I am (still) amazed by how AI seems to be permeating all walks of life from call center agents (Google Duo) to ‘no-staff’ stores (Amazon Go) to popularly known areas like self-driving cars or playing complex games.
But two recently developments were especially intriguing. The part that was intriguing was about how they were at opposite ends of cognitive spectrum – Biochemistry (pure science) and artwork (well, it’s art).
The first one was about how Google’s Deepmind improving something called the ‘The Protein Folding Problem’. In non-technical terms: Proteins can take almost limitless number of three-dimensional shapes. The shape itself determines how certain molecules will attach or bind to the protein. And this binding is one way to discover or build new drugs. The obvious challenge is that since proteins shapes can be virtually limitless, drug discovery by this method involves a lot of expensive trial and error. So, every two years, world’s leading biochemists enter a sort of competition whose aim is improve the forecasting of protein folding. In December 2018, this competition was won by Deepmind’s AI – and it was twice as better as the second ranked human participants. Read the reflective blog post from one of the human competitors here and an excellent article from NYT here.
The second development was about an artist, Mario Klingemann, selling his AI artwork in an auction by Sotheby’s. Now this itself is not new – The Obvious Collective has already sold an AI generated canvas for almost half a million dollars in 2018 (read about it here if you are interested). What is new is that Mario Klingemann is not selling an AI generated canvas or artwork – he is auctioning the AI itself. And this AI will keep generating paintings (or images) in a loop. So, what is being auctioned is a system – two screens connected to a ‘machine’ hidden in a piece of furniture. The images themselves are being generated by a GAN which was ‘fed’ paintings from 17th to 19th century. Since what is being auctioned is the GAN itself, the paintings should get better and better over time!
What do these two developments mean? I don’t think historically we have ever experienced one skill set subsuming so many other skill sets! It’s not just about a car driver replacing a horse driven buggy driver or about word processor replacing secretaries. It’s about data scientists replacing drivers, artists, medical researchers, doctors, call center agents, business cashiers, analysts, voice artists, marketers, filmmakers, web designers, journalists and so on and so forth. I don’t think there will be a complete replacement, but we will definitely need a lot smaller number of people doing these jobs than today.
Then, I think it’s almost inevitable that in next few decades, most countries will start providing some or other form of Universal Basic Income. I was listening to a podcast where Andrew Yang, a democratic party US presidential candidate was interviewed. He imagined a scenario where self-driving trucks become the norm – what happens if 50 of those now out of work truck drivers get together with their guns and march to the state legislature?
Universal Basic Income will not be about a handout – it will be a payout to avoid riots on the streets.