Ask HN: Is AI going to replace web developers? | Hacker News Reader

Ask HN: Is AI going to replace web developers?

8 pointsud08 comments

Looking at the trends, AI seems to be breaking new boundaries but is it possible that web development jobs in the future may be handled by AI?

View on HN

TL;DR: Yes it's possible, but it doesn't actually matter if it happens or not.

An AI creating websites is definitely possible, but will it actually replace people? I don't see that happening (at least not at a large scale). It's much more likely to be a tool that developers use rather than a tool that replaces developers. Or possibly something similar to Squarespace. It might be able to manage content on its own and build webpages, but what happens when there's a bug? What happens when you want to innovate and create something new? An AI would have trouble accomplishing that, so web developers would still play a role.

Now the more important question in my mind isn't whether or not they will replace jobs, but rather does it matter if they do? And my answer to that is no. There's two scenarios. A scenario where AI completely eradicates all developer jobs, in which case AI has probably eradicated a majority of today's current jobs and we would adapt as a society. Either by instituting some changes like Universal Basic Income or by developing new job markets (probably geared more towards the arts or things we can't yet imagine). The second scenario is that AI is introduced more as a tool or a simple solution to web/app development. In that case, there will still be a need for devs to do the more complex stuff. There will also be other development fields to move into, like Machine Learning Engineer for example.

Doubtful. Most of development is discovering, understanding, and refining requirements that may not all be known in advance, and then translating those requirements into deliverables. This is an intricate process subject to information asymmetry, and AI/ML and related fields are fundamentally reliant on a trove of accurate data.

I think it will be another tool in the toolbox.

I.e. I remember that few years ago, large part of web-dev was getting a page design in image files and you sliced and diced them and created proper html, maybe with few placeholders, with css optimized both for desktop and mobile.

Then I have seen a thesis on my uni, that automated all of that with genetic algorithms. You just threw an image at it and few hours in it spews out optimized web-page. I wouldn't be suprised if current photoshop has a button for it :)

Tangentially related to this point, when tools like Rails came along, it seemed that web applications would become trivialized to the point where "anyone can do it". In an environment where anyone can be a web developer, do you need a web developer? It turns out you do, fortunately.

Now, Artificial general intelligence (AGI) is a threat to everyone's jobs, just about, but I can't see an AI short of AGI replacing a web developer's cognitive processes.

It won't likely replace people any time soon, but if it's anything like other technologies, it will make workers more productive (reduce jobs) or the process easier to learn (larger pool of worker -> lower wages).

Very probably, if web development is seen as a closed game with a limited set of rules and numerable outcomes.

Yes it will. The question is, when?

You may have heard of The Grid, a new crowdfunding campaign that would revolutionize website design by incorporating Artificial Intelligence as a design tool. The service is similar to SquareSpace or WIX but include algorithms that essentially analyze a user’s content and imagery to calculate the ideal website design based on a set criterion. Any new content that gets added will result in the entire site reorganizing itself on the fly. Sound too good to be true? Perhaps. But interactive designers and developers have been worried about pre-made templates and DIY client services for years. Some of us fear being replaced. Some of us welcome our new robot overlords. Some of us, myself included, see these new services as a chance to evangelize that there are extremely deep, unique User Experience challenges that a web template (created by a bot or not) just can’t solve. In the Beginning Human-Robot Interaction has been a topic of both fascination and fear since the time of antiquity. Homer’s Iliad described the creation of automatons by the Greek god Hephaestus, who built consciousness into golden statues. Aristotle proposed that human thought and logic could be mechanized. Nineteenth century inventors and artists perfected self-operating toys and machines. These concepts would be lifted to a new plane with the invention of the programmable digital computer in the 1940s and subsequent founding of the field of AI research at Dartmouth College in 1958. Our relationship with AI has always been pretty tenuous. As in all instances of paradigm shifts and technological revolutions, we’re at odds with making our lives easier while feeling a dull sense of displacement or irrelevance. Harkening back to Isaac Asimov’s Three Laws of Robotics, we’ve attempted to safeguard ourselves from AI completely replacing or harming us. But replace may not be the right sentiment as it relates to the Web. We’ve seen design trends, languages and methodologies come and go. Most often they’re superseded by far more efficient ways of doing things and conveying information. Could the introduction of Web AI fall into the same path of progress or are we entering a new era of design? Not for Everybody The Grid promises to free us from the shackles of design and code. It promises that all of our problems are to be solved via a set of constraints and templates, whether they’re conceived from an algorithm or otherwise. While I think AI and template sites will benefit from small businesses and casual users who are on a tight budget, they completely ignore all of the work that User Experience designers and information architects have done in solving problems from an empathetic, strategic and tailored point of view. When we talk about web design, we’re talking about something much more far-reaching than what the end-user sees. Since the design is freed from static, printed paper, the screen is able to contextually change what is displayed based on environment and feedback. How we get to that point of context is the task of a UX designer. Defining who the user is and his or her goals, and what a client’s business goals are, helps inform the final design. It’s beyond the ability of an algorithm to figure that out. We just aren’t there yet. We haven’t successfully taught machines how to think and behave like real human beings, so until we can do that, we have to do the dirty work ourselves. And web designers love those really tough problems. All of the nuances, the complexity, and the compromise. Another issue with an AI-based website is how much control we really have over its output. You essentially abdicate the big decisions for the system to make, resulting in a site based on its own assumptions and programming limitations. (Side note: the code the AI spits out is so poor that you’re likely going to be dinged on SEO crawls!) I don’t think we’re ready to hand over our thought processes as they relate to defining a unique face and voice online either. Although interface conventions are becoming more homogenized, the need to outline a brand’s differentiators in the interactive space are at an all time high.

Reply on news.ycombinator.com