#I think AI will kill this industry for graduates in 5 years

190 messages · Page 1 of 1 (latest)

languid umbra
#

Reddit won't let me post this so ima just drop it here. thanks people
I want to start with a couple of disclaimers:

  • When I say kill the industry I don't mean fully automated. There will always be a few elites overseeing what the AI spits out.

  • I'm just a first CS student who's extremely worried about the future career prospects of this field. Looking at similar Reddit posts, neither side is making sense. If you believe AI will in-fact massively reduce job opportunities and paycheck cuts, you are "dooming" and if you think otherwise, you are "coping"

  • I know that at the end of the day, no one can predict the future. Looking at the trends in the past decade, the field's been on an ongoing decline (minus the COVID era but we are seeing all the layoffs that boost resulted in)

#

Logically, I believe we are going to see a massive reduction in opportunities and compensation. As said before, I believe AI is slowly going to automate a huge chunk of the development process. This coupled with a declining trend in the industry, layoffs, oversaturation, and offshoring being an existing issue, really worries me as a first-year CS student that by the time I graduate I've wasted 4 years of my life studying the hell out of something that is nowhere near an ideal career field with this level of saturation and demand for only the absolute best who grind leetcode in their vacation time.
I also like to mention that I don't go to a top university. It seems like in the future only the top dogs are gonna be able to have a good stable career in this field. I am in a Bachelor of Computer Science degree program, but not from any of the big-name unis. I'm also not looking at FAANG. I just want a fairly stable healthy work routine. While I do enjoy programming and I'm pretty good at it compared to my peers on the same level, I can't say the seeming stability and good work-life balance and compensation weren't why I chose this major. Those are all important factors, or else I would have picked something else. Then again, this whole AI uproar with all the current aforementioned issues is making me overthink my decision.
I hate to add to the negativity that is already looming over this industry and especially Reddit, and I don't wanna be dooming or coping or whatever, I'm just a kid rightfully very worried about the investment of my time, money, and personal aspirations.

Any insight whatsoever is very much appreciated.
Thanks!

forest niche
#

I have worked with Copilot and the most advanced version of ChatGPT for a year now. It is so dumb that I don't see it replacing any SWE anytime soon. AI cannot reason. Sure it can solve some LC questions that it has data on, but if you give it a completely new never before seen question the success rate is like 10%, if even that.

Let's put it this way. If a company can make 100B in revenue with 100 employees and AI makes them more efficient. Why in the world would the company not just keep those 100 employees and make >100B in revenue. Why would they fire 50 employees to maintain the status quo when they can do more with 100 now? In a capitalistic society they will never settle with the status quo. It does make my work faster, but all that means is that the company expects more from 1 employee.

#

Imo this whole AI hype is like the hype with quantum computers and automation in the last two decades. In order for AIs to truly replace SWE in terms of able to code and design everything by itself with minimum guidance and errors, we will need to make a technological breakthrough equivalent to the nuclear bomb or discovering electricity.

normal lagoon
#

this is a pretty long post, is there any specific question you had?

hazy sonnet
#

i truly hope you read that and understand why I am saying it

#

if you have never worked in our field are just starting out in your education on the foundation of our field, you really should step back and think about what you are making claims about first

i believe if you decide to do better you can do better than this @languid umbra

cunning cypress
#

AI is nowhere near good enough to build complex systems

They toy problems you get in school do not represent real-world problems that you work on.

languid umbra
# forest niche I have worked with Copilot and the most advanced version of ChatGPT for a year n...

thanks for the reply. Not all companies are gonna be magically making more money beacuse they can "do more" with more developers though. Comapnies make products in the end of the day and if they can deliver a product with 50 devs, hiring a 100 isn't gonna make them more money. quite the opposite, as now they have pay extra 50 devs they dont really need to get what they want. If you need 10 trucks to ship your goods, you won't order 20 as 10 is already enough to get what you want. as long as the final product is what they want, neither the comapnies or the customers cares how many devs are behind it

I also understand that at the moment the sucess rate isnt anything crazy, but with the developement of the O1 GPT model that internally "thinks" and "reasons" it seems like theyre exponentially getting better and with the current investements in AI it dosent seem to want to stop anytime soon

languid umbra
# normal lagoon this is a pretty long post, is there any specific question you had?

With how AI, oversaturation, offshoring, and rising expectations of the average dev in the recent years, and looking at how this is gonna develope in the future, is this industry even worth starting to get into? as someone who values stability and good work-life balance. Is it worth spending 4 years of time, money, hard work, and emotional investment into this career? if so, why do you think that is?

forest niche
# languid umbra thanks for the reply. Not all companies are gonna be magically making more money...

