BIG MISTAKE? - AI Tools For Software Engineers

Transcript is auto-generated and may contain errors.

hey folks we're going to go to YouTube for a question SL comment um I'm at a Crossroads of allowing AI in my workflow since I am close to a beginner to building my own programs uh went through computer science degree program that was super math and Theory heavy this is pretty common um so I'm not afraid of understanding the uh the code AI could assist with writing but I honestly have such few works I honestly have such few works I think I should be avoiding it entirely until I have a few more projects under my name so how many more projects should I continue to not use AI since I'm very fresh with building thousand line coding projects or am I wasting my time and use all the tools at my disposal to get more things up to help me get a job soon it's

been two months out of graduation I'm mostly a Java guy right now by the way so uh big thanks to this person for writing in it's on YouTube it's the shaking duck these are very interesting user name so uh let's discuss that let me start getting on the road here though um and if you if you want your questions answered leave them in the comments uh otherwise if you want to write in anonymously um you can look for Dev leader on social media that's also my main YouTube channel I literally can't press this button enough to stop this damn beeping um and yeah if you write in then I'll keep you Anonymous of course if you write a c comment your username is public so I can't help you with that um okay but this is a I think this is a good topic of

conversation because I think a lot of people are kind of I don't know getting mixed messages or maybe it's not even mixed maybe it's very much just like Oh no you're bad um you're a beginner you're using AI you're bad I'm going to throw up if I hear that beep one more time please stop beeping you can only press the disable button so many times um which is dumb because it's literally in the disabled mode it's lit up saying that it's off and it will still beep um so yeah people I think are getting kind of told like hey if you're using AI it's automatically like for the wrong reason if you're a beginner and um I'm going to share my perspective on this I guess and if you're new to the channel then uh my goal is to kind of share my perspective my

experiences as a software engineering manager for the past almost 13 years and uh I think this is one where I'm going to try my best to also call out bias and share what could be different perspectives I think that it's important we try to do that when discussing anything um obviously you know the the easy thing to do is just say here's my opinion and like deal with it but uh I don't know I feel like we don't learn in grow as much if we're not trying to do a little bit more than that right so I like to try and challenge myself a little bit as I go through these exercises because that way I get something out of it too so when we talk about using AI tools and building projects and stuff I think there's a a couple of different things going

on right and I think it's important to try and figure out what you're actually trying to accomplish so when we have people uh let's say this is more common for beginners although I would say it's not personally I don't think it's exclusive to beginners but uh especially if you're seeking a job you know one of the things you keep hearing is hey you got to be building projects on the side that's really going to help give you some extra ammunition for your resume and you can have a portfolio and stuff I think these are all good things um and when we talk about this it's like as a beginner or for at any level right are you building projects because you were trying to Showcase something on a resume and what are you trying to Showcase with that project cuz I think these are important

questions to ask or are you just trying to put together projects purely so that you can learn and practice now these don't have to be mutually exclusive by any means because I think that it's great to be able to to put together a project put it on a resume and say this was for learning purposes there's nothing wrong with that um so I think that you just want to go into this kind of stuff understanding what your goal is okay so if your goal was like I'm just going to give you some very different examples if your goal was like I want to put together an app as fast as I can trying to get an MVP together trying to see if I can onboard users like I just want to you know kind of get something working as fast as possible then I would

say especially if you're a beginner like leveraging a ey to go Cobble together whatever you can that would work well I'm not saying when I say well I'm not saying it's going to build a good system but would it let you put together things faster the answer is probably yes okay but it's like important to understand what the goal is because if I said now if your goal was to go learn about how things work and then I gave you the same advice just get AI to spit out whatever code you can copy paste it get it compiling and and you're on with it like I don't think that's the best way to to learn so like I think you need to step back and understand what the goal is because I think this is why you hear so much conflicting opinion and perspective on

