As part of an effort to increase and infuse technology, and related practical abilities into the daily lives and activities of future graduates, a certain university in Romania tries to pilot a program. I am a part of this program, designing two of future courses and running the pilot with them. This activity got me in a meditative state about the current state of Technology (AI and s**t)
My long-lasting statement is that the current state of the new wave of technology is just beginning to show its potential, and the future has way more surprises ahead.
Things change fast, and we have reached a state of profound revolution in knowledge. Last time I gave an argument against the infamous plateau of generative models. Or against transformers in general. I weighted against the problem of “hierarchies of scale”, and tried to frame it in a story.
The current hypothesis
My current argument is that it seems that we have managed to tame certain bits and pieces of the way our own mind figure out data. I do not believe that we know exactly what part of our mind we managed to tame, but we tamed it pretty well. It’s like when we domesticated the dog. We really didn’t know what a dog is, but we used it pretty clever.
Dive into my mind
Let’s take the example of AlfaFold.
The long story short is: There is an active research field in biology and chemistry trying to figure out protein structures. This is extremely important for various fields, like medicine, just to give a grasping example.
Researches have found several hundred structures via a classical exhaustive metod (x-ray cristalography).
Then they tried to move it into the power of computing, with various sub-mediocre results.
Then someone (Prof. David Baker) tought to make a game (called Foldit) where gamers will actually help predict a structure *(following some rules, dooh). Spectacular results.
Then someone tought to do train some DNNs in a reinforcement setup to play the game. Very good results.
And then, (2018-2020) the AlfaFold team threw Transformers at it. In a very clever setup.
Finally Awarding them the nobel prize in chemistry, last year. For discovering the structure of 200 million proteins. And more. I leave the details to be researched independently.
This helped the field of medicine spitting out a vaccine for mallaria, and few drugs targeting antibiotic-resistant bacteria, etc.
That is why
I think what is happening is virtually unimaginable.
Some of my somewhat intimate friends know that one of my predictions is that the medicine will benefit from a new branch, or a new approach, where the clinical model will be challenged. I am not arguing against the clinical model and the clinical epistemology, but bear in mind, that medicine will change dramatically over the course of next years. And it is not because of AlfaFold only. Things started to converge once technology started to offload knowledge, things started to converge once usecases like “drug-repurposing” cristalized their way into mainstream medicine.
And that is why I think, once the dust settles on the current crisis in the IT world, the future looks challenging, and way bigger than it was before.
And finally, that is why I think we have reproduced a valuable piece of the Human Mind at scale.
Cheers!