Keeping the same 100 will allow them to do even more than they are now. I never seen a company go like we X number of money, I am happy with that let's not do anything else. Also I don't think you understand the technological leaps we need to make for to get an AI that thinks and reasons. That's essentially an AGI. If we get AGI then no industry is safe. You are essentially saying that 99% of all jobs will die in the next 10 years.

languid umbra
# hazy sonnet if you have never worked in our field are just starting out in your education on...

your answer is rather vague and i feel is diminishing my opinon soley beacuse I don't have industry experience. I still believe I have valid concerns and I when I see people who do have years of experience still looking for junior positions, that only solidifies it.

"i believe if you decide to do better you can do better than this"

I have no clue what you're trying to say here. Im having all these concerns beacuse I am trying to take a step back and make the best decision I can make and do as good as I can

forest niche
normal lagoon
#

Unfortunatelly nobody can predict the future

forest niche
normal lagoon
#

ive been in the field for some years and don’t see AI taking over anytime soon, it is quite dumb and gets things wrong all the time

#

we can say yeah its good now but we’re speaking from our current reality and experience and right now its giving NO

#

but who knows what may happen 5, 10, 30, 50 years from now

hazy sonnet
normal lagoon
#

and if that time ever comes, can cross the bridge when u get there

forest niche
#

Also we are exiting a horrendously bad job market for tech atm, and even with that I don't see senior ppl applying for entry level positions.

languid umbra
hazy sonnet
#

I just want you @languid umbra to decide to not be like everyone else asking this same type of question

normal lagoon
#

but not doing something now because of what MIGHT happen 20 years from now is not particularly a great way of living live

#

we’d never get anything done

#

do anything we want to do

hazy sonnet
#

Decide to not listen to everyone else who is also not that experienced saying exactly the same thing you’re saying

#

And decide to look at it with a different mindset that isn’t so doomy, especially with such little information

#

Doesn’t seem fair to reality

forest niche
languid umbra
# normal lagoon Unfortunatelly nobody can predict the future

as I stated, I'm trying to argue that looking at how things have been going in the past, and where they are headed, this is a very likely scenario. i'm not looking for a voodoo "heres exactly whats going to go down" response. im trying to gather points to argue against the future scenario in my mind being likely to happen

image.png
forest niche
#

Like I develop Copilot product at Microsoft right now and I can tell you that the reasoning stuff is a complete sham atm.

normal lagoon
#

so not sure what the point is

hazy sonnet
#

I just never see anyone actually say how it would actually work

forest niche
#

If anything you should be worried about how much AI is overhyped right now and see all the tech stocks crash in a few years 😭

#

I am still waiting for automation and quantum computers to make jobs obsolete since 1990.

languid umbra
# forest niche What do you mean how things how been going in the past? Like what?

job market particularly. tech used to be fairly straightforward to get into. go to school, get a cs degree, find something to specialize in, ur good to go. now u need to apply to 500 places for a handful of interviews that are going to reject you. developer expecations have been going higher and higher and AI coupled with oversaturation is seeming to make the issue worst. thats all what I hear from folks on the internet tho im not gonna pretend like ive ever applied to any cs job but the number of times i see stuff like this make me believe this situation dosent apply to only a minority

normal lagoon
#

i have friends applying to hundreds of jobs that are NOT in tech

#

job market in general stinks rn

#

i know people struggling to get a retail job

#

this isnt some tech thing, would step out the bubble a bit and understand that theres issues bigger than AI

languid umbra
forest niche
# languid umbra job market particularly. tech used to be fairly straightforward to get into. go ...

The job market is cyclical you can't expect people to get into big tech positions when interest rates are so high. Back in the mid 2000s and mid 2010s people had to apply to hundreds of jobs too.

If you are concerned about oversaturation you need to think about what type of programmer you want to be. For example FE engineers are insanely oversaturated atm, but BE engineers are less so.

