A SWE's random thoughts in a post AI world

"Software engineers are cooked!" - yet another idiot on X

In my opinion, both sides of the “AI will replace devs” debate are equally obnoxious.

No-code purists speak in absolutes despite not having a fundamental understanding of how software works and why it is difficult to imagine how AI can completely take over engineering efforts.

Devs vehemently oppose against the idea, but the fact they are so emotionally charged reveals an innate fear and sense of insecurity, because deep down they know there is a high degree of truth to how SWEs can no longer just “code” to survive, and we no longer need as many code monkeys per project.

No-code purists are eager to replace devs because they no longer need to feel intellectually handicapped when seeking to materialize their ideas. Devs refuse to accept the new reality because they relish in the intellectual superiority that their (now cheapened) craft once bestowed upon them.

We no longer need as many SWEs per project

Regardless of whether or not AI will completely replace SWEs, the reality is that we no longer need as many devs to build and maintain most digital products today.

Code generation using LLMs is already a reality, and it is only getting better.

Resurgence of the Product Engineer

At the end of the day, what truly matters is the end product, and software engineering has always simply been a means to an end.

Regardless, I think what people seem to forget is that someone still has to be responsible for the entire development process of that product, be it via prompting or classical software engineering.

It is not as clear-cut as one role replacing another. It might simply be the convergence of existing product strategy, design and engineering roles when it comes to building digital product teams.

Product managers will have to become more technical to be able to iterate fast quickly and reliably on their ideas without introducing new bugs every new deployment. Engineers will have to develop stronger business acumen and learn to yap more. Both will require good product design sense to end up with UI/UX that feels good to use.

Hence the emergence (or rather, resurgence -- think full-stack freelancers of the not-so-distant past) of the Product Engineer. There will be room for product engineers with a more technical slant, and there will also be engineers with stronger business acumen and yapping skills. Different types businesses will benefit more from different types of specialisations, but one thing is for sure - both will now have to dabble into what used to be the exclusive domain of the other.

The modern world runs on complex but largely predictable workflows, and a good amount of man-hours have gone into building layers and layers of abstractions around these complexities.

With software eating the world, it is hence only natural that we will end up building abstractions on top of the traditionally complex software building process itself.

Understandably, one can't help but feel a sense of unease.

As with every industrial revolution, public pushback is to be exepcted primarily due to fears of job displacement. However, I am also skeptical that people were truly passionate about slogging away in factories doing manual and repetitive work. In other words, perhaps the short-term pain was worth the long-term gain at the macro level.

I'm not sure the same can be said about the next wave of automation, where we seem to be increasingly automating away work that people actually enjoy doing. For instance, designers and classical programmers tend to find immense passion and fulfillment in honing their now cheapened craft.

Were the cheap thrills gifted to us by the recent Ghibli trend worth the impending existential despair that it might potentially bring by cheapening the value of craftsmanship?

What will happen when quality craftsmanship becomes too conveniently accessible? What will we make of the old adage journey > destination when most of the journey becomes automated?

Existential reasons aside, I also wonder about the economical implications.

"The underlying purpose of AI is to allow wealth to access skill while removing from the skilled the ability to access wealth."

With AI increasingly able to generate commercially-ready music, art, and code, it makes me wonder what kind of crafts will be left that are still worth honing.

Performative crafts will still be valuable. I will much rather watch a human perform a play (ideally mostly written by a human) or watch a human belt out high notes than witness a machine accomplish such feats. I think that's because we naturally gravitate towards art that resonates, and what resonates is human potential and the personal stories leading to mastery and human flourishing.

However, I don't think the same can be said about craft in the context of commercial use. Business executives will gladly replace designers/engineers with AI and end-consumers wouldn't care, especially if it means cheaper products.

Job Displacements

For the longest time, I was worried about AI displacing jobs. However, I have come to believe that that might not entirely be a bad thing. AI could form the foundation of new workforce dynamics driven primarily by intrinsic motivation; people who continue to work do so because they want to, not because they need to. This might have positive implications.

There is also a good number of people who remain in the workplace despite their presence being a net-negative for their peers, be it through sheer incompetence or toxic ego. AI might empower the average employee to stand up against shitty managers whom they no longer need to tolerate in order to put food on the table.

There is also a good number of people in society who hate their jobs and would not want to work for a living if they had the choice. AI might make it possible for a society to accomodate free-loaders. At the very least, 3-4 day work weeks might become the norm.

If this becomes the new reality, then the next question to ask is who gets to stay home and indulge, and who has to continue doing the undesirable but necessary work that AI still can't do (for now). Eg. cleaning toilets. The social contract we know of today would probably have to be re-examinated and re-written from the ground up. Who knows, we might end up rewarding people who do the dirty work most handsomely, which is probably the right thing that should have been done all along.