Software engineering seems safe, if that is what you are primarily thinking of in your question. Developing software at the same level of a skilled engineer is a problem we'd only be able to solve with general AI. Even if program synthesizers (things that take declarative high level requirements and produce code) worked (and they don't at the level required for an non-technical user to use), most engineers today must be involved in the design process, and finding requirements and formulating solutions is a skill that requires humanistic and non-technical intuition.
It's not at all clear that software engineering is safe.
Sure there will probably always be a few programmers, but as tools (and soon I expect AI) gets better those programmers can do more and more of the work. I wouldn't be at all surprised to find the number of software engineers decreasing in the not too distant future.
The only reason it hasn't already is that we've managed to grow the number of things software developers are doing at the same rate as the increase in number of software developers * the increase in efficiency per software developer. And in some parts of software development (e.g. games) it has.
It's possible that you are right. That said, tools are already better than ever before, and yet demand for software engineers has only increased. That's because most tools which are powerful enough to do all the things we need aren't simple enough to be applied without human-level intelligence. Not to mention—the tools themselves are software that also requires programmers.
Personally, I'm of the opinion of the commenter above you—automating most software jobs will require strong AI. Replacement of those might happen someday, but it's sufficiently far away to not think about as an imminent threat to your job (though I agree we should be doing high-level planning about how as a society we will cope with strong AI).
Some of it has nothing to do with ML/AI, docker probably killed more Sysadmin jobs then all of ML combined.
Better tools allow us to do more, faster etc. Thing is amount of things software can do that it is not doing yet is absolutely immense and ML only opens more and more of it.
I think any actual reduction is so far in the future we have no concept of when that will happen and how.
I mean a am I huge proponent of self driving car.. when is that coming? When it works it will absolutely change economy as massive amount of people are now involved or depend on those that are.
It's not at all clear that software engineering is safe.
Software is an arms race. You need new software if your competitors are writing new software, but any organisation can coast for a long time on its “legacy systems”. That’s what we saw after the dotcom crash. If there is a prolonged economic downturn and the software industry just goes into maintenance mode, it will be very, very unlike anyone who wasn’t there last time will understand.
If current trends are any indication, the only way I can foresee the number of software engineers decreasing is if we make the job so complicated that the IQ floor on being able to perform at an acceptable level rises too high. We are currently piling the abstractions higher and higher, and we're patching them up with duct tape fast enough to keep most of the leaks at bay, but when things go south into Here-Be-Dragons territory, there are just a helluva lot more dragons lurking around these days, and they can be some nasty Lovecraftian dragons.
From your comments I feel like I cam safely assume you do not do any serious work in ML/AI. It's not even close, and none of your examples are being threatened by advances in this area.
It already has eaten the lower ends of programming on the web side with centralized services and templated web development. You can keep defining software engineers as those who haven't been automated away yet -- but I'd wager many already have. For example most companies are not spinning up their own hardware anymore, and less and less of their own software.
You’re completely right! I’ve thought this myself: Zapier and the ecosystem that enables it is such a good example of how something that used to take a programmer days or weeks can now be done in minutes.
I wonder if the end result of things like Zapier will ultimately be unemployment for devs. Perhaps, but maybe it’ll just free them up for something else. I mean, is the demand for programming really fixed? Or will more and more powerful force multiplier tools like Zapier just free up devs for new demands that we can’t predict now? Hard to say and I’m concerned like you, but it’ll definitely be interesting to watch.
So yeah, I agree with you and I think Zapier is a great example. I think we all just have to wait and see if that ends up replacing devs in the long run or just shifting what they do.