As in any industry expectations are going to get higher because that's how businesses get ahead in a competitive environment. Doctors today have higher expectations than they did 20 years ago.

hazy sonnet
#

No clue why people continue to forget the last two recessions

#

No clue why people think getting a professional job is easy by any means

#

Or ever has been in recent memory

forest niche
#

This isn't 2021 when everything was just handed to you 😦

forest niche
hazy sonnet
#

Promise you back in 2010 it still wasn’t easy to get a job at Google when they wanted ONLY Ivy League grads

normal lagoon
#

just do what u want to do, no need to grieve about the future

hazy sonnet
#

Which is why these conversations are so odd to me, why are we not using broader and better analysis

forest niche
#

It's also about your expectations @languid umbra. ya it's gonna be hard to land a 200k new grad job. But you wanna make 100k? Lots of those jobs out there.

normal lagoon
#

well thats also an issue, people wanting this big money jobs. somebody has to work at the small and mid level businesses which a lot of the time have better WLB and benefits

languid umbra
hazy sonnet
#

Just be worried about things that are realistic

#

Decide “I’m going to be more critical about my concerns”

#

And then realize so many people say what you’re saying without anything to back it up

forest niche
languid umbra
forest niche
#

It's all about your ambitions and your work ethic. I know people who complain about being poor in Canada, but they don't actually put into the effort to develop their career or study LC/SD.

normal lagoon
forest niche
#

I guarantee if you spend time studying for interviews or are at least a bit ambitious it's not hard to clear 100k in Canada with just a few YOE.

normal lagoon
#

nobody wants to struggle bruh

forest niche
#

@languid umbra which university do you go to?

normal lagoon
#

the time will pass anyway, theres millions of threads out there that talk about this. theres better use of time than to rehash the same discussions over and over and over and over

languid umbra
forest niche
languid umbra
#

still a lot of theory too from my time so far

forest niche
#

Basically every single profession is in danger of AI if you believe we gonna get AGI. Except for maybe doctors and emergency personnel

#

So unless you want to do those not many choices for you.

languid umbra
#

thats how i see it but yall do have some points that maybe im overstressing as i tend to do

hazy sonnet
#

The very first question I think of is “who owns the code the AI writes?”

I work at Bloomberg and there’s just no universe we would allow OpenAI to own the code we ask it to write

#

Then there are huge security issues

forest niche
wide seal
#

If anyone tells me ai can replace my job ill laugh em out of the room.

If AI can code in a dead programming language, and deal with a terrible buggy codebase be my guest.

hazy sonnet
#

Lmaoooo

forest niche
#

Even the most advanced coding AI, DevinAI, was proven to be nothing more than a sham...

#

Do make fakes out there

#

All Chatgpt clones with different prompts engineering implementations.

languid umbra
wide seal
#

I think thats just the nature of graduation numbers increasing at a crazy rate

languid umbra
languid umbra
forest niche
#

All the newly released versions of Chatgpt are just small bug fixes and very small incremental upgrades, essentially patch work.

wide seal
#

From what I understand it’s just the nature of the models we use now, unless we get a whole new model there are limits?

forest niche
#

The general sentiment in the AI research community is that we are hitting a wall with LLMs atm.

languid umbra
forest niche
languid umbra
forest niche
#

Not really anything technologically groundbreaking?

wide seal
#

sounds like good news to me

languid umbra
forest niche
#

With the current technology we don't have anything close. I work with PHD AI researchers and this is what they tell me.

languid umbra
forest niche
wide seal
#

I mean a lot of the stuff we use in ai models now has been known for decades and decades right?

languid umbra
forest niche
#

Also I wouldn't trust openAI stuff. Go with independent papers. They have an incentive to overhyped ai to their shareholders.

languid umbra
forest niche
#

The degree of correctness is still plateauing.

forest niche
#

And asking AI to do questions it has no prior knowledge in is still horrendous.

wide seal
#

Not a scientist but isn’t science kinda slow, when it comes to the fundamentals of the craft at least

#

Not specifically referring to ai but all of science

#

Someone learns something, spreads it, and it takes a long time for other scientists to get the ball rolling

#

And then industry, wayyyyy later

forest niche
#

Until we get AGI SWE will exist

#

Market is saturated ya but it's cause it pays a lot.

wide seal
#

If we get agi what jobs can exist hahaha, not even worth considering because itll be so lifechanging its hard to predict

forest niche
#

Also most of the market is filled with bootcamp developers or like people who can't even code loops even if they have a CS degree

#

Like the quality is def not super high atm.

languid umbra
forest niche
#

People and companies always want more

#

Every single company that has reduced and replaced SWE have failed miserablely

#

All of them just hired them back and even with layoffs

wide seal
#

I feel like applying ai to everything is kinda weird to do. There’s good applications and bad applications of it like any good tool

forest niche
#

Tech worker demand has grown YOY

#

The number of headcount for SWE jobs still continues to grow

#

It's just not 2021 crazy

wide seal
#

Some things like ai image generation will be able to push out artists, maybe llm will replace telemarketers

#

But, the market will have demand for something else

#

And not every skill is ai replaceable

languid umbra
forest niche
#

Anyway my point is if you put in the effort and aren't a lazy developer you will for sure find a job. If you want to do the bare minimum and still get 10 job offers, it ain't working out.

