Will AI reduce the need for software developers?

semaphore said:
Most of the devs in my company are in their late 40’s early 50’s. Previous companies we had a guy in his 70’s. I’ve never really seen age to be an issue personally.
Click to expand...
analyticsindiamag.com

OpenAI is Not Built by 24-Year-Old Programmers

Altman said that seasoned entrepreneurs, rather than first-timers, often make the best founders due to their experience and ability to experiment more effectively.
analyticsindiamag.com analyticsindiamag.com
 
As a side note, I have been using Github Copilot for about 7 months now and I find the code-completion very intuitive and useful.

What is also very handy is the Github Copilot Chat, I often give it a small chunk of code and basically say "simplify" or "refactor" and majority of the time it does.
 
Younger software developer's 2c. 24 and been writing code for 5-ish years. Got retrenched last year May and I've been doing contract/freelance work to get by.

In the last 5 years there have been a bunch changes that I've witnessed. The mass adoption of cloud solutions and microservices, JavaScript ever-changing ecosystem(react, angular, or whatever they are peddling these days), languages morphing in to each other(Java becoming more like JS and JS becoming more like Java), the blockchain hype, etc and now AI.

It gets tiring some days. I've worked with different things but feel I have mastered nothing. Which leaves me with an empty feeling. I like to believe that becoming a craftsman as @shooter69 alluded to, is possible.

I've always loved tech growing up, but lately it's been just "meh". Each new iteration just feels like we are complicating things for ourselves. There are a plethora of ways to say "Hello world", but how many do we need?

Getting back to AI. Benefits: It reduces boilerplate and increases output. Costs: It will just create a new layer of abstraction, making debugging a bit more difficult.
 
The mass adoption of cloud solutions and microservices,
Click to expand...

:sick:

I shudder to think of the amount of LOB application that have 100 users, and it made up of 10's of microservices.


When you get disillusioned because you have a broad range of knowledge, remember that the ACTUAL idiom is

Jack of all trades is a master of none, but oftentimes better than a master of one
Click to expand...

:D
 
randomcat said:
Younger software developer's 2c. 24 and been writing code for 5-ish years. Got retrenched last year May and I've been doing contract/freelance work to get by.

In the last 5 years there have been a bunch changes that I've witnessed. The mass adoption of cloud solutions and microservices, JavaScript ever-changing ecosystem(react, angular, or whatever they are peddling these days), languages morphing in to each other(Java becoming more like JS and JS becoming more like Java), the blockchain hype, etc and now AI.

It gets tiring some days. I've worked with different things but feel I have mastered nothing. Which leaves me with an empty feeling. I like to believe that becoming a craftsman as @shooter69 alluded to, is possible.

I've always loved tech growing up, but lately it's been just "meh". Each new iteration just feels like we are complicating things for ourselves. There are a plethora of ways to say "Hello world", but how many do we need?

Getting back to AI. Benefits: It reduces boilerplate and increases output. Costs: It will just create a new layer of abstraction, making debugging a bit more difficult.
Click to expand...

I know the feeling of "lost" in the expanse of tech that seems to come out every week.

I worked for large organizations for 2 decades developing their in-house system. I learnt a lot and am grateful for the opportunities, but found I become stagnant very quicky as once the system is implemented, its difficult to just start using new tech. I was demotivated, felt I was being left behind, was feeling lost in my career. I tried skilling up, but it was difficult when I couldn't really apply the new tech to anything you are busy with.

So at the beginning of 2022, I went job hunting for a new position, specifically for a smaller company that does smaller projects for smaller clients, so that I could get exposed to different business and tech. I got lucky and a company really liked my skillset and I really liked the setup, it fitted what I was looking for within the first 2 weeks.

Below is some of the primary tech I have used/integrated into over the last 20 months or so

  • AWS - DynamoDB, Lamda functions, Cognito, S3
  • Azure - Identity management
  • .Net (core) - 3 web apis
  • React - 3 web apps
  • Next.Js - 1 web app
  • node.js - 1 web api
  • Jira - 1 integration module

for a range of clients in below industries

  • Mining and logistics
  • Smart home
  • Trader communications
  • Foresty/Nature
  • Vehicle rentals

I have learnt more in the last 20 months than I have in the last 10 years. (from a tech point of view). Maybe think about what kind of development you want to be doing in 5 years time.
 
AdrianH said:
More to get a discussion going than anything.

So being halfway through my career, my current goals are learning the new tech stacks, skills, languages, competencies, etc to stay relevant in the modern age of programming. Fortunately we do shorter projects for customers, so I get exposed to a lot of different tech stacks, languages , infrastructures often which is super. But I have also been thinking about what I should be looking towards with the emergence of AI and code generation.

