Πέμπτη 23 Μαρτίου 2023

Will AI replace software engineers (or any other professions)?

Written by a software developer with several years of industrial experience, and an academic background in Computer Engineering (including Machine Learning).

-----------------------------------------------

Introduction

    Chat-GPT replacing programmers -among other professionals. Writers, journalists, academics, translators, help-desk technicians, and even lawyers. In the long term, no profession seems to be safe from chat-GPT-like Artificial Intelligence (AI) -capable of understanding human language and solving our problems within seconds.

    The case with software engineers (or programmers, if you prefer a less fancy title, though the two are not exactly the same) is pretty interesting, because nothing seemed to threaten our profession in the short term or even in the long term. 

    Virtually all low-code platforms were no more capable than delivering complex code, than 2005 Google Translate was capable of delivering accurate translations on more than a few words. Now chat-GPT writes pretty impressive functions, and refactors chunks of code better than most senior devs would. 

    So, what happens with programmers in the future? Or any other profession actually?

 

General assumptions

    Let's begin with something few would disagree with: chat-GPT is bringing a revolution. And when mentioning Chat-GPT, we don't necessarily mean OpenAPIs Chat-GPT itself, but the level of AI that rendered it possible. More likely than not, there are numerous competitors to Chat-GPT coming within the next months.

    Yet like every revolution, the outcome is tough to predict; it may lead to more freedom, like the French revolution, or it may lead to a disaster. In fact, historically speaking revolutions are more likely to have very controversial outcomes (like the Russian revolution), or clearly disastrous outcomes (like the peasants' revolt of 1381 in England, or the 1523 rebellion in Germany), than leading to something good. 

    One could argue that we are now talking about a technological revolution, but this does not make it very different. The industrial revolution has so far a controversial outcome, with some intellectuals regarding it as the beginning of the end for humanity thanks to climate change. 

    No matter the approach, we can begin with two assumptions. Firstly, chat-gpt style AI is bringing a revolution. Secondly, a revolution may have good outcomes, but it happens pretty often that it leads to a disaster. 

    But what disaster could the chat-GPT possible cause? Isn't AI meant to make our lives easier, free up our time, and lead to a better living standard for all of us?

    This sounds very nice and plausible, but in the current economic model, no technology is meant to make our lives easier; they are only meant to increase the profits of the owning class. 

    We produce way more food than we actually consume thanks to the modern industrial technology, yet a huge portion of it is thrown away without being consumed, while millions of people are starving. The industrial technology that could very easily be used to satisfy the need of all of us, is only used to increase the profits of people investing on a vague notion of productivity.

    The case with housing is pretty much the same, as we build houses easier and quicker than ever, and in many countries the amount of empty apartments far surpasses the number of the homeless. Still, working class people have to pay large portions of their income just to retain the right of living in a house or apartment they will most likely never get to own. 

    There are plenty of examples one can come up with, and they all "converge" to the same conclusion: no matter how technology makes our lives easier too, it is mostly meant to increase the profits and power of the owning class.

    I have no intention of writing (yet another) "lefty" rant here. But understanding that our current economic model is meant to serve only a handful, is vital before pondering our original question: what happens with our professions during or after the AI revolution? 

 

What happens to our professions then?

    Before abstracting and giving a general reply for all professions, I will try to give an answer for us programmers. 

    We can regard it as certain that AI will decrease job postings. It is already known that in many (if not most) companies, there are periods when developers don't have much to do, and spend their time on improving existing code or working on side projects. And of course, like virtually all professions including a desk, pretending to be working (a well-studied and widely known phenomenon we are typically reluctant to openly talk about).

    Now that improving code and even generating it becomes easier, these periods will become even longer. In a completely rational economy following the current economic model, one would expect lay-offs, but our economy is not rational. 

    We constantly invent new professions that barely produce anything just to keep everyone busy on an 8+ hour schedule. Millions of people in the western world spend their days on pointless meetings or editing excel sheets, just to retain the right to live in a house. Book lovers might appreciate the writings of David Graeber on the topic.

    Whoever has worked in a modern company is aware that many people with fancy-sounding professional titles, do actual work less than a couple of hours per day, and spend the rest of it literally pretending to be busy. It is the result of insisting that everyone must work at least eight hours to get access to food and housing, while industrial production and population increase have rendered it unnecessary that everyone works that much in order for society to have such needs covered.

    So, even though one might expect that many developers will now be laid off because they won't have much to do in their 8 hours, our economy is actually irrational and does not lay off professions that do not do actual work in their 8 hours. 

    But there is a vital difference between other office-related professions and ours: our profession was meant to be productive in the first place. The fancy-sounding "digital product  marketing communications super manager" was never meant to be productive all day long; the primary function was for other managers and CEOs to have an audience in their meetings (my apologies to any digital product marketing communications super manager who felt offended). 

    Now that the periods we won't be doing actual work will be even longer, fewer positions will be available. Why hire someone for a technical position if there isn't very much to do there? Why not hire him with a fancy title and use him as audience in meetings instead?

    The need for developers will drastically decrease, though never cease to exist, for the following reasons:

    1. We are often busy debugging and adding features to existing projects, and it is extremely unlikely that AI will be able to do it anytime soon. This will keep us busy for at least several years to come.

    2. Even if de-novo development  is carried out exclusively by machines within 10 years, someone will still have to check if the codes are safe to use. 

    3. Corporations will always need someone to bear legal responsibility for bugs that result in data breaches or lost money. 

    4. Corporations that have human developers too will be de-facto more trustworthy than random startups that just use generated code. 

    5. Someone has to program the AI platforms. Relatively few programmers will be active here, but there will be some.

    But AI will dramatically decrease job postings for developers, not to mention that many of us may switch to other professions literally because of boredom. Checking and refactoring generated code has nothing to do with the creative and joyful process of creating the code from scratch, solving problems, and knowing that the product is your creation.

    This is of course nothing that is very likely to happen withing this decade (not even job positions for train drivers and cashiers are drastically decreased in the last 30 years). But it will happen pretty long before we, millenials, are about to retire.

    You can use your brain's capability for abstraction and use more or less the same logic for all professions. Legal reasons and trust issues will always keep humans busy, so there isn't a danger that we are replaced. Instead, there are the very real dangers of us ending up being the constantly bored and depressed supervisors of the machines, or the equally bored and depressed audience of managers and CEOs talking to meetings. 

    That is the most likely (and best case) scenario. The less likely and worst case scenario is that mass layoffs are coming, and we all imagine the consequences.

    In either case, with the current economic model, AI will not be used to make our lives better.


Δεν υπάρχουν σχόλια:

Δημοσίευση σχολίου