This post has broader implications for software development beyond the Drupal community, but I feel fortunate to be part of an open source community that can lead the way in addressing the widening productivity gap among its contributors and maintainers.

The title of this post is meant to draw you in by highlighting a problem, but my goal is to get us thinking about a solution. I realize the term "not-so-great" may sound negative when describing a developer, but this comparison bluntly highlights a major problem developers and communities face when working with AI. The truth is, I have never met a "not-so-great" developer in the Drupal community because people are engaged and curious about the software we build.

My realization is that "AI is making great programmers even greater and not-so-great programmers, well, not-so-great."

For me, a "not-so-great" programmer is someone who writes code like a factory worker. The difference between a "not-so-great" programmer and a beginner is curiosity. Curiosity is the secret to being successful with AI. A curious beginner can easily accelerate their learning experience with AI. Anyone with curiosity can move from beginner to novice in a matter of hours with AI.

Everyone agrees that AI can be a force/capability multiplier, ranging from 2x to more than 10x. The reality is that some people are simply unable to leverage AI and have a 1x multiplier. Very experienced developers report they can now accomplish tasks that would have taken months in days or even hours. Observations suggest that the more capable someone is, the more effectively they can leverage AI.

Let's say we were rating programmers on a scale of 1 to 10, using a system similar to a chess rating system, with 1 being a beginner, 10 being a legendary programmer (aka a super grandmaster in chess), and 5 being an advanced programmer. Frankly, a not-great developer wouldn't even be ranked.

Here is a 1 to 10 ranking for programmers

1. Beginner
2. Novice
3. Intermediate
4. Capable
5. Advanced
6. Expert
7. Master
8. Elite
9. Distinguished
10. Legendary

Now, you apply AI as a capability multiplier to programmers, ranging from 1 to 10, and apply the AI multiplier to their ranking to get an overall productivity score. The beginner programmer at 1, when working on AI, might see no productivity gain, meaning their productivity gain is 1x. The advanced programmer, when working with AI, is at 5 and sees a 5x acceleration, increasing their overall productivity from 5 to 25. Now, the legendary programmer at 10x acceleration is increasing their overall productivity from 10 to 100.

I know it seems crazy that I am suggesting some legendary developers have gone from a 10 to a 100 in productivity using AI, but there are some anecdotal stories worth noting.

Legendary developers are recalibrating.

This entire thought experiment underscores the reality that AI may be creating an exponential productivity gap in software development. My hope is that, as an open source community, we address this widening gap honestly, openly, and publicly.

The widening productivity gap in software communities is creating three situations

• Developers are being laid off
• Developers have to embrace AI
• Junior developers are not being hired

There is no easy way to say it, but some people are going to lose their jobs because a senior developer, especially a very talented one, can use AI to automate their work. The reality is that AI is not replacing someone's job; rather, someone using AI is replacing someone who is not using AI or not using it efficiently. In the long term, there will be a lot more software produced using AI, and there will always need to be a human in the loop, especially talented developers.

Still, these layoffs are inevitable, and those who want to be rehired will need to retrain and adapt to using AI. For Drupal, this could be an opportunity to grow our community, especially if we embrace AI.

To keep our jobs, we will need to embrace AI. Specifically, to grow our community, we have to adopt AI in a way that supports one another and those joining our community. Everyone will have different levels of AI capability and productivity. What frightens me about the 100x difference in productivity between a Drupal grand master and a beginner is that the beginner will be left behind, and most people will feel like impostors when they compare themselves to a Drupal grand master using AI.

Discussing and sharing our experience is the first step toward collectively adopting AI as a community. I see some developers rocketing around the earth, with multiple agents running and building things. I'm in the middle of the productivity scale, and if those above me are exponentially more productive, I am replaceable.

I feel replaceable, and I don't expect to be 10x more productive with AI. With my experience, I can help onboard new developers to the Drupal community and support them in becoming more productive. I am old enough and experienced enough to happily mentor someone who goes on to build bigger and better things than I ever would. Mentoring and embracing junior developers using AI might be the Drupal community's biggest opportunity.

Developers are being laid off, and companies are citing "AI efficiencies." This reduction in force is also leading companies to hire fewer junior developers, since senior developers can use AI to do the work of several junior developers. I am not really worried about a curious junior developer quickly leveling up to an intermediate/capable developer, as long as they use AI as fuel for their curiosity.

Still, I think companies will be short-sighted in their hiring of junior developers, defined as those who code, debug, and maintain software under a senior developer's guidance. AI has replaced basic coding, debugging, and maintenance; we will have to expect more from junior developers and from the senior developers mentoring them. Junior developers will need to use AI to learn to code and write code, and they should start with "vibe coding." Senior developers will need to provide the tools to ensure that "vibe coding" is productive, secure, and scalable.

The Drupal community takes great pride in the quality of its code, and Vibe coding could produce "AI-slop". At the same time, I've been able to easily de-slop code with skills and code review. One of my recent AI "aha" moments was posting feedback on an AI-generated MR with some AI-slop, and the AI fixed everything I asked for. Personally, I am not a fan of the term "Vibe coding", but I am okay with the concept.

My definition of "Vibe coding" is the ability to use AI to build software without prior knowledge of how to "code". The most important skills for vibe coding are having an idea and the curiosity to build something; AI is a tool that helps you build it.

I am a self-taught programmer who built my first personal website, https://jakesbodega.com/, using Microsoft FrontPage, creating an interactive experience that still works today. Even with MS FrontPage, I could peek at and tweak the HTML source to create a crazy, nested-frame page: a virtual tour of the Big Blue House and Gina as Mrs. Potato Head.

I got my first job because my prospective employer didn't care how I built my website; they saw that I was curious enough to figure out how to build something and that I learned how it worked under the hood as I built it.

The fundamentals of being a good coder, or a vibe coder, are curiosity, patience, and practice. It is very likely that we will see more brilliant software applications built by people who don't know how to code. In Drupal, we are open to "no-code" solutions for our UI, but we need to consider "no-code" contributions to our software, while still recognizing that there is always value and, frankly, a real superpower in knowing how things work under the hood.

"Vibe coders" can't be scorned by our community, and we need to figure out how to include them. Vibe coding could be a tool for onboarding new people, and we need to talk about it further.

A massive productivity and employment gap is opening in the developer and software community. With Drupal, we can work to fill it and help people avoid falling into it.

To survive and succeed, we need to evolve as a community by embracing AI. We also need to do something challenging and unexpected; throw out our playbook and preconceived notions about how software is built, and take Drupal and its community in a new direction. Our motto, "Come for the code, stay for the community," can be revised to "Come for the community."

Personally, I am done playing with AI, and this will be my last Drupal (AI) Playground blog post. Next, I'll start "Vibing Drupal" and explore what I thought was once impossible and is now possible using AI.