AKA The Pendulum of the Anthropocene
Ten years ago I recall myself meta-presenting to my high school class Iain McGilchrist on the TED stage, via RSA Animate. He wrote a book called The Master and His Emissary: The Divided Brain and the Making of the Western World, about the left brain right brain dichotomy and about how the history of humans since the Enlightenment is a story of the shift from right-brained religious order to left-brained scientific discovery.
But I being an art teacher, and they being young people who had no reason to believe that the entire history of humankind is any indication of human futures, I speculated aloud, for their sake -- are we really destined to continue on this trajectory forever? What if the pendulum were to swing in the other direction? Could there be a world where the Right brain of feelings and emotions are more important than facts and data?
I didn't think what I was speculating could actually be true. My job was to ask the craziest questions imaginable, in the hopes of stretching the most malleable material on Earth -- the minds of young people. I myself could never imagine a world where the right brain became the dominant force. But I always held back my own biases, because one thing I was certain about was that the world these kids would grow up in would be very different from the world I grew up in, and they only way to prepare them was to forget everything I knew, and say nothing else but "what if".
And then it happened. Facts became optional. Computers became creative. Only 500 years after its appearance, Reason has begun to lose its appeal, and its utility.
First, we see algorithms generating theories without any data:
The Ramanujan Machine - Researchers have developed a 'conjecture generator' that creates mathematical conjectures
Feb 2021, phys.org
"The Ramanujan Machine" generates conjectures without proving them, by "imitating" intuition using AI and considerable computer automation.via Israel Institute of Technology: Gal Raayoni et al. Generating conjectures on fundamental constants with the Ramanujan Machine, Nature (2021). DOI: 10.1038/s41586-021-03229-4
And then, we see algorithms generating data without any theories:
New machine learning theory raises questions about nature of science
Feb 2021, phys.org
Instead of teaching the program the laws of physics, he just shows it all the orbits of all the planets until it can produce its own orbits. No more rules baby:And he goes on: "I would argue that the ultimate goal of any scientist is prediction. You might not necessarily need a law. For example, if I can perfectly predict a planetary orbit, I don't need to know Newton's laws of gravitation and motion. I go directly from data to data.-Hong Qin, physicist at the U.S. Department of Energy's Princeton Plasma Physics Laboratoryvia Princeton Plasma Physics Laboratory: Hong Qin, Machine learning and serving of discrete field theories, Scientific Reports (2020). DOI: 10.1038/s41598-020-76301-0
Another one, we let the algorithms imagine, just to see what they come up with:
Enabling the 'imagination' of artificial intelligence
Jul 2021, phys.org
They're really good at synthesizing, but not at creating imagery from scratch. Here, "controllable disentangled representation learning" or what the article calls "imagination".via University of Southern California: Yunhao Ge et al, Zero-shot Synthesis with Group-Supervised Learning. Open Review. https://openreview.net/forum?id=8wqCDnBmnrT
No comments:
Post a Comment