To gain understanding for the future, look at the past

“AI” is changing the way that people live and work. This is particularly true of my occupation as a software developer.

To try and predict the long term effects of AI on my occupation and on other creative fields, I think it’s helpful to think about times in the past when people’s jobs have been replaced by cheaper labor.

Book reproduction

Before the printing press, books had to be reproduced by hand. A lot of the time this was done by monks (the Bible was and still is the most reproduced book). A large book like the Bible was expensive because it took a very long time to create one copy. Sometimes the reproductions were simply the words written legibly. Other times they were done with more style and craft — with embellishments and imagery drawn in (example). These monks were experts in their craft, but their occupation was made obsolete by the printing press.

When the printing press was created and production capabilities dramatically improved there were large effects:

Most, if not all, people reproducing books by hand eventually lost their job.
Books became drastically cheaper.
Book design became more similar and less unique.
Owning books became less of a status symbol.
More people read books.

Are there still jobs in book reproduction? Obviously. There are authors, publishers, and smaller print studios (which are great! Try out your local one). Additionally, the artistic side of book reproduction lives on through hand lettering (example).

Photography

For a very long time, the only way to have an image of something you cared about was to have someone draw or paint it. For anything past a crude drawing, this was prohibitively expensive.

When cameras became cheap and commonplace the effects were similar to book reproduction: lower cost, increased access, and arguably a lower quality on average.

Are there still jobs in paintings of real things and photography? Yes, though it’s hard to establish yourself in these areas (i.e. the barrier to becoming a professional is hard). But also the ability to practice the craft is cheaper. You don’t necessarily have to go and learn from an established person, you can practice with the tools and resources that are cheap and develop your skills.

Car manufacturing

As car manufacturing has matured, the average car from any given company has started to look more and more the same and has a similar quality bar.

But again there’s a very sizable market of top tier cars that are in their own class in terms of cost and uniqueness.

Cobol programming

This is a bit different of an example, but I think it illustrates a different set of effects that should be considered.

Cobol was one of the earliest programming languages. Many of the first institutions to digitize built their systems around it. As time went on, most companies switched to newer languages and tools but some didn’t.

Today, these companies still need to get people to work on these Cobol systems, but there are fewer and fewer experts working in the language.As such, those experts are very well paid.

Will Cobol be around forever? No. But Cobol programmers can make a very good living in this interim period.

How does this history apply to the future of coding and creative work?

Without question, “AI”, meaning at least the use of LLMs, has and will continue to shape the future. As a programmer and a creative, I am exposed to its effect in the world around me and I am in communities where professionals are grappling with its implications.

What are some of the big picture effects of this transition? Time will tell for sure, but here are my predictions:

Lower barrier to entry of creating low quality goods with AI

Most of the time, everyday people and purposes don’t require goods and services to be finelytuned to their needs. For example, people generally don’t need a custom pair of underwear. Are custom boxers a fun change? Do pregnant or breastfeeding women need better bras for their needs? Yes, absolutely. But on average the generic piece of underwear is sufficient.

The same principle applies to our digital and even creative lives. Most people are fine using apps created by others for mass consumption. Even most one-off apps can be created using a WYSIWYG editor or, more likely these days, an LLM. Non-coders can and will be creating applications, images, videos, and music like this for themselves, specific to their needs.

But they will lack the craft that true professionals can provide. In most situations that’s an acceptable tradeoff.

The barrier of entry for becoming a professional (in fields where AI is prevalent) will be greater

Why hire a junior designer when you can create mockups with some prompting more quickly? Why hire a junior software engineer when you can get the same output faster with fewer mistakes?

AI makes the decision to hire junior professionals an even more long-term decision, because the initial cost is even greater. Unfortunately, most companies these days do not hire people with long-term thinking in mind. But if all companies exclusively rely on AI for junior-level work then the professional pool of senior workers will eventually shrink and there might be a lack of good seniors.

A whole article, or series of articles, could dive more into the effect on people trying to get into software engineering and the effect that AI has on software engineering personnel development. I’m not sure what will happen, but this is definitely an area that I’m interested in following.

The cost of experts will go up

People will create apps with AI and then run into issues that the AI can’t fix. Consultants to fix this will be comparatively expensive. Codebases will be generated that then have to be refactored in large ways and that will be costly. Perhaps greatest of all, experts will be able to produce work even faster than before, so the value that they bring is greater. Thus they can charge more.

In the creative world, some companies will be happy with the small touches of flair that LLMs can add. Many will probably be happy with AI-generated images. But companies that really care about craft and pushing the bounds of what is possible will still hire people or agencies to add the polish and creativity that only people can add. These people will probably charge even more for these services because some of the simpler work that used to help them pay their bills will no longer be available to them.

Taste matters more

People who have taste and use AI effectively to accomplish their vision will be the most successful. Being a person who can create a clear vision will become even more valuable.

Brand matters more

In a world where most things are generic, things that are truly unique will stand out truly more. One of an infinite number of examples is making everything hand-drawn like the video game Cuphead. I think I can safely say that an AI will never have the originality and ability to independently think of and create something like this by itself!

Individually, your personal brand will be increasingly important because some of the things that used to define us (writing performant code quickly, being a Figma expert, etc.) will not be as important any longer.

Many companies will expect even tighter deadlines

When you can create an MVP of a product in a few minutes, companies will expect that you can take it the rest of the way in a short period. People are notoriously bad at estimating how long “the last 20%” of the work takes. Timelines will get even shorter for most work, which will be particularly harmful for work that is truly new.

We will have to spend more time defending estimates and efforts, outlining exactly why it’s necessary or better, sometimes using qualitative data and appeals to emotion as support.

In the AI world, how am I doing things differently?

At work I use an LLM (GitHub Copilot) regularly to write code more quickly. Occassionally I use it to give me a first draft of copy for a particular warning message or content block. Even more rarely, my team gets to implement an AI feature for our customers to make use of, usually to try and automate the initial creation of something (an email, a form, a survey, etc.).

In my free time, I’m more willing to try out some different project ideas with AI because I can get off the ground more quickly to test the idea. Recently I’ve been trying to be more hands off and let the AI drive almost all of the programming (using Augment).

I’m also trying different programming languages and libraries that I haven’t used before and might not have ever used, because I have a little helper to point me in the right direction without me having to read through all of the documentation at once. Sure it gets stuck and I have to figure out how it actually works, but I am able to find those places where I am not understanding something more quickly.

Tools like AI are not innately good or bad. The world will change, as it has since the beginning. The AI cat is already out of the bag. There’s no turning back to the pre-AI world so long as computers still exist and develop.

As professionals, should we be using the tools in spite of the questionable ethics of training, the effect on the environment, and the occupational risks that come with them? This is an ethical question that each person must answer individually.

My view is that we probably should use them because we need to make a living and if you don’t evolve / adapt you die (at least professionally). With that being said, I think we should try to personally use them responsibly and set norms (and help set policies) of using them responsibly. That applies from an application level to how much energy and water they use.

Am I sad about some of the things that will be devalued in a world where AI is commonplace? Sure I am. But there are some benefits to it as well.

We might look back at the pre-AI time like we now look back on the monks reproducing the Bible. We made some beautiful work. But it took a long time. AI will help us create more things more quickly, even if some of it is a bit less personal and lower quality.

But just because we can make something more quickly doesn’t mean it has to be soulless. Putting effort into the right things will be an even more important skill. Caring is still the key.