Imagine an infinite library of new fields of study

Every problem scope has different key features derived from within its own system. For buildings, tensile strength. For pipes, fluid flow. For circuits, voltage and current. For interplanetary travel, lightspeed travel and cryonics. Without the fields that house these features, they would be nameless—we do not care for features that do not belong to a system of interest. 

Systems of interest give names to nameless features—they turn sound waves into music, binary into data formats, or prime numbers into cryptography. Intent births fields, fields define systems, and systems house features. Interest, like the big bang, expands our universe.

Imagine if we charge an AI to create new fields of interest. AI seems well suited to this—imagine giving an AI the task of building a building in an open world physics engine. Would it not derive the importance of features such as tensile strength? 

AI seems uniquely suited for declaring an interest and optimizing for that interest, and in the process, creating a completely new field of study. 

A couple of ideas based on the above thoughts:

1) There should be a way to clearly and simply define interest for an AI. Dall-E is already playing around with this in infancy through the english language, but I wonder if the interface to ask AIs to do things is itself a place that AI can improve. Imagine Dall-E but learning was applied to its input system as well, not just art. 

2) An AI that "discovers" new fields. Take prime numbers and the fact that multplying two primes is easy, but deriving what two primes was multiplied is hard. Without the need for security, this attribute would be useless, but because of the need, this created an entire field of cryptography. I wonder if an AI could help discover new fields based on interest. 

3) A sci-fi story. Imagine a species similar to ours that achieved global sustainability, effectively captured all the available energy of its solar system and used thousands of years of computing time for AI to discover new interests and creating systems around them. It would be an infinite library of wild, crazy new fields of study.

Comments

Post a Comment

Popular posts from this blog

A Policy of Activism in the Workplace is Anti-fellowship

Of course, this is a reflection on the happenings in Basecamp. If you are all Basecamp'd out, please ignore. --- A policy of activism in the workplace is anti-fellowship.  A fellowship is "a group of people that join together for a common purpose or interest". This dictionary definition fails to capture the benefits of fellowship - namely, through an aligned sense of purpose, the group can fully apply the group's collective resources, willpower and intelligence towards one direction and mission. The classic book - "The Fellowship of the Ring" - vividly showcases the benefits, struggles, and nuances of fellowship. What drives the story is the characters putting away their personal preferences and goals to accomplish something greater than themselves. Everyone has to put away something; hatred of other races, pride, fear, ambition, and mistrust of one another, and it is the great sacrifices from all parties that allows the mission to succeed. The book preaches...
Read more

When is the pandemic over?

It's almost June of 2021, a year and a half after covid first turned the world upside down, and yet globally we're far from returning to normal. While some countries like the US seem to be on track, other countries like India and Brazil are still battling the worst spread of the virus to date, and countries like Singapore, Taiwan and Malaysia are re-engaging in lockdowns as covid cases spike.  This all begs the question - when is the pandemic  truly over? Covid cases will never return to zero; we can never completely go back to pre-covid times. However, we've reshaped our lives to combat covid's impact with quarantine rules, work from home, travel restrictions, face masks, lockdown, alcohol wipes, travel bubbles, and more - but a lot of these inconveniences we'd expect to go away. So... when's that happening? The answer for those in the US and Europe is now. For them, the pandemic is already over; borders are about to open and citizens are returning to their pr...
Read more

Upholding values is a dangerous game

Value , n. - principles or standards of behaviour; one's judgement of what is important in life. Values are extremely useful. They're shortcuts for how to live a good life. They range in size from small to great; from "brush your teeth twice a day", to "all men are created equal". Values are also extremely dangerous. They can fall short of solving the real problem but give a false sense of security that the problem is solved. Values are relied on because staying alive and happy is difficult, and a good set of them makes that easier. They uphold everything in human society - culture, social norms, ideologies, economics, law, morality, identity, and more.  But should they?  --- Biologists don't usually ask if traits are universal - they instead ask which traits are evolutionarily beneficial. And some positive traits in one species can be detrimental to the next. Take bravery, for instance; needed for lions or cheetahs, might get you killed as a rabbit. Ind...
Read more