this stuff it's because people will go into these conversations and if you haven't aligned about what the goal is in the conversation then it's like yeah of course you can have completely different Tak takes on this stuff so um for this person right if your if your goal is like hey I want to get together stuff for my resume and and demonstrate that I was able to learn about different Technologies and stuff then like number one I think that's great right I I definitely encourage that and does that mean that you should or should not use AI I think that you still have options here right if if you're the kind of person who when you're talking with an llm you're like all that I'm doing is copying the code out of there putting it into my ID fixing up you know variable names and

stuff if that um and then I try to move on to the next thing if you're like I have a habit of doing that then I might say okay well if your goal is to try and learn more effectively either maybe step back from the llm a little bit or change your approach right like challenge yourself to go ask the llm questions every single time after it gives you some code right you have a tool in front of you that you can use in different ways and I'm not saying it's easy especially if you have a habit of of doing it but I'm just saying that you can use the tool in different ways um but you have to kind of understand how how you work and if you're like ah like it's it's almost too convenient and like I'm not going to like thanks

for the tip Nick but I'm not going to do that I'm just going to keep taking the code then I would say yeah like maybe take a step back then you know if it's too accessible for you and you find that it's hindering your learning take a step back right um even that's probably easier said than done but the point here is that you have to understand what your goal is if your goal is to learn and you have this tool in front of you that is very powerful can you think of way to leverage it to help learn if so then I would lean into that like my personal take on this would be like uh given how prevalent AI is and like the direction everything's going or that I I see it going I feel like it's almost silly for me to be

recommending people to not use AI right um I have this I don't even know if I want to draw this comparison um I'm going to do it and I realize that this is probably going to set me up for different different perspectives so let's talk for a brief comparison it's like um screens with children my wife and I don't have kids and we've talked about when we have kids you know there's this idea of like when do you allow your children to have screams and when we've talked about this at a high level and like she's not here to have the conversation with me and you know kind of defend her perspective on it but like I'm of the mindset that like what I don't want to do is say no screens until you're like 10 years old or something like that and the reason

I don't want to do that is because it's such an integral part of the direction that everything's going that I think that you're actually going to be hindering them by disallowing them from from leveraging it but I think the is that and what we both agree on cuz I think she's of the mindset like you should you should wait and I'm like I don't think so but what I don't want to have happen is like you put a screen in front of your kid just so that it's mindless activity like if I put a screen in front of my kid that screen is going to be loaded with things that are going to be learning tools right if you're going to be glued to a screen you're going to be learning so uh and I realize you know easier said than done I am not

a parent I'm not going through this but that's my kind of philosophy right so the parallel I'm trying to draw here is like I see that if kids are going to be technology enabled much earlier than I ever was and I was enabled like with technology pretty early right like when I was really little not quite like a toddler but when I was really little I had access to the family computer I was able to play You Know video games in DOS and not really know what I was doing um you know like just I had access to this stuff pretty early relatively speaking for that time but like kids these days they have digital devices like almost like as soon as their eyes can actually lock on to something so I think that if you're if everyone else around you is leveraging technology and

tools in this way and you are sort of preventing yourself from doing that I think that in my opinion that's not the right answer instead it's how do you understand how to leverage the technology effectively so instead of avoidance it's proper use and understanding that's that's my personal take right um but this is why I said I think if you understand how you operate and you're like I can't trust myself to do that or I don't have the willpower or or whatever it happens to be then like yeah maybe take a different approach but that's why I don't recommend just avoidance because I don't think that avoiding is like a sort of like a productive path forward it's it's kind of avoiding the problem altogether or the challenge all together so when people especially juniors are saying like should I be using AI I go

back to what your goal is right so I and I I would would imagine that for most people that are junior like your goal is actually learning right I know that it gets confusing because you're like okay I'm trying to get my resume together trying to get my portfolio together I keep hearing all this conflicting advice of like can't have another calculator app or another to-do app on my portfolio because no one's ever going to hire me it's not that like it's not that you can't do that right the goal is all about standing out right that's that's what it's going to come down to there's going to be tons of people submitting resumés in their applications if they all look the same because a to-do app looks the same then that's not going to stand out it doesn't mean that it was a bad

