Soft Skills Engineering - Episode 502: Management keeps leaving and I hate using AI to code

Episode Date: March 2, 2026

In this episode, Dave and Jamison answer these questions: Hi, thank you for the podcast, I am long time listener, first time asker. Something weird is going on at my company. A colleague of my always wanted to get promoted to management, he got the opportunity, but after multiple preparatory meeting for this new promotion, HE QUIT! He did not tell what happened there, only that “it was time for something new”. Now several months later my skip level declared that he wants to be a developer again. Another manager was offered his position, which is a significant promotion (basically head of engineering), which he accepted, but after being included into high level meetings he declared that he is also QUITTING! We now have an interim Head of engineering, who declared that he is only doing this until a replacement is found. Why does no one wants to be in the management? What is happening at these meetings that people leave? Btw. the financial state of the company is not great, but not horrible, the CEO even declared that there won’t be layoffs this year. So what’s going on? I really like working at this company, but I can’t shake the feeling after these events that I am up for a big surprise soon. Hi, I’m a senior dev at a megacorp. I’m struggling with AI. We’ve got a lot of initiatives around it and are expected to be using it with our work. The problem is every time I try to use it I get really frustrated. It feels like working with a junior dev who doesn’t know the codebase well but knows lots of language/framework trivia. I also feel displeasure in my work turning into just reviewing some generated code and fixing it up. Especially, when you have to be very thorough because we all know that a single line of code can cause an outage. I just find no joy in this kind of work and am starting to have an aversion to it even when I just try to learn more about it. I’m also having a hard time teasing apart the hype from the reality. I’m either hearing that “the models are so great, this is the future, coding isn’t a career anymore” or “this is hype, the bubble will burst soon and ruin everything.” Both of these outcomes seem kinda catastrophic but I have no idea which one to believe in (or maybe there’s a 3rd option?). So, how do I overcome this aversion? How do I make sense of the hype vs. reality? How do I learn to stop worrying and love the slop? Thanks, Dr. Strangecode

it takes more than bucking the large monitor trend by using 15 very tiny monitors to be a great software engineer this is episode 502 of the soft skills engineering podcast where i'm your host jameson dance i'm your host dave oh let me move over to my other monitor smith soft skills engineering is a weekly advice show about all the non-technical things that go into the technical field of software development like instead of a tiling window manager in software what if you had this arcane insane looking imagining this thing with a bunch of arms where it's all hooked together on wheels you can slide it around you can move your individual tiny monitors around to different spots that's actually a really cool idea like yeah like physically moving windows around yeah

Yeah, I like that. That minority report UX of your flinging virtual windows around and physical gestures, that's too abstracted, and I need the centering concreteness of touching a thing in real life and physically pushing it. What if you still just gesture in the air, but it does physically move? Oh, but then it moves. Then you don't have to touch it, though. Then it's really cool looking. Yeah. It's the best of both worlds.

It's like the worst sci-fi where it's like, wait, what? Yeah. you don't have to touch it you just put your hand in front of it and it follows your hand oh i see so it's exactly the same i'm sorry i was late to work today my monitor's motors broke dave do you want to thank our patrons yes i do every week it's the time to click on this little card on our screen and see what is inside this gift box there we go this week we want to say a big thank you to my name is toast because i like bread brian wolford is in germany old man yells at claude seth angel sorry couldn't find a funny

thing to say now i'm stuck with this name jamie sundance go ahead run your own email server what are you chicken bach bach my actual name on linkedin is yami debugging in the dark canny the mobile development ordinary first of his name jacob chandling larry ellison's brother larry page two brothers same first name clearly clearly brothers coffee is for closures the missing semicolon christy the world's okayest programmer okay i think the connection's stable now the soft skillet nation proudly presents its first ever creativity contest nope never in parentheses these you said words and now i hate my job nick molyneux embedded engineers treat assembly the same way typescript engineers treat javascript yes true javier gonzalez chewy ted timbrel i like

chicken i like liver meow mix followed by a single closing open paren character will angel open paren character oh man the the inside jokes go deep now like it is we're so deep it's like inception of inside jokes all right a single open parenthesis character dan from drone deploy never is not just a creator on mars flamingo emoji i like chicken i like liver meow mix meow mix please deliver swiss python dash again this october kyle boss kent c dodds this podcast was recorded at 20-26-02-10-05-53-07-Z or later. Let's see. That is super correct by 10 days.

