Though not advisable, I spent unnatural amount of time, thinking abt capitalism VS communism; about the final stage of AI; what makes startups boom; what kind of personality shud I have!
I think today, I found the connection b/w all of em!
The answer lies in how GPTs are trained and their superpowers or the lack of em!
GPTs are literally trained over the WHOLE internet. They have read ALL the books, ALL the physics laws, seen all the case-studies, used ALL the softwares.
The result is superhuman intelligence, GPTs are literally the ‘Masters of Trades’. See the diagram below, on how GPT4 ranks top percentiles in ALL the major exams be it MIT or medical.
Google Med-PALM 2 also aces it. Showing expert-level medical knowledge
Yet despite these achievements, the goals of AI giants are still not met. Their aim is to make AI systems do the MOST valuable works humans can do. But, aren’t the most valuable tasks already met Haven’t we already aced ALL the papers, gained ALL the knowledge?
Yes, but still the final and most imp stage remains, the holy grain of AI models:- Creation of new knowledge!
What is that?
You see, with school systems it seems that the whole world is known and all we can do is work hard and learn everything. But, just like Bohr didn’t the whole of atom, we still don’t know the whole of truth.
The net total of knowledge is still pending, there are terrains left to be explored, science laws yet to be gained, and above all ideas yet to be had.
And that is what is unique amongst us humans, we can generate new knowledge. We can see coal, have no experience with it and figure out it burns better than wood. We can see an apple falling, and realize the same force pulls in the moon
This doesn’t just end, at science. Most valuable stuff is just new ideas!
The idea of @amasad to make IDEs online & multiplayer
The idea of Brian Chezky for making strangers live together in @Airbnb
Big companies aren’t just a great products, they are fundamentally new ideas. And these new ideas and new knowledge are the one thing AI models can’t do. They are the MOST valuable things humans can do
This thought startles me more when I think deep abt it. I remember reading biographies of great founders, and a pattern strikes me.
I remember reading about Zhang Yimming (TikTok founder) on how choosing to marry was a mathematical optimization problem How he choose his college by just narrowing his priorities and then shortlisting the top ones. He applied that to TikTok later, modelling human behaviour thru maths.
How for @reidhoffman networks always mattered SO MUCH!
How @bchesky just loved hosting folks even before @Airbnb
How @amasad hacked @Replit before it was close to a company.
This showed a pattern to me, this showed how the MOST important stuff for humans i.e. new idea generation happens!
New ideas get made by being true to yourself. Cringy but true, it happens when you are actually you!
New ideas don’t get made when u imitate! You can read all the books in the world, get ‘mentored’ from the top leaders Nothing will work. Truthfully every human has a core inside of him, which when nurtured and brought forth brings a NEW IDEA!
It was networks for @reidhoffman,
Hosting some1 at ur home was for @bchesky
It was running in movies for Tom Cruise. Singing of teenage crushes was for Taylor Swift, while singing for teenage insecurities was for Olivia Rodrigo.
It will be something else for you! And finding that and bringing forth to the world will be THE MOST VALUABLE stuff u can do
And this sense of respect for individuality is what capitalism nurtures and communism bans. Though cringey it is true:- We are all unique, just like everyone else. And finding that uniqueness is the MOST VALUABALE stuff u can do
So don’t imitate the toppers of ur batch; or that successful colleague; or that Series A founder. Genuinely make your path, its what AI can’t do (yet)