project or the wrong project but if the goal is to be able to stand out then that's probably not going to help much because it's a common one right one I just did a couple of resume review videos I recorded on my main channnel Dev leader at the time of recording this they're not posted now I had two resumés come in and uh each one of them had projects listed on them that I thought was very coincidental and quite convenient for me to be able to talk through because uh in many cases when people are submitting resumés it's a lot of like uh pretty front-end heavy web development nothing wrong with that right I'm not saying that's bad or whatever I'm just saying that's the pattern I'm seeing a lot of and not only that with resumé submitted like kind of what I'm seeing a

lot of patterns of online there's a lot of people that get into software development very heavy into the front end for web development now when you start seeing a bunch of resumés for people that are doing this you're going to see some really repetitive things right so in the examples it's like okay like at least people have put together projects awesome that's good news right that helps you stand out and differentiate from especially from the people that don't because I have already reviewed resumés where someone didn't have any extra side projects right my biggest feedback to them was like think about the people that you're competing against if they have side projects and you neither of you have work experience the people with side projects are very likely going to stand out from a resume perspective so these couple of resumés that came in they

had tools that they had built that were like uh one was like uh sort of like a file system Navigator with file compression and the other one uh was almost like not quite like a a file recovery tool but like something along the lines of that and I was saying like hey look like I used to work in a digital forensics company for 8 years if we take the example of having a lot of really similar resumés come in and then I have these two that have these projects where someone just because they were interested went and built something that is like I don't know adjacent to digital forensics like that's pretty handy for standing out right it's pretty handy for standing out for that particular job space if we're comparing it to something like a calculator app or a to-do app those don't really

stand out they could be great to list on a resume for learning to saying like hey I built this here's the Frameworks I learned here's why I did this here's what I learned awesome right that is valuable it's just that compared to something else that might be more job specific domain specific it doesn't stand out as much okay so if you're trying to get something on your resume that like is maybe something that's kind of eye-catching I don't think there's anything wrong if you wanted to leverage AI to pull some stuff together and say hey look I built this cool thing because it's Unique and that might stand out nothing wrong with that I do think that if you end up getting a job interview and then someone in the interview asks you about that and you can't explain any of it because you actually

have no idea what's going on you just kind of covered your eyes copy pasted then like that's not going to be a good situation to be in so I have this like recommendation and this has nothing to do specifically with AI tools but if you're building software and you're putting code into a repository it's going to be something you're building and or shipping you should probably understand what it's doing strong emphasis on the probably it's it's likely pretty dangerous if you don't know what that code is doing right this isn't unique to AI it's just that I makes it so much easier sorry we had to go around someone that was going under the speed limit you don't do that in the fast lane and there's no 360 cam on right now because it's like kind of misting outside and that would just be you

know 2 seconds into the video just be water droplets on the camera 360 cam is cool but like completely useless as soon as this a little bit damp outside so I mean I'm kind of saying the same thing in a bunch of different like the same way slightly different ways maybe but uh the reality is like I don't think there's anything wrong with using AI tools I wouldn't personally I wouldn't discourage beginners from using it um I think that this is you know a way that a lot of development is going there's a lot of uh you know support for AI tools for developers like there's a lot of investment in that direction I I don't see that slowing down okay so I just see it as a missed opportunity if you're not learning to use the tools effectively so my recommendation is leverage the

tools but also understand how you're leveraging those tools try to understand what your goals are and if those tools are compromising some of your goals and it's like it's it's okay to have an honest conversation with yourself right I'll give you I'll be very transparent okay like just to give you an example of this literally from last night so I've um I've said this multiple times on videos I uh am heavily using AI for my SQL queries okay and so I'm building this product called brand ghost for social media scheduling it's how I get all my social media content online and um there is a I found a bug because we just had the time change and we had done all of this work to accommodate that time change now the scheduler that runs like I said had a bug in it and even though