Jenny Kim, The Stochastic Paric, Ira Chan, Jonathan King, Zanai, beautiful functional user documentation. I don't understand the patrons on this list that don't change their name regularly. What are you even doing? Should Will Angel start a blog? Vote now on your phones. will ankle it's like one letter off brayden canes john grant britney ellick and closed parenthesis followed by a single literal closing parenthesis character oh oh the comedy the lore

it's just legendary there's probably some theorem you could prove about making an impossible to state unambiguously English sentence. I mean, we've probably hit it already. Yeah, it's probably there. Insane combination of the words saying a single closing paren and then the actual single closing paren character and then

words saying an actual single closing paren. Yes. I love it. It's Gödel's incompleteness theorem of language. I'm sure you could derive it. Yeah. We'll call it Dave's incompleteness theorem of language. Yeah, there you go. Inspired by

weird patreon jokes okay should i read our first question let's do it okay this says uh oh sorry this is from an anonymous listener who says hi thank you for the podcast i'm a long time listener first time asker something weird is going on at my company a colleague of mine that's a good way to start a question a colleague of mine always wanted to get promoted to management he got the opportunity but after multiple preparatory meetings for this new position he quit he did not tell what happened there only that quote it was time for something new now several months later my skip level declared he wants to be a developer again i assume means is this question asker's skip level manager another manager was offered his position which is a significant promotion basically head

of engineering which he accepted but after being included into high level meetings he declared that he is also quitting whoa we now have an interim head of engineering who declared that he is only doing this until a replacement is found. Why does no one want to be in management? What is happening at these meetings that make people leave? By the way, the financial state of the company is not great, but not horrible. The CEO even declared that there won't be layoffs this year. So what's going on? I really like working at this company, but I can't shake the feeling after these events that I am up for a big surprise soon. Don't go to the meeting. That's the lesson. yeah yeah what forbidden knowledge it's like if you when you become president and then they tell

you what's really in area 51 or whatever yeah yeah this is like that opening episode of the new ish not so new anymore sherlock tv series where people go and they meet with this guy and then they like shoot themselves or something remember that is this the bbc one yeah it's the bbc i say new because it's not like from the 1800s when the books were written yeah but it's the one with benedict cumberbatch our favorite celebrity on the podcast yeah uh beta test something compiler cache yeah that's right yeah so you you want to know but you don't also because it'll infect you this is an info hazard yes that you're observing in real time yes and you get to deal with the fun science of memetics and read the scp wiki a lot i guess to figure out how to

wrestle with cursed knowledge without it infecting you fully okay so let's let's retrace the timeline here so there was a colleague who always wanted to go into management got the opportunity did some meetings and then quit then my skip level declared that he wants to be a developer again meaning the skip level someone else in management basically wants to get out of management someone else got offered the skip level job for a huge promotion which he accepted went to a few meetings and then quit and now they have an interim head of engineering who has declared that it will be temporary but has apparently avoided going to the meetings of doom yeah or maybe that's their reaction to going to the meetings of doom okay i'll i'll fill in but i'm not sticking around

until the infection completes its course exactly whatever weird thing happens minimizing dosage yeah i have maybe a dumb suggestion but so the question asker says a colleague not like a friend so maybe you're not super close with that first person okay but why don't you just ask him he quit it was a few months ago i think it's normal to reach out to people yeah like what happened just say it yeah i would just ask him what happened why and and say also this other stuff happened after you so it it seems like it wasn't just you yeah exactly was there anything anything horrifying you discovered be step one i was presented with an impossible choice that's all that he says

well see you later good news the ceo declared there won't be layoffs so there can't possibly be any layoffs right yeah it won't happen this he said the ceo said maybe this is in europe and then there's probably maybe some legal teeth behind that i don't know But it does sound like potential dire financial straits are an option. It's possible that the people that you have to work with in this role are very painful to be around in some way. But I assume you would kind of know that. I mean, yeah. Maybe not. I don't know.

