3 ms·

A question for programmers in their 30s wanting to still be doing this in their 50s and 60s... Are you worried about the advancement in AI, i.e. programs writi

A question for programmers in their 30s wanting to still be doing this in their 50s and 60s... Are you worried about the advancement in AI, i.e. programs writing other programs... will there still be enough programming jobs in 20 years for humans? Ps I'm only a prototyper, I don't create production ready code, hence the slightly naive question.

alpha_squared 10y ago
I'm still in my 20s, though fast approaching 30s. I'm not sure if I'll still be doing this in my 50s and 60s, but I'm not concerned. Maybe it's naive of me to not be, but there is a pretty significant delta at the moment between current AI and true, sentient, AI. We're quite a ways away. And, if we ever do get there, I think the least of my problems will be, "oh, well, there goes my job."
karllager 10y ago
No, not worried about that, particularly. But there are few other options, that might be worrying to some: The routine tasks will more and more go away. Still amazed how much is done by hand today (like research, data preparation, communication, marketing, business to business processes, scheduling, HR). Just as with every profession over the past decades, if you want to stay employable, you will have to level up your game significantly because the simple software stuff (of which there is a huge amounts today) will go away, let's say in the next ten years. In 2030 software related employment will look more like in the 1960s. A highly selected and highly self selective group of borderline asperger guys having a blast thinking, proving theorems, playing piano and writing code. Why I believe that? Because we see the 80-20 rule play out in so many fields. And in 2030 the leverage you get from technology will just a tad more intense that in is today.
jimbokun 10y ago
"will there still be enough programming jobs in 20 years for humans?" Maybe not, but that means there aren't any other jobs for humans, either. Machines that no longer need humans to program them is pretty much the definition of the singularity, which means machines quickly surpassing humans at everything.
uxcolumbo 10y ago
Found some more good answers here https://news.ycombinator.com/item?id=13929633 https://news.ycombinator.com/item?id=13929633
randcraw 10y ago
Yes, I'm confident that AI will have very little impact on software jobs in the next 20 years. Fact is, every usage of AI still requires labor intensive setup (choice of hypothesis, choice of data, assessment of signal-to-noise, choice of algorithm, intern of results), unless the 'intelligence' is embedded in hardware for easy (automatic) setup and reuse. But that requires a clear simple hypothesis and pose of problem, which most problems are not amenable to. Specific repeatable tasks that follow strict guidelines (like smart cars) can easily be automated by AI (once the sensor problems are solved). But nonspecific tasks that follow loose guidelines or require substantial effort to organize and implement (like programming) -- these jobs will be very hard to automate. They just require too many custom steps and non-obvious decisions that require a lot of 'hand-holding' for a machine following a simple set of rules to succeed. If you can describe the rules of the task simply and completely, then AI can do it (drive a car, play a board game like chess and go). If not, AI might be able to help a human with the more formulaic aspects of the job, but it can't glue all those pieces together and replace the human planner... not for the foreseeable future.

alpha_squared 10y ago

I'm still in my 20s, though fast approaching 30s. I'm not sure if I'll still be doing this in my 50s and 60s, but I'm not concerned. Maybe it's naive of me to not be, but there is a pretty significant delta at the moment between current AI and true, sentient, AI. We're quite a ways away. And, if we ever do get there, I think the least of my problems will be, "oh, well, there goes my job."

karllager 10y ago

No, not worried about that, particularly. But there are few other options, that might be worrying to some: The routine tasks will more and more go away. Still amazed how much is done by hand today (like research, data preparation, communication, marketing, business to business processes, scheduling, HR). Just as with every profession over the past decades, if you want to stay employable, you will have to level up your game significantly because the simple software stuff (of which there is a huge amounts today) will go away, let's say in the next ten years. In 2030 software related employment will look more like in the 1960s. A highly selected and highly self selective group of borderline asperger guys having a blast thinking, proving theorems, playing piano and writing code. Why I believe that? Because we see the 80-20 rule play out in so many fields. And in 2030 the leverage you get from technology will just a tad more intense that in is today.

jimbokun 10y ago

"will there still be enough programming jobs in 20 years for humans?" Maybe not, but that means there aren't any other jobs for humans, either. Machines that no longer need humans to program them is pretty much the definition of the singularity, which means machines quickly surpassing humans at everything.

uxcolumbo 10y ago

Found some more good answers here https://news.ycombinator.com/item?id=13929633 https://news.ycombinator.com/item?id=13929633

randcraw 10y ago

Yes, I'm confident that AI will have very little impact on software jobs in the next 20 years. Fact is, every usage of AI still requires labor intensive setup (choice of hypothesis, choice of data, assessment of signal-to-noise, choice of algorithm, intern of results), unless the 'intelligence' is embedded in hardware for easy (automatic) setup and reuse. But that requires a clear simple hypothesis and pose of problem, which most problems are not amenable to. Specific repeatable tasks that follow strict guidelines (like smart cars) can easily be automated by AI (once the sensor problems are solved). But nonspecific tasks that follow loose guidelines or require substantial effort to organize and implement (like programming) -- these jobs will be very hard to automate. They just require too many custom steps and non-obvious decisions that require a lot of 'hand-holding' for a machine following a simple set of rules to succeed. If you can describe the rules of the task simply and completely, then AI can do it (drive a car, play a board game like chess and go). If not, AI might be able to help a human with the more formulaic aspects of the job, but it can't glue all those pieces together and replace the human planner... not for the foreseeable future.