you're not just learning the tool but the tool is learning how to interact with you how to understand your sentence structure your grammar your reflections a lot of times it's interesting because when I first started with chat GPT I'm a bad speller it would more or less at early on just correct my spelling say did you mean this now it takes almost anything I throw at it and it returns some type of message now it may not be the correct response but it is a response based on the history of the interactions with the tool so if I were to log in anonymously through chat GPT so this is a good test for those of you that have a chat GPT account if you use chat GPT all the time log out clear your browser cache or go incognito log in to chat GPT or don't log in to check GPT but just go to chat GPT and start using it you're going to get different responses and different interactions you're gonna find that either it's going to send you down a different path or it's not gonna give you that kind of personal experience that you're used to from being logged in so these are just some of the things to be cautious of as you continue to learn and grow with the tools it takes time and you need to question it even if you're just starting out heck I would ask it almost once I started like building projects like Rob was just talking about like with code I started out with okay I want to build a web page so give me a web page I'm like is this the best way to build the web page yeah okay well all right maybe we'll do this I'm like all right and then it would prompt me well who's the web page for so if you are asking the right prompt it will prompt you back if it's not getting what it needs now with 5.0 I'm not sure if it still does that would have to go back and retest that but I would probably have to retest it incognito because I've already gone through that learning process with chat GPT so as you're learning the tool the tools learning you and it just takes time be patient with it but definitely spend time with it and work through it if you question it Google it and call it out if it's wrong and I think that's that's really I think the context is so critical and it goes back to it reminds me of Google food Billy people's ability to write searches in Google to like get useful information back and AI is so much like that is it really is and I I was saying about this today as I was like I was having a chat with chat GPT and I was actually getting it to do some work for me and I knew I knew I knew I knew I was like I really am gonna I like it was one of these that I asked it to do something and then I had to ask it for something else and I wanted to go back to the not the prior but to prior and I was like if I don't ask this right it's gonna go to one prior and not to prior and sure enough it did and it's really is it's like it's I has it like sometimes I like joke that it's like talking with the child that like has a very short attention span that you have to like make sure you're very clear on what you provide I do think as Michaels mentioned like the folders and stuff like that there are some things that it does allow it now to do a better job of keeping context but so often it still will like wander off and so it's things where I have to and it's where it actually does become worth it to have like the paid version so you can have a bigger question where you can say okay based on this I want you to do this and make it look like that and it's like sometimes that's a decent amount of stuff that you need to work with the cool thing in building a better developer and then I'm gonna let you jump into this is it the best way to get that kind of information is we dealing with small chunks and this goes back to something we have often said is like try to break your problem that you're solving down into small chunks that are very easy then to work with and maintain and this will also help you immensely when you get an AI you had a thought so once you jump I did so you made a comment that when working with check GPT if you're like two or three chats down and you need to go back and it doesn't go back to the right one a small tip I found with this is if you go copy the selection you want to go back to repase it but start out with reset the conversation to this chat paste your chat in sometimes it will reset everything to where you need to be and then you can continue your thread from there I found that to be more successful than just assuming it knows where to go back to that's actually had a conversation with the guy not too long ago that's a AI maven or whatever but that was something that was one of the things that he we talked about in our conversation is how it's it has a very limited memory and so there it's very easy five or six requests will say into it to suddenly lose stuff and it just like it's sort of reset and he said that was something that was he was struggling with as he was building out longer term conversations and that is to be able to do it and I think that's honestly and this is like a little pro tip or whatever because this is something I've been working with I think there's a point where what you need to do is if you're building like chat bots and stuff like that for longer term conversations is to have ways to actually freeze and update the context is very much this is very very much a Python thing but very much when in Python where you if you use pip if you use their library and there's other things are like this but there's a way to like say freeze like use what I have right now give me a script so then the future I can go back and then just and this is called requirements basically but I can use those requirements and I can suck all those in and say I want exactly this environment I want these libraries these versions that the cut that is literally the kind of stuff that we need with AI and if you're building something that is an evolving AI I think you need to do that but because AI has a it has a crap memory it's worse than mine and so what you need to do is you need to be able to do something where if you really are building it's isn't like the chat tools if you're building a chat tool you're going to need to be able to do a feedback loop into that to say okay now we have knowledge that we want to like maintain and we want to push that back into the AI so that now the AI is keeping that as part of it's like this is what I know it's just like if you go to like notebook LM where it's like load these documents in it's like well I need to have those documents there those are the things it's like marking knowledge that is key knowledge and honestly it's funny because Michael and I have had conversations in business stuff outside of developer in recent days weeks months where