I had all of these tests to go check time zones and stuff like that and do time conversions there's a funny scenario where um given the window of time that we're looking over the um sort of the windowing function I'm using doesn't account for the fact that if you shift your time zone it might put you outside of that that filtering window but I I'll kind of explain in a little bit more detail here so I was like okay I have this function the time stuff is completely mind-bending and the reason that's the case for me is that historically the way I built literally everything to do with time dates and times is anchoring things to UTC right just anchor stuff to UTC it's like you know that's absolute zero everything you can do off sets from there you're good to go it doesn't work

with scheduling stuff and without getting into too much detail um the problem ends up being that people schedule things when you anchor it to UTC and then they go through a Daylight Savings Time shift as the user because their stuff was anchored to UTC it now looks like it's an hour off so the problem for me is that it screws up my brain when I'm trying to rationalize how this stuff works I know what's going on right I know that there's an issue but I'm like my brain just stops working and this is not not because I have a SQL query that's too complicated uh it's not because the code is too messy or something like that it's literally this is just a a thing I've talked about where when I have to map different I call them like coordinate systems it's more common for

me with geometry when I have to map different coordinate systems in this case it's time zone offsets it really messes up my brain it's like one of my biggest weaknesses in all of software development it screws me up so um I acknowledg that I have this problem I have a scenario and I said perfect I have this SQL query and I've been using chat gbt very effectively to uh you know to manipulate SQL queries and stuff especially if I want to optimize them or get ideas on on what to optimize it's been super helpful so I said great I'm going to give it this query I'm going to explain exactly what the problem is and it's going to help like I've seen this work time and time again so Chad gbt right away gives me this um suggestion and I'm going ah like that's not

going to work because the way that I schedule stuff like the example I gave it was too broad of a a Time range so I'm like it's not going to work it I've kind of confused it that I'm that I'm able to actually look over an entire week to schedule but really this thing runs on like a minute interval so I'm like ah it's not really fair I spent an hour I spent an hour with chat GPT trying to go back and forth it kept giving me queries and I'm like this is garbage it's like the same problem repeating I was so frustrated and I was like this is absolutely back firing if I would have just stopped from the beginning I could have sat there looked at the query and figured it out myself so after an hour that's exactly what I did I

stopped using chat GPT I looked at what was happening and you know the conclusion I came to it was the very first recommendation that Chad GPT actually gave me that I dismissed as being not going to work it was right the very first time and then I was the problem for the rest of it so this is just an example of me like you know going to use it going too hard into it being like it's going to work it's going to work and then being completely screwed over by it in the end it was me screwing over myself but you know we we have to acknowledge like like I probably should have stopped before wasting a full hour on that like this isn't working there was one point like 15 minutes in when I saw that it was repeating mistakes I was like maybe

I should prompt it to tell it to consider a completely different solution right but instead of doing that I should have just stepped back and said maybe this isn't the right tool for the job so I'm sharing this because it's like you know I've been programming for over 20 years and it's like I'm I'm making mistakes like this because I have this tool in front of me so I think that you just have to acknowledge like how you're using these tools right for me it's not a learning thing I'm not using chat GPT in this case to sorry that was people are real stupid sometimes when it comes to like on- wamps and stuff so someone just like came up and then like sat in my blind spot so I think they I think they were trying to like give some space to like let

me merge in but I don't know like just anyway um I think you have to understand your like your usage of of these tools understand where your weaknesses are but then have this honest conversation with yourself right so next time if I'm finding the like I have an llm that's repeating the mistakes maybe maybe my first thing is like I maybe I need to tell myself you haven't prompted this effectively and you might not know a good way to prompt it effectively that's why it's doing this take a step back you go think about the problem a little bit more you come up with a different perspective and if you so decide to go back to the llm like it's on you to be presenting that different perspective um but I think like that's the honest conversation I need to have with myself so if

