The AI is even better thought companion than I hoped or imagined

I wrote in the previous post about the limitations of LLM's that I experienced. In this post we'll talk about the strengths and the parts I absolutely enjoy working with.

First, this is going to be only about Bard AI, I haven't worked with anything else and therefore am unable to speak to that. Second, I've also tried local LLM's but it's clear the large models are better in many subtle ways, but I'm not an expert to tease out the details.

One thing Bard is quite good at is following templates, enforcing them, and creating essays and articles based on that. So I can give it a template, I can give it the 'meat and potatoes', and it'll write the essay in that format. Mechanical organization is great. Idea generation using LLM's is also one of the better use cases. If I want to do a project, but have no idea where to start, this is where I'll start.

Speaking of which, the models are great to bounce ideas against, because they really force you to clarify your thought, your intention and goals, and how you want to present them. For example, I wanted to write an essay on how exploring unrelated technologies can prove useful for companies and employees. The bard ai explained to me I can't just say 'companies or employees' as the consequences and actions plus the recommendations to make would be different, it gave me an example of how making my primary thesis to be targeted at companies would go toward a direction away from one that was targeted at employees. Then it kept asking me more and more questions until I had a really good idea about my thoughts.

Second example, I had this idea about improving cleanliness in national parks, but my thoughts were jumbled, unclear and not directly implementable. As I answered the questions Ai asked me, my thought got clearer, I came up with new ideas, and eventually it solidified around a context that was way more interesting than what I originally thought.

Third, I wanted to explore long-term tourist rental idea for Nepali villages, and had that 'conversation' with the AI. The quality of the questions it asked was really good, and it felt exhaustive. I talked about thiis to PS whom I met yesterday and he said he would have asked me the same questions as a consultant to tease out my ideas. And his job is consulting! So there's definitely lots of 'bouncing and evaluating ideas' opportunities with AI.

Maybe I'll use it to help me write my fiction one day. Who knows if fiction will even survive by then, by the looks of it things don't look great.