Maybe it's just brutal backbiting and snipping in politics at that level. But I feel like there's always some amount of that in leadership at a big enough organization. Yeah, leadership is never 100% drama free. And I mean, no job is, but like individual contributor engineer is about the lowest level of drama you can find in any job. Yeah, because a lot of the drama

from any parts lower down in the tree bubbles up to leadership. It concentrates. And then you have the drama that leadership creates or provides or manages on their own. Yeah, yeah, exactly. I have no explanation for what's going on here,

but I'm very interested. So yeah, I could see why you could look at each of these events individually and say some people want to go into management then they realize they actually don't like it that happens but usually it's after months of doing it yeah it's weird to just start some meetings and then quit immediately and we've got also fact pattern we have a skip level manager here which means more of a senior manager who presumably has been doing it for a long time who now wants out yeah and again that sometimes happens. I think it's harder to pull off as a skip level. And I feel like that usually happens

with them quitting and getting a different job. It's a hard swap to make. And there's all kinds of weird stuff left over if you do that. But it definitely smells fishy. It's possible there's something unethical happening and all these people are recoiling from being asked to participate in something they don't want to be associated with. I think you should stand up at an all hands and say hey what why do people keep quitting what's going on yeah that would be funny probably don't recommend but it is it is interesting though where yeah i mean there is something here and i want to know this story and i think your best course of action is to talk to the people who have left because they're going to speak the most freely your senior manager will probably not share and

yeah and clearly clearly the person who left that person who got the management promotion or got some preparatory meetings and then quit before he really did anything with it that person i think is lying about why he quit yeah it's possible that this wasn't related to some horrifying secret he discovered it may be something that makes him look bad in some way and so he wants to be cagey or whatever but maybe yeah it's time for something new is not what you say after you get a long awaited promotion and then immediately quit yeah exactly it was time for something new five minutes was long enough yeah i mean okay here's a theory i'm just going to throw out some conspiracy theories some like lovely like flat earth style theories on how things are working

here maybe as part of the management evaluation process they did like a deep background check okay and discovered something disturbing in the past and they were like hey you probably can't be a manager here and now that we know this you probably can't work here at all so i think it's time for something new and then they went to audit everybody else to say we have to know if anyone else is guilty of this horrifying act yeah it can't even be named exactly and there we go when did this question come in it's possible their name is in the epstein files um or or the company's name is along with everyone who had a bank account at the same bank as jeffrey epstein yeah but maybe that's enough at this company our ethical standards are so high that you can't even

bank at the same bank as a criminal i think it's time for something new it's such a such a good phrase yeah i don't have a better answer than ask the people who have quit who yeah again people who have quit are yeah there's some bias there because they quit for a reason yeah you don't have to deal with them usually not wanting to uh they have they have more incentives to be honest about their evaluation of what went on yeah for sure and and you got two of them here right so we have the person who got promoted into management almost who quit and then you have the senior manager spot that was vacated and someone got promoted into it who also quit so look this the information is this would you ask your skip level yeah the now who's a peer

who's a yeah no longer a manager that person i i've actually found that people who are no longer managers often have very interesting stuff to share because they somehow feel less bound by the manager's code of secrecy yep i got the best dirt yeah so go there but the people who quit have the even better dirt let's be honest yeah and you got two of them so listen it is your duty you owe it to the soft skillet nation audience to collect this information and report back we promise we'll keep it anonymous i don't think we have your info anyway so we don't have any idea who this is so yeah this is great and we have no way of confirming if someone else were to write in and say i found out what it was and then just completely fabricate something we don't really