there are things like that that I realized I recognize now that is very much a part of the human conversation as well you'll be like hey this is a watershed moment or a milestone or something a marker that we want to say we have agreed on this this is what requirements documents are all about really we've agreed on this and so this is what we're going to use as our foundation moving forward it's literally no different when you're dealing with AI other than AI is a little more obnoxious and I know a lot of obnoxious people I'm never going to say is more obnoxious than people but I think that's what we we didn't embrace in this season is that we didn't feed back into it every time we asked it about a season about a topic for and give us a couple of things to talk about we didn't actually give it feedback on it so that is I think why we found over and over again we got roughly the same response it got I mean we we asked a question a little bit differently but generally speaking we got the same response because we were asking the same thing over and over yeah so one of the things as you went through that that kept coming to my mind is there are some tips and tricks you can do with AI and this may be more application desktop version of AI versus web but I have found as I go through the process of working with AI and resetting AI if it gives me a response there is that little checkbox that you can check under the message hey this is right but what I have found to be more useful is if it gives me hate this is like exactly what I want I respond save this in the memory this is a good response and for those are not I say okay this was a poor response reject this but do not follow this path and if I do that I get better results in the long run but if I don't do that and I just let it keep regurgitating stuff you take kind of that long way to get there like it takes longer to get to the solution than where you literally tell it hey bad bad bad bad go back here's a cookie that's right here's a cookie that's good but I have to interrupt because this is you and I have this conversation I'm like okay timeout stop right there let's let's I don't say snapshot it's basically let's snapshot that can we agree and we can move on I know I've done that with you I've done that with a lot of people I will say because you probably won't hear this my wife gets really annoyed when I do that sometimes because I will cut her off in the middle of us and it's gonna be like stop right there that's a point let's move on from there but that is a I'm gonna let you go because that is really a very important insight into how you find ways to move AI forward instead of get into those like loops or let it spin off into space exactly and worst case scenario if you find yourself five or six messages into a thread even after trying to get a corrected stop create a new chat delete do not archive delete the old chat say before you delete it say this was a bad message thread deleted from memory then delete it start a new thread nine out of ten times I find that the new thread starts in a different path usually closer to what I want and it gets me there quicker these are not things that happen overnight a lot of times I can now jump on the chat GPT probably within about five ten minutes I get enough of what I need from it to go do what I'm trying to do or I get enough of a correction to like a document an email or something that I can just jump in and go take it from there and move on I use chat GPT now more days as a good virtual assistant and a quick boilerplate like code generator like here just throw this out here and I can go from there sometimes it works well for more complex things but I'm still struggling with most of the chat bots to get something that will build me an enterprise solution and I only want to touch on this one other thing because I read an article on just just recently it seems like more sea level fortune 500 companies are actually now starting to shift away from AI a little bit because they're finding that it's costing more to get to that chat bot level of maybe this is right to something that is actually going to help the customer and that's honestly that's the wave we've seen in technology from forever is it's like everybody like oh this is now finally the silver bullet this is the thing that's going to solve all our problems and then everybody throws a lot of money at it the next thing we know it's like oh no it's not solving our problems I'm just gonna say if it's just throwing a lot of money at it you're not you're not paying attention throwing a lot of money is not it you've got to be there has to be wisdom in that you can throw good money after bad all day long AI is no different from all of the other silver bullets that we've seen all the way back to like when I was started out it was things like you eyes and object oriented and things like that actually AI I did a class I was just talking again guy today I had a class 35 years ago on AI it was basically expert systems in artificial intelligence and how to do that way and things have evolved massively since then but the interesting thing is which is actually a really cool little observation this guy came out of the gaming environment the gaming world and he had dealt with a lot of stuff there there is a lot of AI built into games the gaming world of so like think about it like if you're sitting there and you're fighting now granted the the MMORPGs and stuff all the massive online stuff it's just other people but when you're dealing with stuff and there is a intelligence in the application that is your opponent a lot of times we know we've seen it like you pick all of your games like the Dark Souls series of stuff there's there's patterns that they do the things that they struggle with usually is patterns with some sort of randomness to it or things like that but that's honestly what we know if you really look at it if you look at a fighter if you look at a football team if you look at I'm trying like a fencer or things like there are certain moves the ultimate fighter and stuff like that wrestlers there are there's moves that they use there's certain things that they do if you look at dancers you can look at across the line of what humans do there are patterns of what we do the things that are different for those that are the best is that we add randomness to it is that we try not to be predictable enough so that our our opponent doesn't know which path or which sequence we're taking at any given moment that's not that much different from what AI provides what we're trying to do of course is we want AI to be consistent however the bonus is that like AI is not consistent partially because it doesn't have the same set of information each