Comment by systematical | Hacker News Reader

Comment by systematical | Hacker News Reader

Parent Full thread

When most engineers automate repetitive work flow, they seek praise and then find other useful things to do. I think automating, then being lazy, is an exception rather than rule for most.

With that said, I do believe AI will replace most developers in our lifetime. Particular ones like me, that create run of the mill relational systems for SMBs. I didn't need to go to school or get a fancy CS degree to do what I do. Nothing cutting edge. AI will cut deep into development like everything else eventually.

A lot of what we do is converting business rules into code, storing that data, returning it, and performing some useful operation with it. The remainder is making the systems easy for humans to develop on (be it DevOps or clean design). That whole last part doesn't matter to AI, in fact it's really a waste of time when you remove human engineers. Even still, AI can figure out SOLID principals, cyclomatic complexity, style guidelines, and general human readability if it needed to, but why?

> With that said, I do believe AI will replace most developers in our lifetime.

I don't see this happening. I don't believe AI is anywhere near sophisticated enough to tackle the majority of problems that developers are solving on a day to day basis. Even if we somehow manage to have a major breakthrough with AI, it's not going to be commercially available/viable enough for average-joe dev shop to implement it.

I'm far more worried about AI replacing lower income jobs and widening the pay gap even more than I am about tech workers losing their jobs.

>A lot of what we do is converting business rules into code, storing that data, returning it, and performing some useful operation with it.

Sure, but you still need someone to make those business rules. I don't see AI being able to interpret the direction of MBA filled meeting rooms without some sort of translation later (i.e., developers).

AI is hype right now. I have not yet been convinced of its practical application, or anyone's real desire to invest in it (other than the handful of companies who would benefit off of its implementation).

I just find it interesting that the audience of this site, can't fathom engineers being replaced by AI. Not even in the next 50 years? That would put me in my 80s.

My Dad will be in his 80s in 15 years. If we look at the landscape of computers when he was my age, 30s, the big thing was still mainframe computers. The internet barely existed. Smartphones? Nope. What was the big programming language in 1990? Was it Pascal, BASIC or C++? Java wouldn't even come out for another 5 years. 15 years before that he was still loading punch cards into machines.

I guess we will just see. I won't be the surprised one though.

I think you vastly under estimate the process. We cannot even get a state of the art AI to recognize large stationary objects on a highway, so how are they supposed understand customer requests? If, and that is a big if, we ever get to a Strong AI scenario I'd be worried, but not until then.

It is, IMO, far more likely that we'll see someone come up with a general purpose SMB system like WordPress.

I've been automated out of a software development job before..

Only when I can see a human shaped AI put a duvet cover on a duvet would I be worried about AI.

>With that said, I do believe AI will replace most developers in our lifetime

I'm skeptical about the idea of coders programming themselves out of a job, and I can make that skepticism rigorous. The following slides explore structures intended to capture the core essence of computer programs writing computer programs, and these slides culminate in a theorem suggesting doubt. https://semitrivial.github.io/MeasuringIntelligence2019.pdf

It boils down to the fact that if we create a Strong AI that can improve itself, the first AI is as intelligent as it can get. Only thing holding it back is available processing power. We will know if we've created a Strong AI, even if it's dumb.

> I do believe AI will replace most developers in our lifetime

AI can't replace a developer, it can only perform the tasks that developer currently does. The developer can go do something else. The possibility that AI can make humans obsolete in the next 100 years is extremely unlikely IMO. By doing the low-level tasks, AI opens the doors for humans to spend their time on more important things.

It was roughly 70 years from the demonstration of the first working transistor until a sufficiently large collection of transistors could pilot a commercially-reasonable car on public streets.

I can't begin to imagine what the next 100 years will bring.

Reply on news.ycombinator.com