AI & the Story of The Limits of Knowledge
Hi there, ✋
I’m Miguel, living in Barcelona, nice to meet you! I truly believe that we are on the edge of a new period for the Humankind, analogous to the discovery of writing, printing press, or computer science — all these evolution turning points, involved a leap forwarded in technological processing & collective intelligence.
Altough the word “edge” may lead us to the idea of single critical point, I lean more towards the concept of “overnight success, that have been building for over than 30 years”. Highlighting the fact, that we (All) are on the shoulders of giants, and there was an insane groundwork being developed behind curtains, which now allows people like me playing around with these far-fetched AI tools. 🤖🧠
I have always found fascinating the idea of learning in the web public square (i.e Internet), where you don’t need to be a tech savvy person to find the right insights, you are looking for in any type of field. There are a million of good examples, but one of the most illustrative ones, it’s the behemoth of visual content: YouTube. 📽️
If you seek to learn anything, anything at all… most likely, you will find a how-to, guiding step-by-step video on YouTube. You want to learn how to code, learn a new manual craft, improve your physical sports skills, help studying for science or history class, anything related to any type of domain… you know where to find it!
Maybe this is just a way for me to say “THANK YOU”, to all content creators out there, that helped to build this humongous knowledge archive. 🙏
For example, through Ben Tossell’s NL Ben’s Bites (here), where he goes over all the current major advancements in the AI world, I was able to inadvertently find enigmatic_e YouTube channel, and learn how anyone can easily set up Stable Diffusion editing “text-to-image” AI model, to create crazy looking, original videos! 🤯
Btw: Maybe you like to read weird stuff too. I have other small, quick to read, and curious posts on my profile that may interest you ➡️ Miguel Parente
The concept of learning anything autonomously due to the widespread, powerful tech resources, that it’s the Internet… it’s actually a good transition to today’s exercise.
A couple of years ago, my girlfriend at the time (now wife), to embrace the Catalonia tradition of Sant Jordi, surprised me with a gift, the book “What We Cannot Know: Explorations at the Edge of Knowledge” by Marcus Du Sautoy.
I’m a contrarian at heart, who loves the idea that we don’t hold the absolute truth, meaning that are a lot of new concepts out there waiting for us to unravel them. Therefore, it was the perfect gift! 😻🌌
This book was the main source of inspiration for today’s AI story.
Quick tangent, if you are not aware of Catalonia’s Festival of Sant Jordi (I wasn’t before moving to Barcelona), it’s worthy to dig a little bit on it.
What is the Festival of Sant Jordi and when is it?
It is one of the most original celebrations in Catalonia and takes place in spring, on 23 April.
It is a very popular festival that combines culture and romanticism, celebrating both World Book Day and Valentine’s Day.
On this day, it is custom for couples to exchange gifts: the men receive a book and the women receive a rose.
It comes from the famous legend of Sant Jordi and the dragon:
The knight defeats the dragon, who was terrorising the village and the princess, and after penetrating the dragon with his sword, Sant Jordi gave the princess one of the beautiful red roses that spilled out of the beast’s body (here). 🐲🌹📕
Quick disclosure: please bare in mind, that I’m no expert on AI topics, and that these exercises are just for fun & game’s purpose.
Furthermore, I will not bother you with all the details & logistics upon the AI Stories exercises — you can find the full extent of it in the first medium post (here).
So… Let’s play
TL;DR: I’m leveraging two different AI generative tools, both from OpenAI, one to generate the cool looking pictures (DALL·E — here), and the other one to immerse ourselves in a new story and/or theme (GPT-3 — here).
📏 Guidelines & Rules — on how to differentiate “me” from “AI”:
Paragraph style ➡ it’s written by yours truly
Normal text ➡ it’s the AI magic
Asking AI for Human Limits of Knowledge
Humans always looked up to the stars, down to microbiomes, and around to Nature.
They have always been inspired by it, tried to imitate it, and even worshiped it.
“Biomimicry” is the term coined by Janine Benyus in her book Biomimicry: Innovation Inspired by Nature.
Biomimicry is an approach to innovation that seeks sustainable solutions to human challenges by emulating nature’s time-tested patterns and strategies. The goal is to create products, processes, and policies — new ways of living — that are well-adapted to life on Earth over the long haul.
In her book, Benyus introduces the core principles of biomimicry as they apply to sustainable design:
- Emulate Nature’s Efficiency: Look to Nature and ask, “How can we do this better, using less energy, materials, and time?”
- Sustainably Manage Materials: In Nature, everything gets used. “Waste equals food.” Nature runs on sunlight.
- Use Renewable Energy: “When we look to the natural world for clues about how to sustainably power ourselves, we find that Nature runs on sunlight.”
- Redesign for Abundance: “Nature produces an abundance of food, using very little energy or material.”
- Emulate Nature’s Patterns: “In Nature, form follows function.”
- Create Closed-Loop Cycles: “In Nature, one creature’s waste becomes another’s food, and wastes are continually recycled into new life.”
- Apply the Precautionary Principle: “When in doubt, leave it out.”
Benyus offers concrete examples of how these principles are being put into practice in the design of everything from buildings to airplanes to computer chips.
In one example, Benyus describes how the Japanese company Sony developed a new line of energy-efficient televisions by studying the eyes of moths. Moths have large eyes that allow them to see in low light, but they also have a layer of cells that reflect light, so that predators cannot see them.
Sony applied this principle to its televisions, using a layer of cells to reflect light. As a result, the televisions use less energy and produce less heat.
In another example, Benyus describes how the German company Daimler developed a new type of car paint.
We have come along way, however Humans are starting to understand that maybe are things that we cannot know, explorations at the edge of knowledge.
Therefore, we must change the way we think, how we perceive the world and how we relate to it.
Our minds are like a computer, the way we were programmed will determine the outcome, the results we get in life. Maybe it is time to re-program our minds, our subconscious mind to get different results, to get what we really want.
We can reprogram our mind to get what we want.
The question is: How?
There are different ways to change our mindset, our thought process, and our way of perceiving the world and life.
We can use affirmations, positive thinking and visualization.
We can use meditation, yoga and various relaxation techniques.
We can use hypnosis, NLP, EFT, Time Line Therapy and other modalities.
We can use various tools and techniques that will help us to change our mindset, our way of thinking, our way of perceiving the world and life.
The choice is ours…
…THE END 📖
Happily ever after?
Mind-blowing, right???? I would not know how to continue to express myself on a such complex and abstract topic. It’s astonishing, how the AI was able to do it in less than 1 min!! 🤯⌛
Every time I try it, it feels like magic! Going back to my initial thought in this post, that’s why I feel that these new AI generative models will become standard tools that empower us to keep building & creating (e.g. like writing, computers, and so on)—enablers of the evolution of collective intelligence.
Please, let me know down in the comments, if you agree with the POV of the AI, or if you find no limits to the Human Knowledge.
Thanks! ❤
Carve-out (acquired.fm style): one of my favorite substacks Packy McCormick the author of the Not Boring NL (here) has started to explore the YouTube world!
Related to what we have been talking, of learning in the Internat public space, Packy McCormick has been releasing podcasts co-hosted with Anton, who indeed is an expert on the AI realm and help us understand in layman terms AI research papers and breakthroughs.
The latest video about Chinchilla by DeepMind, shows us that the best way to improve the performance of LLMs it’s not by increasing the number of parameters (i.e. wide of the model), but rather by increasing the length of training data.
Anton summarizes it best: “for a given amount of compute, without more data, scaling parameters gives you diminishing returns in performance”. 🤔