Once upon a time, there were people who had the job title “computer.” Then we invented computers. All those “computer” jobs shifted from actual calculation, to coding, configuring, and using computers.
Once upon a time there were people called “programmers.” Then we invented LLMs…
[you are here]
Talking to computer and computer talking to us is a big deal though
It’s hard to imagine the outcome of it on the long run
If the LLM coder bots are good enough to take all the jobs, they'll also be good enough to contribute code to open-source projects, so the open source development process will probably evolve into a bunch of humans nitpicking each other's LLM code.
Git-wikis could allow people to discuss and edit code collaboratively, like how contributors work on Wikipedia/Wikifreedia.
Highly active projects could have every line of code accompanied by lines of comments from critics and defenders explaining how the conclusion was reached to use that exact line of code.
This way, humans could retain control, and human knowledge of programming could continue to grow, instead of atrophying.
Or maybe I'm putting too much faith in people again.
(Copied and pasted from a previous post of mine)
You are too faithful in people that only care about money.
If IA is more efficient ECONOMICALLY than a humans, whatever the human can include, if the result is acceptable by managers you will loose your job.
It is just that simple.
And the opposite is true. For any job where robotics or AI is too expensive for the human result then the human will be more paid.
It is the way the current world works
Not there yet and not even close
This. Junior devs may feel some heat but it’s a far cry from threatening even mid-levels.
note1wh6qndx4w8rcgmgqmc2lyh0cgvrtg5zes9sfxc9kjmtv9zw0z2dspn3vr2
I see what you’re saying - thanks for the clarification!
In my experience, it’s a LOT of work to get the thing to produce something that works in a real production environment with real customers. Not just on the prompting side, which often requires a lot of prep work like identifying and bundling relevant files and such, but also on the verification side, which is like a huge code review.
That said, I’ve found usages. For stubbing out tests of all the the corner cases of our business logic in one domain at our work, given the right prompting, it saved me hours and hours of manually creating JSON files.
note1wh6qndx4w8rcgmgqmc2lyh0cgvrtg5zes9sfxc9kjmtv9zw0z2dspn3vr2
We should just go back to punch cards
I’ll start to worry for my career when “Vibe Debugging” works across systems.
It’s not about you losing your job, it’s about the fact that you will be working with LLMs to do your job.
Or losing your job to someone who knows how to use LLms to do your job.
Yes agreed, but engineers who develop an over reliance on these tools will dull their skills and understanding of their product. IMO there’s a difference between using AI to streamline your core competencies and using it instead of learning a skill.
And that’s not to say it can’t be useful in learning skills, but the onus is on the user to develop their understanding of the subject matter.
Technology will keep evolving. In the end, regardless of the words we use, we’re all builders 🙂
A lot of programmers seem to now be saying that there's no risk to them losing their jobs to AI and I find that super hard to believe.
Tim Cook of Apple said one year no more computer programmers. Two years no more artists or mathematicians. Three to Five years AGI and one year following AGI is ASI (artificial super intelligence). ASI being more intelligent than all humanity combined.
I believe we are just past the event horizon of the Singularity. Hold on its going to to be one hell of a wild ride.