I do think about how things changed over the last 25 odd years in software development, and all the times I have had to skill up over the years, how tech stacks 10 or 20 years ago is now redundant, how some skills and languages I have learned are obsolete, and how I have had to change from waterfall to agile, etc.

Now I am wondering what I should be looking at as a long term goal 10 to 20 years ahead so that I will still be able to develop software.

It's inevitable AI will take over some roles and functions from software developers, but I also don't believe they will make software developers redundant, well probably not while I am still working.

What's your thought fellow old programmers? What are you planning to do to ensure you are relevant in 10 to 20 years time?

I am also interested in hearing what some of the younger software developers think of the changes, and how they think it will affect their careers as you have to stay relevant for even longer.

PS: I don't want to change careers, I want to write code until I retire.
Click to expand...
Well you need a good amount of code to be able to train the system... that aint coming from a bot. Then there's the constant framework updates...
It might eliminate the need for code monkeys.
 
NeonNinja said:
Plus isn’t the ability of AI to learn a big factor? Also it keeps on improving. I think it will (replace). In the next few decades.
Click to expand...

Not with the current ai, all the current ai uses batches of fixed datasets to train off of in order to make an ai model from the machine learning. This means that the ai knowledge is only as useful and recent as the data it was trained on, if there was a new language or framework that exploded in popularity tomorrow then the ai can't provide any code for it as there is no dataset available that it could learn from, you have to wait until there is a critical mass of data available that allows for machine learning to train a new ai model.
 
lkpat said:
LMAO not sure if serious.
Github fills itself?
Click to expand...

CoPilot was trained on GitHub.

I think we are talking past each other here.


tenor.gif
 
AdrianH said:
CoPilot was trained on GitHub.

I think we are talking past each other here.


tenor.gif
Click to expand...
I think you're misunderstanding.
Copilot was trained on output from developers. Github is mostly a repository of code from people with jobs.
Github does not generate content, employed developers do.
Add to that the speed at which new framework releases came out and you quickly realise that ML is useless without people for data. People will still be employed, but ML can augment the role.
 
Mystic Twilight said:
Not with the current ai, all the current ai uses batches of fixed datasets to train off of in order to make an ai model from the machine learning. This means that the ai knowledge is only as useful and recent as the data it was trained on, if there was a new language or framework that exploded in popularity tomorrow then the ai can't provide any code for it as there is no dataset available that it could learn from, you have to wait until there is a critical mass of data available that allows for machine learning to train a new ai model.
Click to expand...
This.
 
Also current ai is innovation-negative since it is backwards looking (based on past data) and can't create anything *new*, it can only regurgitate existing methods in new forms, this means that there is a knowledge plateau that leads to stagnation if there is an over reliance on ai and no significant growth in [human] development of programming innovation.

The day developers can be replaced is probably when ai becomes true machine based intelligence.
 
Mystic Twilight said:
Also current ai is innovation-negative since it is backwards looking (based on past data) and can't create anything *new*, it can only regurgitate existing methods in new forms, this means that there is a knowledge plateau that leads to stagnation if there is an over reliance on ai and no significant growth in [human] development of programming innovation.

The day developers can be replaced is probably when ai becomes true machine based intelligence.
Click to expand...
When AI can secure frameworks and create new frameworks, things change, but yoh.... that's a looooong way away.
 
I do not think AI will replace the need for software engineers in the short or medium term. I see AI as an assistive tool, which can help developers get the answers they need or clues when working on something. I think a developer will still be needed to incorporate the best solution given the scope any other constraints to the project. I have gotten code generated from ChatGPT which was outright wrong, or code which was not the best way of accomplishing a given task. I think AI has a long way to go if it is to ever eradicate software developers.

I switched off the MS Edge landing page with news content after it started showing a lot of far-right media headlines. I then learnt (via YouTube) that MS had gotten rid of its news content selection team and replaced it with AI.
 
Mystic Twilight said:
Not with the current ai, all the current ai uses batches of fixed datasets to train off of in order to make an ai model from the machine learning. This means that the ai knowledge is only as useful and recent as the data it was trained on, if there was a new language or framework that exploded in popularity tomorrow then the ai can't provide any code for it as there is no dataset available that it could learn from, you have to wait until there is a critical mass of data available that allows for machine learning to train a new ai model.
Click to expand...
I agree, plus I did say in the next few decades. Lots of time for development and improvement.
 
Come, come to the "DevOps" side, somebody has to keep that AI dev running on kubernetes...
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X