you're I don't know like would love to hear if you have different thoughts on this right cuz I I think that the popular opinion right now is like I don't know if it's the popular opinion it's the one that I see a lot of on line right I have to call out the bias that I see um I see basically more tenured people in the software development industry saying you know like stop using AI it's making you stupid like developers that are getting started don't know what they're doing all they know how to do is copy paste and I'm like I don't number one I don't really believe that okay like I think that's a little bit of an exaggeration at least a little bit but number two like they have a really the new developers have this really powerful tool at their fingertips that

we never had so maybe it's just a matter of like having conversations around how to use these effectively for what our goals are right for for me as someone with a lot more programming experience than someone just getting started like I can almost get away with copying and pasting code at a chat GPT I won't but I can almost do it because at least when I go to paste it like if I have a couple of seconds to glance through it there's probably some stuff I can catch or at least question right for the most part because if I don't get it I am going to ask but if you're brand new to this stuff you might be saying like I don't even know what questions I should be asking I don't understand anything I'm seeing here but it gave gave me an answer so

I should just trust it right the I'm not saying that that makes me better than other people are better than Juniors it's just that my experience pushes me in that direction to be able to question things because that's what I've spent my entire career doing and if you haven't built up some of that experience to go question stuff like that then I don't blame people but I think that this is a skill you want to build up right you want to be ious you want to make sure you're taking time to understand should I go in this other lane I can't see far enough ahead we'll see what this person does and then I'll make a decision it's hard to see there's three lanes that kind of opens from one into three and most people that I'm behind are going to be in the far

right lane but it means that like this left lane is completely open if anyone like I don't know like 500 ft back wanted to turn left they could just fly up past everyone but then you have people that actually try to do this and they're going to cause an accident what are you doing buddy this is a good 360 cam time so what I was trying to say is if you wanted to go up and turn left so in the lane beside me you could have but then there was a Tesla that decides to do it except instead of going up into the left lane they just sat in between two lanes the entire way and they're not turning left so it was just really weird so I couldn't then pull out to try and go around anything cuz they were kind of in the

way but this is what happens when you don't know how to drive a vehicle um yeah I think moral of the story here is like the these like Ai and the llms and whatever we're integrating them with they're tools for us to use and I real some people seem to get triggered by this I keep saying like AI is just tools and no no no it's way more than a tool like I don't call it whatever the hell you want man it's it's more than a tool sure okay it's it's more than a tool um you have you have the option to use it like a tool it can help you like a tool it can also hinder you like the wrong tool so give it a different name if it makes you feel better but my point is that regardless you can have the

same conversation about things that aren't AI if you're misusing something because you're not understanding what your goal is or how to use that tool effectively or whatever not a tool I don't know then then you might find that you're ineffective I think that's what it comes down to personally so I'll wrap it up here because I'm just going to be sitting in traffic for the next couple minutes um so thanks for tuning in I have another question from the comments that I'll get to on the drive home I got the dev leader live stream tonight so uh if you're watching this it will already be after Monday but the good thing about Mondays is they happen every week so um if you are interested I do a live stream on my main YouTube channel at 7 pm Pacific and so that's on dev leader usually

what I will do is I will pick a topic from code commute uh generally it'll be the one that has the most either comments or the most views basically it had the most engagement I write a newsletter on that at weekly. deev leer.com me talking at you if you're like hey it would be fun to be able to ask questions and kind of have a discussion join the live stream because I will literally dedicate the entire live stream time just to asking questions in the chat I would much rather do that so um yeah if you're interested that's Mondays 700 p.m. Pacific on my main Channel Dev leader and um it's also streamed on like basically every platform so like Tik Tok Instagram LinkedIn x Facebook I don't know what I forgot um twitch kick it's there um so pick your favorite streaming platform look

for Dev leader if you're watching this on YouTube I I'm making assumptions that probably like watching stuff on YouTube so I hope to see you there thanks so much for watching and I will see you next time take care