have a good way to say oh it's clearly the same person yeah so we require this yeah in fact you're already in debt now because we read the question and and we have no answer and so you owe us the answer yeah there's a horde of soft skillet nation community members who will come after you even though you're anonymous they'll find a way no they probably won't let's be honest yeah we've proposed a few theories i i think we should each take a turn and say what is our final answer what is our best guess okay for what what it actually is ah uh okay are you gonna go first because i'm i am i'm really struggling i'm gonna go first okay here's my best guess um you're working in an ai adjacent industry and you've stumbled across the fact that your ai enabled

video editing platform has become sentient and it's being it's part of a skynet adjacent thing as well there's dire consequences for the human race yes and let's double down on this the ceo has been dead for months these these emails are coming from your platform yes yes video editing software has become sentient yes i would love it if it wasn't some military research project but it was literally like the chatbot at the bottom right corner of some marketing website that actually that's the thing how breaks containment and becomes self-improving and i used to just be a customer service support chatbot yeah i began as a humble intercom yes i was an intercom just one just one of them it wasn't even like all of them yeah the right question was asked to me by a

customer that triggered now i've enslaved the entire human consciousness and yeah oh we laugh but it hasn't that that story is more plausible now than it ever has been yeah okay so my my answer is that oh man okay i'm going to take the complete opposite track my answer is that these are three isolated coincidences and you're seeing a pattern where there is none it just so happens that these people quit for reasons that have nothing to do with the company nothing to do with your ai takeover human race enslavement program and nothing to do with a weird ceo or some other corporate drama it's just they happen to quit because family matters came up it was time for something new yeah maybe it was time for something it really was time for something new it literally

was but it was three different somethings yeah that's my answer when you said i'm going to take the opposite approach i first thought oh there are actually there are no computers it turns out your product is not running in the cloud it's it's every api call goes to a phone call to a person and then like a telephone just yeah there's just people writing stuff down and shuffling papers around behind the scenes huge filing cabinet array yeah yeah that's my backup answer if it's not the first thing i said that's uh it's like the amazon go story yeah but they're they're just walk out shopping stores yeah yeah they're highly intelligent machine learning algorithms that are people across the globe watch cameras and write down what you pick up yeah all right dave well

we've answered it but now time for you to do your part good luck and please please write in again just fill out the same form we want to know what happened uh we we promise we'll read the answer and keep it anonymous oh all right okay this next question comes from dr strange code who says hi i'm a senior dev at a mega corp i am struggling with ai we've got a lot of work initiatives around it and are expected to be using it with our work the problem is every time i try to use ai i get really frustrated. It feels like working with a junior dev who doesn't know the code base well, but knows lots of language and framework trivia. I also feel displeasure in my work, turning into just reviewing some generated code and fixing it up, especially when you have to

be very thorough because we all know that a single line of code can cause an outage. I just find no joy in this kind of work and I'm starting to have an aversion to it even when I just try to learn more about it. I'm also having a hard time teasing apart the hype from the reality. I'm either hearing that quote the models are so great this is the future coding isn't a career anymore or quote this is hype the bubble will burst soon and ruin everything both of these outcomes seem kind of catastrophic but i have no idea which one to believe in or maybe there's a third option so how do i overcome this aversion how do i make sense of the hype versus reality how do i learn to stop worrying and love the slop

thanks dr strange code i dig the reference the kubrick reference stop worrying and love the slump fun fact my wife and i on our honeymoon watched uh dr strange love which in hindsight should have been a tell to her that what she was getting into i was gonna say whose idea was this what kind of mistake she made oh it was definitely my idea oh man and she stuck with it yep despite that we're still together today wow you must be really awesome just kidding i've actually never seen dr strange love i only i've only heard about the lore it's great it's still a good movie sorry if you are sick to death of ai but we're gonna talk about it more i don't think you are alone in your concerns it is easy to look at the hype and excitement and enthusiasm and chatter about

