Learning AI prompting

I'm currently scratching the surface of this AI world. As a novice, what started off as this scary place that I had no business being in, has become somewhat of a favourite past time. Now if you'd told me 6 months ago that I'd enjoy learning about LLMs, I'd had laughed in your face.

Now to be clear, this is neither a tutorial nor a deep dive into Large Language Models. It's a point of view through the lens of a true beginner at the start of their journey.

Now like a lot of folks, I was using these LLM's like a search engine. But they speak a language that if learnt, unlocks a whole new world. This language is prompting. Think of a LLM as a speakeasy and prompting as passwords. The passwords are: articulation, clarity, and specificity. It's not that deep but you get the drift.

Basic inputs receive basic outputs so there's a bit of finesse required that come in the form of prompt techniques. I don't have the expertise to explain this in detail but a quick online search will lead you to people who are authorities on the subject.

While ChatGPT may be the most popular there's a long list of others. Each one with their own unique specialty. Gemini, Grok, DeepSeek, Claude to name a few.

So what are people using LLMs for at the moment?

There's a long list but here's a few.

Want a LUT for a video? Code for an app? A personal training plan?

How about a business strategy? Market analysis? Technical docs, emails, newsletters?

It’s all on the table.

How about a novice like me?

I'm using it to learn how to write. I'm not sure if this is a popular method but I'll give you a rough outline of how I go about it:

I come up with a blog idea, like this one about AI Prompting. I write my first draft, copy and paste it into an LLM to assess whether it's achieved my goal of clarity, precision and impact. I'll take the feedback it provides and make any adjustments I feel are necessary. I'll repeat this process until I end up with a product that I'm satisfied with. I'll take the end product and run it through other LLMs, only this time I apply a prompt technique called "role and audience". This is where you assign a role to the LLM and a target audience. In this case, the role would be an expert writing critic and the target audience would be a mix of casual readers, creators and people like myself. I'll then ask the LLMs to provide a rating based on the roles I've assigned. Once it reaches an average that I'm satisfied with, I publish the piece.

This sounds like a lot of work but it's actually a fun exercise and one that's helped me improve my writing skills. Now is this the most efficient way? Not sure but that's the beauty of being a novice. You wander around a new place trying new things until you either get it right or run into someone who can point you in the right direction.

Is this cheating? Depends on you ask. I'm not a purist and I'm still learning the ropes.The introduction and use of new technology always results in split camps. The closest example I can think of, while overused, is the internet. I remember my parents generation calling it a fad. Now they're all streaming Netflix and doom scrolling tiktoks. Change will never lose its fangs. Ultimately I do believe that this is just another tool. The soul of creating will always lie in the hands of the storyteller. An arrow will fly where the archer points it.

There's still so much to learn and it's a slow process. But the speed this technique is moving at means that there is no perfect time to start learning it. It means that everyone will be learning in real time as it evolves. Suprising thing? For someone like me who's always been a bit "meh" when it comes to technology, I'm actually enjoying this.

Funny thing this learning business.