AI Markdown (AIMD)

Edit: not called embeddings due to confusion with the more common use of the word embeddings in relation to AI, and not called markup due to AIML (AI Markup Language) already being a thing since the time of A.L.I.C.E

So, here’s the deal: the tech world is changing, and it’s changing fast. It’s like when the internet first hit the scene in the ’90s – subtle but game-changing. And in this new landscape, it’s all about programming, specifically how we scale it. I’ve been down this road before, following the tracks laid by folks like Alan Kay. His take on scaling and programming, especially with DSLs, had me convinced they were the next big thing. But, you know what? I think I was barking up the wrong tree. AI, with its knack for breaking down complex stuff into simple explanations, is the real MVP here.

AI’s not just about fancy tech tricks; it’s a serious time-saver in loads of areas. Take my current gig, for example, where I’m whipping up an automated proposal generator. But let’s zoom in on something that really hits home for developers: how AI can make our lives easier and maybe even a bit happier.

Writing up detailed comments for functions can be a drag, right? But AI’s got our backs here. It can figure out what’s what in your code and spit out these spot-on explanations. The catch? This approach is static, and in the AI world, standing still means falling behind. Plus, let’s be real: keeping these comments up-to-date is a pain, and it can be a real downer for us devs.

Enter the concept of “AI Markdowns” or “AI hooks.” This isn’t just some fancy tech jargon; it’s a shift in how we mesh AI with our coding routine. By Markdown AI in the compile or documentation phase, we’re looking at some pretty sweet perks:

  1. Keeping Up with AI: AI’s moving at warp speed, and Markdown it means our code does too, staying fresh and relevant.
  2. Tailor-Made AI: We can tweak AI to understand the context of our projects better, making it more on-point.
  3. Speaking Human: We keep instructions human-friendly, which is right up AI’s alley these days.

Sure, there are hurdles – costs, a bit of unpredictability, and some lag in processing. But the payoff? Huge. We’re not just talking about smarter docs; we’re reshaping how we write and think about code.

Now, about the nuts and bolts: we need a straightforward Markdown language, something like Markdown, to weave AI into our code. To start, we’ll run it before compiling, so it doesn’t slow us down.

Here’s the kicker: someone’s going to crack this code, and it’s going to be a game-changer. It might be me, it might not. But either way, I’m diving in headfirst with my current project. I’m putting together this AI code-gen step, mixing it right into the development process.

Keep an eye out for more updates from my end. I’ll be sharing the nitty-gritty, the successes, and even the stumbles. We’re on the brink of something huge here. It’s not just about coding smarter; it’s about redefining the game. And that, my friends, is what makes this ride so exciting.

Here is a teaser of what the internal stuff I am using looks like

def foo(bar):
    # AI:template generate_docstring
    # AI:context function
    # AI:target inline-above
    # AI:param insertModificationDate=True
    ... the code ...

Generates output like

def foo(bar):
    """
    :param bar: The Bar object representing the bar food

    :return: The ID of the generated document.

    The `foo` method applies a template representing bar foods
    """
    # AI:template generate_docstring
    # AI:context function
    # AI:target inline-above
    # AI:param insertModificationDate=True
    ... the code ...

If you want like metacomplication removing the aI bits, instead of inline-above you could do replace.

AIMD  AI Markdown  Tools  Automation