#

If you are truly scared of AI just drop CS.

languid umbra
forest niche
#

We don't think so but we can be wrong and you can be right. Just do your own research and make your own decision.

#

Like the other posted said no one can predict the future.

wide seal
#

Become a nurse

languid umbra
languid umbra
#

takeaways:

  • even with productivity boosts, companies will hire even more to make more software(?) cuz money(?). so its not really an issue. and jobless people rn is just a result of overall job market and not just swe

thats what ive got so far as counterarguements against my points

mighty nimbus
#

in regard to AI replacing jobs yeah no lol

#

you'd have to be pretty dumb for an AI/LLM to even remotely replace what value you bring to a company as an engineer

#

IMO AI rendering software jobs obsolete is years away if never. If you as a software developer can be wholly replaced by an AI then you're just a shit dev. If AI makes devs more productive and gives us the ability to do more things, why would anyone want to cut them? This entire field is built on fast innovation and rapid expansion/growth. AI tools help that; there will be MORE jobs

#

In terms of companies scaling down I think that really just comes down to the company. If you need less devs to run operations then that means you have more devs to do innovation/development/expansionn. I get the idea, like with AI you can output more with fewer people, but I feel like most companies will just take that extra people/devs and focus them towards expansion and growth so ultimately I think it comes down the company and its goals

#

Software eng as an industry is the absolutely fastest moving and evolving field out there and the current business paradigm heavily HEAVILY encourages ever-expanding growth and innovation. To think that the entire industry is going to just take AI and be like "aight time to coast and lay off" is pretty silly

#

Yes there are periods where economy isn't so strong and there are pull backs no doubt but to think that the entire industry is heading that way permanently? long-term? not a chance

#

tl;dr AI replaces code monkeys so if you want to be a code monkey then you can be worried. otherwise you have nothing to worry about. learn to use the new tools just as every dev has since the dawn of software eng. that is literally the name of the game

cold belfry
#

I was worrying about this quite a bit for a period of several months roughly a year ago. And I'm not a junior developer or first semester student, but somebody with almost 3 decades of programming experience and an M.Sc. degree in applied computer science from a relatively high reputation university.

Long story short, I have completely stopped worrying about AI by now. This doesn't mean that I'm now certain that "it" is not going to happen ever. But I came to a couple of conclusions.

#

  1. "It" (whatever we exactly mean with that) is almost certainly not going to happen nearly as quickly as I still assumed/imagined during the first year of the ChatGPT super-hype.

  2. Even if "it" might happen at some point in the future, there is no value at all in worrying about it now. What happens, happens, and we have no idea when or if "it" will happen at all.

  3. If "it" happens, it affects (almost?) everybody's job. Not only mine/yours, and not only software developers. Some sooner, some later, but I have little doubts that at some point, even if this point is still decades away, machines will be better than humans at literally everything. Then, the question will no longer be whether they can replace a human, but only whether we let them or not.

  4. Which leads to a very fundamental, "philosophical" argument: Even if, at some point, machines can do everything better and faster than humans, does this mean that we humans should just lean back and stop learning and doing things? I definitely think not. Such a reaction would not just only accelerate this process. Most importantly it would mean that we give up all control and transfer all responsibility to the machines. Do we really want to do this, as a civilization that, at least for now, still defines itself as "human"?

#

That said, I can well imagine that the safest jobs in the future will not be defined through high complexity / skill requirements, but though social dynamics / the "human factor". The safest jobs will be those where either other people simply prefer a human to do it instead of a machine, even if the machine is actually more competent. Or where you have enough social influence to prevent the machine from taking away our job, even if the machine is actually more competent.

#

But for the moment, I think my primary realization is that there was (and probably still is) a lot of "hot air" being blown around about the capabilities of AI, and about the capability and willingness of society and organizations to really optimize everything with AI so quickly, even if some things are perhaps even already possible today from a purely technological perspective.