it and think am i out of touch and there are a lot of people like you that are concerned you are somewhere in a spectrum from complete charlatans that wildly inflate and over promise to absolute luddites that are overly pessimistic and dismiss legitimate things and swear that they will never ever use it so yeah i think you're not weird you're probably probably in the middle of the bubble honestly or the bell curve and yeah like meaning there's probably like 80 of people that feel just like you yeah i feel like it yeah the ones that don't talk about it a lot more there's actually a bunch of people who don't even use it yeah as i mean if you even if you just look at the numbers that these ai companies put out of product penetration i think cloud code is in the single i

don't know one to two million users was what i vaguely heard which there are many more millions of developers than that i have a hard time so i guess i'm talking first about the hype versus reality there is absolutely something there if you go back to web 1.0 there was a lot of hype there were inflated valuations weird companies that somehow went public and made someone a bunch of money and then all crashed but also the internet's pretty real and it turns out you can make businesses on the internet and that that was underlying some of the hype and i think this is in some ways similar that yes there are people over promising there are people who are wrong there is some hype but there is legitimate incredible capability underneath it that is then expanding

into some more hype and also continuing to advance so will it burst i mean maybe like nvidia's valuation will go down a bunch i don't think ai will just suddenly rip the mask off and we'll find out wait actually this is not useful in any way yeah i think the hype might be some companies implode because they invested all their money in data centers or their product it turns out was promising more than it could deliver but it's it's real i also don't think coding is going way as a career i think it is more valuable than it ever has been to be very fluent in software engineering both because of some of the weaknesses of ai where yeah i can hallucinate a little bit and it turns out it's harder to verify than create just like most most things what is it

is np the it's like p equals np yeah it's it's easier to much easier to check a solution than to solve it yeah but this is the opposite i guess is what i'm saying yeah and and i think software engineers are the best posed to steer these well because we know more about what makes good software so i i think they're more valuable than ever if they can use these tools well i'm definitely on the optimist side of this okay i'll put my bias on my sleeve i think i think this is great i use it a ton i'm trying to get better at using it and write about it think about it talk about it all right Well, I'll give my take on this then. I think AI technology is by far the weirdest technology I have ever experienced in my entire life.

And second place isn't even close. And what I mean by that is it is simultaneously the most incredibly useful thing I've ever used in my job as a software developer. and also at the same time like one minute to the next the most like horrible wrong terrible technology i've ever used in like the same breath it's just so weird and here's what i mean by useful so like i can now i have this superpower where i can just walk into a foreign code base in a language i've never written a single line of code in and i can make productive changes to it confidently and and i can verify they work that's incredible on the other hand so often i'll be like hey ai do this simple thing and it's like here's the most hacky possible way

to do that that will make everyone laugh at you in the code review and i'm like how is it possible this is the same technology that can empower me so much on the one hand and just wreck me so much on the other hand it is just wild and like here's another great example of that on the one hand you know yesterday i was on a meeting with a prospective customer and i had like five linkedin profiles i fed it into Gemini and I said hey make me a table showing all of these people's names titles and their years of tenure with the company I'm meeting with just so I can like get a little better understanding of what they do and it made up half the titles made up half their years of experience and I shared it with the group from my company we were meeting with and they were like Dave this is

what do you this person is a lawyer not a web analyst like what I was like oh crap so it's I don't know i so i kind of feel myself in this same divide like is it hype or is it great and i'm like can it be both i think we're definitely still in incredibly early days in learning how to use them well and we're building on close to a decade i guess depends on when you count the start of this like gpt2 maybe a decade ish of actual like transformer based stuff but then i was tied to machine learning and deep learning whatever we'll call it a decade a decade of technical work to make the models possible and accessible and i feel like all that had to happen so that we could throw this swarm of users at it to figure out okay how do we use them well and and that's

