Starting somewhere with Generative AI

We're all being bombarded with 'AI' this and 'AI' that at the moment. It's exhausting, annoying, over-hyped and bad for the planet, but is undoubtedly going to be everywhere and may eventually be accurate enough in certain use cases to be useful.

Generating emojis or fake videos showing nightmare fuel is fair enough, but since GenAI is really just Machine Learning using Large Language Models which in turn is really applied statistics, there are places where the limits of the models are useful enough that they should be able to save a lot of time and perform repetitive tasks, especially ones based on searching, that it's worth spending a little time investigating them.

So over the next couple of weeks, I'm planning on investigating the best way to take (or generate) a list of risks, threats, controls, policies around cybersecurity and see if the time spent on learning and feeding the model is quicker than modelling it by hand.

I'lll start with Amazon Bedrock, since I need to learn about it anyway, but might revert to ChatGPT since it's probably going to be quicker, which, since laziness is the whole point in this, is probably more what I want...