— thinking out loud — 3 min read
I Skate to where the puck is going to be, not to where it has been - Steve Jobs
Hi,
I am Aman, and this safe space is where I am most raw with my thoughts, hop on if you'd like to interact with them :))
I am sure by now, many of you must have interacted with a version of LLM either in the form of chatGPT or in the form of an abstraction,
Github Copilot comes to my mind when I try to think of the useful use cases.
2023 is an interesting year to be an engineer in my opinion, the wave of AI applications being built is just mind blowing and I am very much excited to see the kind of interface problems this shift solves.
Here's my take on this category in general:
Interface Problems ⊂ Technical Problems
I have been somewhat obsessive about tracking my history on the internet, since the lockdown of 2020 I have taken various stabs at this obsession and I think I have found a model which better suits me now.
I have been manually tracking all the content I consume since last 1 year with about 90 - 95% accuracy.
Open Source Content Consumption
The thought of doing something like this was first inspired by this tweet, which eventually turned into an experiment
I am not first one to be doing this people like Julian, have been doing this long before me
Tracking personal data as a concept is not new, in fact it has been around for a long time now, but why is this important now?
I believe in 5 years from now, everyone will have their own personal search, which will scan their internet history and cumulate result on the basis of that.
with LLMS interacting with the web becomes more like a magical experience than it has ever been, some experiences which I can think of now are
Better Recommendations on your data, I have already fallen in love with Spotify smart shuffle, now imagine a model 10x better doing that
The internet serving as a time machine and reminding us that we have been to a particular site before, this just makes internet more multi-dimensional I believe.
Better model for creating clusters of people on the basis of content consumption, which will make it easier to find interesting people in a particular domain/niche.
Geographically mapping similar people, resulting in physical spots
As most of you must have guessed by my blogs that I am a fairly optimistic person and choose only to look at the world from an optimistic angle and I have projected the reason for doing so here:
Polarization of protagonist as a way of outlining optimism || pessimism || dystopian outcome
I have expanded a lot on the purpose of technology helping us move closer to fiction,
At this point it's fair to say that I am very much excited by the power of LLMS and what it enables for people living in the internet
Stories like Upload are not far way from becoming a reality.
Expanding further on this notion, I think personal databases will soon be a commodity,
A new marketplace of tools is emerging, we will see more and more vendors allowing us to play with our data, just like upload, with that being said
I will be iterating on the idea of making our internet history multi dimensional, probably a market place where you can plug in your data and build all kinds of schtuffs.
Iterating on personal searches
Having personalised FAQs
Multi dimensional bios on the internet
Distributing personal recommendations
If I start to think about it, the list is very long,
My goal with this blog post is to attract smart people in my cluster to help me build it, so to anyone reading this blog who shares the same curiosity, hop in the rocket ship ;)
There two intersections where I'd want to be working in the future are very clear to me now,
I think this blog will see more of my raw thoughts taking shape, definitely drop me a Hi, if I made you at least a little bit curious ;)