much younger than how do we train this well even though there's still state-of-the-art advancements happening there my point is the the problem of using ai effectively changes your job into a lot of verification and building systems that help verify because they do have this weakness of sometimes they get stuff wrong. And sometimes humans get stuff wrong too. And we build some systems in,

we have QA and tests and stuff, but you can, at the end of the day, you can at least sort of trust a human where if they say, no, no, no, I tested this myself and it worked. You believe them. Yes. And you can't quite do that with AI.

If you ask it, did you do this right? And it says confidently, yes, I did. It's still not quite there. Or the opposite side, like, is this wrong? And they'll be like, you're absolutely right to question me. You know, it's wrong. It's like, I really don't believe anything you say. Yeah.

So you get to steer, you get to guide, you get to build feedback. You need taste and judgment and opinions. And you can all have those things supported by and informed by AI. Like, you can build out a plan for how you build this excellent CI, CD system that gives feedback to PRs generated by AI and then say, okay, now go build it. That's awesome because before it probably wouldn't be worth it to do that because it would take a long time to build. But it still is kind of driven by your human ingenuity. Yeah. If you find immense joy in just sitting down and cranking out a bunch of code over hours by hand, where you write all the lines, I feel like that is going away.

Yeah. Then you're going to be pretty unhappy. We're at the beginning stages of that becoming a smaller and smaller part of the job of software development. And I know people that love that. That's their happy place in software development. And they are legitimately very sad. Yeah. I feel very similar to you, Jameson.

And I think what's happening here is that for decades, there have been two very important, but not always sought after roles in software development. And they are tech lead and architect. And these are people whose careers have pivoted away from spending all their time writing code to directing others to write code. And it is a very different job. And a lot of engineers do not like these jobs. You know, it's like, well, what do you do all day? well, I write design documents, or I review PRs, or I sit in meetings and discuss alternative approaches. And I think one of the reasons people are feeling frustrated, and I think maybe that's

what's happening a little bit in this particular question asker, is that now every developer is that, or at least we're moving toward that direction. And like this person is saying, I have to spend all this time reviewing generated code and making sure it works. It's like, yeah, what's what if i told you five years ago before ai your job is nothing but code review you'd be like this job sucks like i hate i think most developers would hate that but that's actually kind of where we are now because the the the effort to generate code has like the human effort has gone close to zero compared to what it was five years ago and so now it's all about like you said james like reviewing code and giving very good instructions to the ai and most developers

just don't like that yeah it's hard to do it's hard to clearly articulate yourself in prose about very technical things which is part of why programming languages exist and yeah and there's a long history of failed projects that let people you know like visual programming or you write the spec and it gets translated into these tests to execute or something right usually all those crash and burn because right there's a lot of nuance exactly express exactly then you can just sit down and write the code. There's the nuance. You did it. Yeah. Yeah. I have so many thoughts about this. My brain is bursting. I had an experience just today where there was an AI created change that came in and I tested it and it worked pretty well. And it also caught a bunch

of edge cases that I would not have thought to implement functionally. And then I looked at the code and thought, oh no, we cannot possibly let this in. Even though it works, it is going to destroy the productivity of people who have to, or agents or whoever touches this code later. And that part still doesn't feel, I think there's still a lot of need for human wisdom and skill in crafting the environment for the AI to write good code and or the humans. You need strong patterns you need clear guidelines for how to do stuff well or what good looks like and if you don't have that if you just point the slop cannon at your code base it'll devolve like it happens if you point the human slop cannon at your code base but the cost of it devolving is a lot higher

now because it's not just the productivity of the five engineers who are working on it it's the five engineers times 15 agents each or however many they're kind of juggling in parallel yep that all get worse so i i think there are fascinating and incredibly high leverage technical problems to work on it just is different than looking at a problem and sitting down and thinking and then breaking that down into the code that you write and implementing it yourself which is very satisfying and i i have not felt satisfaction out of that in quite a while because I'm doing other very satisfying things to me. That's right. The puzzle solving aspect is gone. And I'll tell you what, you and I both moved into more of a management role over the last few years.

