This notion that "those who leverage AI in their work will survive" is so ludacris. I am aware obviously the LLMs in their current state hallucinate at a rate which makes it unfeasible to eliminate engineer positions but in the future how is this progress not concerning? It seems like most of blind is on copium and thinks that GPT/some iteration of AI has no/insignificant effect on human capital which to me makes no sense considering how rapidly progress is churning out. People keep saying "those who leverage AI in their work will survive" without understanding that entails a MASSIVE reduction in the size of a work force. Some future iteration of a LLM will be able to read proprietary code and understand closed source codebases to the point where basic debugging/feature development or creation of unit tests is so trivial and can be done instantly. Junior and midlevel SWE impact is so minimal anyways that the notion of an LLM creating a new branch, quickly solving x issue, and creating a PR is not all some "crazy" far in the future idea. Capitalism's quest for constant reductions on inefficiency seems to have been its catalyst in suiciding itself. Thoughts? Any potential good subsets of CS to pivot into? Perhaps hardware? TC - 150k #hardware #semiconductor #software #swe #chatgpt #openai #ai #llm #hallucinate @google @facebook @meta @amazon
Didn’t read. Im happy for u tho, or sorry for that
Tf is Hexplora? You are so naive to think you are safe. You trivial CRUD api can 100% be churned out by an LLM and you are in denial lol
Ok keep going 👍
Here’s the thing: SWE is corrosive to many other jobs and professions, such that when GPT corrodes SWE, SWE can expand their scope and corrode other jobs. The level of digitalization and automation in today’s world is actually way lower than you think. You have no idea how many low-end mundane office jobs exist in corporate settings which softwares could have automated. GPT will arm SWEs and help them invade into traditional, under-digitized industries. As long as there’s one single white-collar profession left, there will be SWEs building stuff to automate that profession. The end of SWE comes AFTER the end of all other white-collar professions.
Yes but you aren’t denying the copium we see on Blind. Sure SWE is corrosive relative to other professions, but AI and its future fine tuned iterations will be the epitome of corrosive to SWE
there are endless scope to SWE jobs and human are 1 million miles from full automation. So if GPT increase productivity by 300%, then our scope will expand, but swe headcount will not reduce. FYI I’m not was really talking about GS, goldman still sucks as it always has been
Can someone stop these clueless PMs and low level SDEs from making doom posts about engineering?
Maybe in 10 years, we look back and find out OP was right 🤡
Or not 🤔
Health & Wellness
Yesterday
1447
Indian street food is disgusting
India
Yesterday
1401
Indians will be the death of Indians
Health & Wellness
Yesterday
746
I have a Toyota Camry and got made fun of by Tesla owner friends
Ask Blinders
Yesterday
464
What does "Abortion after birth" even mean?
Software Engineering Career
Yesterday
1297
When Will The AI Bubble Burst?
You underestimate how much work it takes to maintain software systems and how much work is pending. Even if AI makes some things easier, it's not going to result in the world running out of software work to do any time soon. People said when Java came out that now there will be fewer jobs because Java is much easier to write and debug than C. This was in the 90s. Same with SQL, Wordpress, etc. Also, AI is not the biggest impactor of dev productivity that I've seen. There have been many similar leaps in dev productivity over the years - new languages, frameworks, cloud, no-code, new IDEs etc. There's just such a mountain of work to do that these things don't put a dent.
Dude anything full stack/mobile related has 20 billion resources available for an LLM to train on. That means it can rapidly identify issues, create boilerplate code, explain what certain endpoints/components do in legacy codebases, and drastically decrease the amount of time to make new features. There WILL be head count reduction. Maybe not like 90% but I guarentee you all these CRUD positions will be a thing of the past for the most part.
Well, I think it's simple then. Move on from your position if you're doing CRUD. IMO, more code for experienced developers to fix. I can't wait for code in production environments to 10x because some unfamiliar launched a product solely built on AI. AI will get better, but not before it builds a bunch of questionable software and then iterate further code changes in more questionable manners. I can totally see some higher up screaming"turn the AI off and role back to a working version". Companies don't have unlimited $$ and compute resources to allow it to iterate endlessly.