Remembering

Currently I'm finding that we are having somewhat a slowdown in new type of devices and technology. It looks like it have been some years already with no major change besides the Steve Jobs Iphone. Many major innovations are coming from software and right now generative models like LLMs (Large language models). But new devices, I haven´t seen different ideas. I know that probably there are some hidden ideas that come and go but they aren't lasting.

Discussing this point with some friends, one suggested me that this new developments are getting so complex that it is difficult and slow for teams create new things. It seems to me that these new LLMs like the GPT, Gemini and etc are coming to alleviate this difficulty. But we are somewhat behind the schedule then, because we will need first to understand these points, before moving on.

I'm currently reading the book (Exhalation from Ted Chiang)[https://www.amazon.com.br/Exhalation-Stories-Ted-Chiang/dp/1101947888] and one of the stories re reflects in many points about Remem a software that you can use you entire lifelog as a source of research and reflection and I couldn't sat back and don´t do a parallel with the fact that this software really resembles what LLMs can do with the input of texts and images. Sometimes I really imagine what this can do with access to a big context and a person with a huge curiosity and a potencial problem to be solved in his hands.

This also gets me thinking in things that I don't want to remember exactly as they where, sometimes the mere association between the fact and the description of the fact is enough, these mangling that is done in our brains bring in many times joy to review mentally these ideas, that is what make our stories!

People are made of stories. Our memories are not the impartial accumulation of every second we’ve lived; they’re the narrative that we assembled out of selected moments - Ted Chiang

But returning back to the point, by using LLMs we can leverage knowledge that otherwise would´t be accessible to us. Besides working with code I always had some difficulty doing rewriting to get the most optimized code like re-writing to get a O(n) version of some loops or optimize some processing types to the given processor that we are working on. Last week I was looking into it and probably by using LLMs I can do this in a faster way and also make changes between languages. The secret lies on the prompt.

Probably we are on a point that working with this tools will began do be second nature, and we should learn how to interact with them.

To dive more into this topic you can go to Code Helpers