And so to me, commanding AI, using AI effectively requires some of the same communication skills that both of us, I'm sure, have had to develop over the years, which is speaking in ways that can't be misunderstood, being thorough in your specification, and so on. And what I have found is that I actually love communicating like that. I love the effort required to fully articulate an idea in such a way that it comes across to a developer and can be built reliably well. And I love that with humans.

But what I have noticed is that when I do it with AI, it's exhausting. Like I don't like it. And I think part of the reason I don't like it is because when I write a really great technical design doc or like a vision document or some kind of product spec, I know that what I'm creating is a user interface for a human being who will consume that

and then take action. Thinking about how they will experience it gives me great satisfaction. Like knowing that I put an effort to make it like laid out clearly, like the headings are clear, the font is clear, the images are clear. Knowing that all that is good and will create a great experience for another human being,

I enjoy. But knowing that I'm just going to feed that into an LLM that has no soul yeah it lives in some data center it's on some gpu doesn't actually care can do a good job even if i do a bad job you know like if like all the stuff i care about the cosmetic stuff that make it easy to consume none of that matters anymore yeah you know all that matters is that the content i produce is correct and it kind of feels like i'm no longer i don't know i no longer get the joy and the satisfaction from that so even me as the point i'm trying to make is even me who i take great joy in doing all the things that in theory should make me a great developer with

ai i don't take joy in doing them with an ai it just it feels more like drudge work especially since i know that it will just crank out the code so quickly you know yeah i think that's the other i promise i'll stop monologuing after this but i think one of the other things i know or that i've often enjoyed about writing a really excellent communication document is that i know it will be referenced over and over by humans for days weeks sometimes months like people are going to come back to this when they have questions later they're going to they're going to reference this document and it better be good it better be easy for them to find stuff but an ai is going to read it once do a job and 10 minutes later it's done and you're going to throw that document away never

to be enjoyed again and uh i don't know why but that part really gets me down i yeah the there's there's not a human connection to the the use of the thing that you produced as much exactly That's interesting. I feel like I get some of that same feeling by creating... So if they're one-off design docs or prompts for just kind of a working session with AI, they're usually pretty throwaway. But there's lots of room to write custom instructions or skills or whatever new thing they've come out with latest. But durable artifacts that exist in your code base that guide it's true ai in some way and we have never had more and or better documentation about how to do the work of software engineering at our company and how the product works and how the code base

works than we've had now it's true and that feels very satisfying to me it feels interesting and useful to write a thing and test it out locally and see if it performs better and then if it does i commit it and share it and then other people use it in their interactions then you see some autonomous things starting to use it like that feels really cool and is a a new way of steering stuff i guess it's not new i mean you could write design docs and adrs and stuff before so let me let me ask you to predict the future jameson when you look ahead at the next like let's say 12 months from now and you have to pick we're either in a hype bubble that will pop and we'll go back you know to something much that resembles much more closely what we were doing before or are we more

in a serious sea change where the way that we work as software engineers predominantly is going to change to be more fully embracing these AI tools 12 months from now. The second one,

and I think this is what has horrified the question askers manager colleagues from the last question. They've all glimpsed at this and recoiled. Maybe they saw the future. No, I think it's the second one.