time we ask that question now sometimes we'll ask the same question and it's close enough but it's still gonna be like there's I think there's percentages and stuff like that behind the scenes not getting too deep into the eye and I'm trying to simplify it for like not getting into the deep the deepness of what this is but honestly at the bottom line it it does work down to like you're rolling the dice you're throwing it you're asking a question you're rolling dice and says what things am I gonna like most used to color my opinion on this and we've seen that we've seen this in this season we've asked effectively the same question we've gotten slightly different results slightly different approach and it's based on time of day or whatever else is going on so I think that's very key for us to take I think that's a very key takeaway is AI is literally like another person is like throwing out there I think if you look at it that way but it's saying hey and Michael mentioned using as a virtual assistant I really think before I like kind of this is a rabbit trail I could go to I really see AI is replacing the VA I see so many things that I do right now that I'm like I can throw that to AI AI I'll do that and I could honestly this is your automation stuff is like I could build a bot that does that stuff for me and I talked to people that are like yeah the the people that are technical that are trying to leverage AI that are not just throwing random stuff at it say yeah I can spend a little bit of time and I can build some sort of an AI solution that gives me this report this dashboard this look this overview this summary this outline things that are like all they do is just get us to the solution a little bit faster and I'm gonna throw something to you before I like get to the end of this of this episode faster yeah so the last thing I want to touch on so as you were talking the spoilers for those that have not seen Superman the new Superman the scene with Lex Luthor where he's literally throwing out numbers he's basically controlling the Ultraman character in the movie through the fight so it's a real-time person interacting with another person for battle a lot of the best AI tools that are out there are still the best procedurally written applications basically if then do this if then do this it's building upon a procedural logic it is not AI it is a rules-based engine that is driving where your applications going in the interaction with your customers I have seen more often than not especially with testing tools and since Rob's been publishing all his applications I am probably going to get mine out there again soon I wrote a testing application ten years ago a testing framework needs to be updated a little bit but speaking of legacy I spun it up the other day ran it against a website and it worked so the short short answer is procedurally driven applications is essentially the current form of AI you have to ask certain questions in a certain way and AI is essentially building you a procedural response to those answers to give you a solution based on your input everything I see for was it SAP and all these marketing tools and business tracking tools they are literally just freaking integration systems that are basically procedurally driven based on rules not necessarily AI where based on customer rules we're going to give you this generated code or this generated report yes it's AI ish but it's still procedural it is not really thinking outside the box it's thinking within this rule engine to generate these responses these reports these applications so be very careful relying on AI as the be all and all answer to your problems and that's I guess that's about it was it's like it is it's an advisor like anything else it is not going to always be correct it is there it is wrong at times it makes mistakes and things like that just get used to that but that doesn't mean that it's not helpful it's just like everything else it's like everybody else you talk to there's going to be value and what they say and there's gonna be some stuff that is not very valuable and what they say there's a lot of value in shooting us an email at info at because we will use that we will like build that into this thing you will find the option the opportunity to say something to us and actually impact season topics episodes stuff that we do if we do blog episodes I mean maybe even training and stuff like that we do there's a lot that you can do if you look back we've done a lot over the years you go back especially to like some of our mentor classes some of things those are all out on the YouTube YouTube.com developer channel you'll find a lot of stuff where there were discussions that we had that ended up becoming entire products and things like that so feedback is awesome you can leave us feedback wherever you listen to podcasts you can check us out on YouTube you can check us out at X we are at developer nor Facebook there's a developer nor page Facebook page you can leave us comments anywhere on the developer nor.com site which is gonna be in the months ahead updating and enhancing and we're gonna do some cool new stuff I assume cool we'll see how cool it is when I get to that but I will just say I was very happy with some of the things we changed we made on the RV consulting site I very much expect to bring some of that stuff forward and do even more on the developer nor site plus some other side projects I've got going that will sort of feed into be like a little feedback loop for these kinds of things so we're not done we're not just giving you content we're gonna find better ways to bring you content and bring you the stuff that you need but it also helps to give us feedback so that we can you know know what we've done correctly and where we need to do it better that being said I just want to say once again as we are probably wrapping up this season I'm guessing that our next episode we will just dive right into whatever a new episode or a new season is which in recent seasons is sort of what we do is we're like I don't know we'll figure it out when we get there and it's turned out pretty cool so far we've gotten a lot of stuff we've learned a lot hopefully shared a lot that has been valuable to you appreciate you so much and the time that you've given us as always go out there and have yourself a great day a great way great week and we will come back to you next time with season 26 oh my god have a good one folks thank you for listening to building better developers to develop a newer podcast you can subscribe on Apple Podcasts Stitcher Amazon anywhere that you can find podcasts we are there and remember just a little bit of effort every day ends up adding into great momentum and great success