AI hype

There's a lot of hype around AI right now.

At my current workplace, the hype is relentlessly generated and pushed from the top down, seemingly every day.

Apparently, we're now an "AI-first" business. I don't recall there being much notice about this—mostly whispers and rumours.

Then suddenly, a new in-house generative AI framework was introduced. Only a select few engineers were privy to its ideation and creation.

This new framework, and the topic of AI in general, is now consuming all the air around us. Every week, we're being asked to show what we're doing with AI, as though anything else is irrelevant.

I think a lot of us saw this coming. I just don't think we were all prepared for how intense the push would be or how sudden the change would be.

The FOMO is real.

Threats

I don't have an issue with AI as a tool. I do, however, believe it's a tool that requires responsible and respectful users.

It was a shock when I started getting replies to comments on pull requests that were clearly AI-generated. It wouldn't have been so bad if the user had identified themselves as an agent. At least I would have known what I was interacting with.

But the developer was communicating with me as though I were speaking to them directly.

It was a weird and disturbing experience trying to engage and interact with a human but getting AI-generated responses—responses that demonstrated a clear lack of understanding about the points I was making.

In this case, I think the developer was hiding behind their AI agent, hoping it would help them out.

Then there's the slop.

I've seen some dodgy code produced this year—code pumped out by developers who have put complete faith in their agent, with no idea of the risks inherent in the output.

In some cases, it's understandable. When the business is pushing the message of velocity at you every day, the pressure can build to deliver on these expectations, and this pressure can lead to lower quality output.

Is it possible to meet these expectations? I don't know—it's too soon to tell.

But augmentation and the transition process need time. Especially at large tech companies with a tonne of legacy systems.

Successful integration and use of AI isn't going to happen overnight, and it's going to require some deep and balanced thinking to get right.

Opportunities

At first, I resisted augmenting my workflow with AI. Part of my satisfaction as a software developer comes from crafting code.

But I see where things are going, and there's no point in resisting.

The question is how I can make the process interesting and satisfying for me in this new world.

I'm side-stepping the in-house agentic framework. In my opinion, it's just a bunch of markdown files.

It's the quality of the content that matters most to me, and if I'm being completely honest, I don't really trust the quality when it's open season.

Sure, I can contribute to the source, but there's a weird political and social-media-type reality surrounding the framework right now.

Contributing to the framework is definitely not my buzz at the moment, and I'm trying to stay clear of anything that might mess with my hauora.

In saying all that, I understand why it exists.

Our executives want to ensure widespread AI adoption, and the easiest way to do this is through an opinionated framework.

What's most interesting to me is developing my own set of AI tools to improve my personal process and workflow.

The more I work on AI skills, the more I realise how loose the results can be. I'm used to deterministic output.

So this has been an interesting challenge: finding the right balance between what should be deterministic scripts and what can be left to domain-specific context and knowledge.

I feel like I'm gradually discovering a place for me in this new world—a place where I can continue to think deeply about problems while also increasing the speed at which I can produce support, reviews and solutions.

Speed has never really been a personal goal of mine. But it's clearly a goal of a lot of businesses in this new AI world we're in right now.

Personally, I feel the true value AI brings is not in its generative capabilities, but in its ability to help me think through problems and bounce ideas around.

Approaching discussions with agents from a critical perspective, and with a healthy dose of skepticism, I am starting to see the value of AI in my personal process.

Reflections

In the end, perhaps the real challenge isn't resisting the AI hype but finding our own relationship with these tools amidst the noise.

While businesses chase velocity and executives push adoption, I've discovered that AI's greatest value comes not from replacing our thinking but enhancing it—when we approach it with the right balance of openness and skepticism.

The hype may be relentless, but our personal approach to these tools can still be thoughtful, measured, and ultimately, our own.