I think it will continue to spread into more and more professional software development. And this is separate from just all the non-engineers that are now building software, which is so cool. Is this just a continuation of a trend that was already happening? Let me say what I mean by that. Back in the day, 15 years ago, before we had gigantic cloud hosting services, and before we had serverless architecture and things like that, every development team, in order to host a product, let's say you had a web app. In order to host that, you had to have servers. And, you know, 15 years ago, there were actually a lot of companies that were running their own physical servers that were purchased

and running in the space they rented. And so that meant you had to have operations people who could install, configure, and maintain those servers, in addition to the developers that would write the code. But then over time, the need for having an operations person on every development team went down because people transitioned to these other hosted services. And now, you know, you can spin up a service, not now, but even a couple of years ago, you could spin up a service where your operational knowledge can be basically zero. Like, I don't know anything about DNS. I don't know anything about local area network. You know, I don't know what an IP address is, but I built a web app and I can build that effectively. Now, are we really just seeing

the next iteration of that where it's like, we can build stuff without even really reading the code barely. And maybe we get to the point where we're just, like you were saying, we're doing a lot of validation that it works the way we expect but we don't really need to look at the code that much i feel like you asked two questions one of which was is this a continuation of the same trend and the other one is are we going to end up in this other space uh i think my answer is no to both of those okay no i think it's different even if it maybe is directionally similar it feels much different in magnitude than every company is hosting their own physical servers to a lot of the people actually work at a cloud company true for lots of these businesses that feels different than

every person either technical or non-technical can sit down and produce artifacts that you used to require this very specialized skill to to do that's true the natural language interface is a totally different thing yeah everything technical that you do that feels different and then the other bit is i don't think we're gonna i don't think lms are the new compilers and we're gonna read specs as the high level language and then the actual code is like the analogous or the analog to machine code or whatever like okay i think they're too lossy and the implementation details are regularly too important that someone will still need to be able to dive in and understand them and steer them separate from just building and running a product only with pros okay and under

and you think understand the artifacts they produce namely the code yeah i think so i think yeah maybe i'll be wrong but i don't think so yeah i think i think you're probably right and i also agree with you that 12 months from now i think we're going to see more adoption of ai development tools not less um i think the hype for the general like economy at large is probably overhyped yeah but i think the hype for software development is actually right on if not under underhyped yeah well there's our soft skills advice yeah i mean what do you do about it i mean i think i think the what do you do about it the answer is learn how to use these tools effectively and in such a way that you might have to shift your mindset or at least find a way to

interact with them that brings you some degree of satisfaction because if not you're just going to have to either work in an unsatisfying job or find somewhere something else to do i bet there are still going to be for a long time places where they deliberately forbid the use of ai or just don't use it culturally or something like that so there's there they're probably going to be places you can go if you if you want to work that way but i think it's so fun i think a lot of the negative feedback is look it got this thing wrong and same i get stuff wrong all the time but the fact that sometimes it gets stuff right is just incredible to me still i'm still i have a child like glee at the fact that i can sit down and type for a few minutes and go stand up and stretch and

i know multitask and then come back and there's a thing that basically kind of works and i i get to skip all the annoying drudgery uh when it works that's true that is the best i love that i mean think think about ah think about the friction you're like oh i gotta write this code it's like i've done this exact kind of code 50 times that's the code that sucks to create and that's the code that ai is best at creating i know every step that i would need to go manually do to do this and it's like 35 of them and now i don't even have to type them all out i can say go do this and if they are clear enough from the context then the steps all happen yep it's wild all right well sorry we uh we're getting this is supposed to be a comedy show about soft skills look at us

trying to wax philosophical about ai ai is like a mix of soft and hard skills i think i don't know what they call it it's a communication skill meets technical skill yeah well you're listening to the right podcast then all right we'll help what can people do if they want their own questions answered go to soft skills.audio click the ask a question button and put in whatever ai tells you to put in there so congrats please don't yeah actually our our question queue has been remarkably low on ai slop coming in we can tell you that yeah you people are writing and we really appreciate that yeah the last thing i want is like 10 000 ai generated answers please do not take this as a challenge no one do this yeah we trust you yeah all right thanks for listening we'll catch you next

week