I think AI is contributing to a golden age of invention.
First of all, AI itself is this phenomenal invention. It's advancing faster than almost any field that has ever existed. To be fair, it's been an exponential curve. Where we are now is directly the result of 60 years of investment, some of which was like a 40-year winter where people thought the people working on this were crazy and wasting their lives. So, it's not like AI came out of nowhere.
We really need to pay our dues to people like Jeff Hinton, Marvin Minsky, and Claude Shannon, who worked so hard on this topic for so long.
But when we're talking about artificial intelligence, we should talk about where it is today and then where it's going.
AI as fuel to invention
Today, artificial intelligence is almost like a compression technique. It's something that is able to take knowledge, represent it, and maybe have a small amount of rudimentary reasoning skills.
What it's really doing is helping you explore the space of invention more efficiently. And there are other applications too.
When we think about AI, it is actually a broader category than LLMs. There are actually lots of areas, some of which are actually way more developed than LLM, like machine learning.
People are using this to do all kinds of things, especially in hardware fields or drugs, where it's expensive to try things. In these examples, what you want to do is take a space of possible candidates and then narrow them down by desirable qualities.
That requires to not only sift through a lot of information, but to develop good simulations. And AI is really good at this! It's really good at developing candidates like drugs, materials, biology, things where it's the study of the specific and helping us get ideas of what to do.
I think that that's a huge advance that we're seeing already in some of these areas, like AlphaFold, a 200 million protein structure database to accelerate scientific research, that people talk about a lot because it's this super transformative breakthrough.
DeepMind just put out a paper talking about an AI tool developed with deep learning, called Graph Networks for Materials Exploration (GNoME), with which they can develop candidates for 2.2 million new materials that can be used for things like batteries. And we're going to see a lot more of that!
AI for a faster validation phase in innovation
It's really going to be exciting because AI will allow us to do more inventions and also scale the science that we're doing. The input for a lot of this is based on research that people have already done.
So, if you're someone like John Goodenough, who invented the lithium-ion battery that we use today, and inquire if you can take the insights of a ‘good enough’ and scale that up in a computational way? Where now you can generate more candidates that can be tested in a rigorous manner.
We have all this scale up manufacturing to the validation phase so much faster, 'cause that's where a lot of these inventions fail.
There are lots of things that we can actually create in the lab, but then they turn out to have some instability, practicality, expense or problem when produced at scale or put into the real world.
Getting to the validation phase faster is something that will allow us to determine if we can go to the scale up faster. The scale up part it's super hard, but we've been doing scale ups since the industrial revolution started.
With this knowledge, are you ready to speed-run the validation phase of your business ideas like never before?
