The State of AI

Day 247 / 366

This blog started as an AI-only blog. My goal for the year was to learn more about AI. And to be consistent about that, I thought of writing about it every day, so that I would be forced to learn or apply something new daily.

But slowly I lost my interest in AI. We all know that AI has considerably underdelivered. Any new AI product coming out these days is what we call in the tech space — “Vapourware”. These are products that look great on paper and in demos, but they are never actually shipped to the general public. Even if they do, they are too basic and lack most of the promised products. And so, I pivoted this to a normal daily blog, where I wrote about what I felt.

In the past few months, I’ve had the chance to work on freelance projects that used AI and LLMs in really creative ways. This has convinced me that while LLMs might not be the scary or mind-blowing thing that we were promised, there is still scope for using them to improve our existing tools.

Take summarization for example. Every day, I spend time on YouTube, Twitter, and Reddit to get my news and updates. If AI could fetch this data, and summarize it in a nice report for me to quickly review, that would be a huge time saver for me. And there are countless more ideas like this.

So I have decided to give AI another try. And I will again start working on some sensible products using AI that could make things easier for us.