#

I have started to use GitHub copilot a couple of weeks ago (late bloomer, I know). It is nice, but it does not "revolutionize" my work at all. And its impact on the human interaction side of my job is zero.

hardy bay
normal lagoon
#

what a sentence

hardy bay
#

im just sayin you aint guaranteed a great life just cause ur in canada, u gotta work bro

languid umbra
languid umbra
#

my worries been that the productivity boosts that AI offers, on top of the all the other things you yourself mentioned like competiveness and oversaturation, are gonna leave new graduates in an extremely tough spot in 5 years

languid umbra
#

if theres THAT much demand that there is always so much stuff to make, how are so many people struggling to find a swe job rn? (i mean to be fair i just remembered the whole job market is not great rn so it may be a result of that)

languid umbra
mighty nimbus
#

Just because there are layoffs and unemployment, doesn't mean that there is less demand, certainly not long term (maybe temporarily owing to aforementioned factors)

#

Either way, trying to connect the current conditions to AI is also entirely circumstantial

#

But in any case, to my knowledge layoffs are down and job market is much better now than 1-2 years ago

languid umbra
# cold belfry 1. "It" (whatever we exactly mean with that) is almost certainly not going to ha...

1 & 2. its rapidly growing since the inception and with the amount of funding big corpo is putting into it, it seems like "it" may happen sooner rather than later but what would i know? if "it" would be on the near horizon, i dont think its a good idea to say whatever happens happens and spend 4 years of hard work, money, and lots of time on a craft thats gonna not get me the life im trying to build

  1. I believe there are a lot of jobs that put value on whos behind the craft. many forms of art, healtcare, anything that has human intraction in the mix is not gonna be as easily replaced. but software, wether its 1 dev behind it or a 100, the product is the same, and customer will get the same not caring about whos actually behind the process unlike the other things i mentioned. thinking software dev automation = every single job on earth being automated is not very realistic cuz of the "human factor" you mentioned later

  2. i totally agree with this one tho

languid umbra
#

thanks for all the reassurance everyone. from all i got so far from here, and considering i cant really afford switching majors at the moment, ill stick with cs and will do my best. i think picking a niche may also be a good way to bulletproof myself from the oversaturation (currently intrested in VR)

cold belfry
cold belfry
#

One argument that I have often read over the last 2 years is "when AI is good enough to do junior work, then how can human juniors ever build experience to get beyond junior status, if they can no longer get a junior job?". I think that there are a few things wrong with that question.

One answer could be that people should no longer (and no longer need to) even try to learn the things that AI can already do faster and better anyway, and instead focus on things that AI can't do yet. And of course this applies to everybody, not just juniors.

But I also think that, at least for now, the question overestimates what AI can already do. There are still many complex tasks that AI can not do. And the irony of this is that these complex tasks often still require skills and knowledge that are seemingly "simple" when broken down into individual "pieces", but become complex when put together. AI might know all the simple parts when you ask it about the simple parts in isolated artificial situations, but fail at the "big picture". As a human, you can understand and solve the "big picture", but you still need to know the small and "simple" pieces. And that's why I think that learning the "simple" parts is still important.

cold belfry
# languid umbra thanks for all the reassurance everyone. from all i got so far from here, and co...

Learning to build things that not everybody knows how to build definitely sounds like a good idea. IMHO this is also a lot more interesting than building the millionth run-of-the-mill app. One "strategy" which worked very well for me is to get into very interdisciplinary environments at the borders / cross sections between very different fields, and between academics/research, industry, and public administration/government. There are not only interesting projects there, but these are also areas where your job will have a high degree of "human factor", reducing the probability that AI will be capable, and will be accepted, to do all the things you do, in the near future.

tidal trenchBOT
mighty nimbus
wicked coral
forest niche
#

FE positions are way easier and faster to fill too

languid umbra
mighty nimbus
#

You learn new stuff at work

#

This thread is in response to a guy in school, so what I said was that they should incorporate new technology and trends in their learning to prepare for the future market

#

But yeah if I were giving advice to someone currently working full time, in regard to AI jeopardizing your job, I'd be saying incorporate it into your daily work

full birch
# mighty nimbus But yeah if I were giving advice to someone currently working full time, in rega...

Yeah I am going to second your take here, you need to look at this like a tool to add to your tool belt. It is a very good tool if you can learn to leverage it. In my current role I use it to: explain things to me like I am an idiot (Spoilers I am), use it to quickly develop the boilerplate for test, POC, anything really. Populate data sets with dummy data that helps me understand the problem better. I use it to write nice emails to non-technical peeps (im lazy) ultimately you need to use it to be as lazy as possible IMO use it for the boring things and it will allow you to focus on the things that bring you joy.

languid umbra