Uncertain Ruins – From The Big Bang to AI

Posted on Mar 1, 2020 in AI, Art and Science, Machine Learning, News, Outreach, universe

I was invited to contribute a piece responding from the perspective of astrostatistics to the timely and exciting show “Uncertain Ruins”, a “a site-responsive collaboration by artist Julie F Hill and Gauld Architecture that draws on the social, material and historical context of the Swiss Cottage Library in which the gallery is located”, part of the Passen-gers site-specific exhibition series.

Closing event of the “Uncertain Ruins” show and launch of publication, Jan 2020

From the Big Bang to AI

For thousands of years, humankind’s only instrument with which to observe the cosmos has been the naked eye. Our ancestors paid a great deal of attention to the movements of the heavenly bodies, for they believed that the sky had a powerful and direct influence on human affairs. 

The geocentric model of Ptolemy held sway for thousands of years, until the Copernican revolution in the 17thcentury put the Sun at the centre of the solar system, thus dethroning the Earth from its privileged position. With his home-built telescope, Galileo saw the craters on the Moon, discovered the four largest moons of Jupiter, the rings of Saturn, sunspots and the phases of Venus. The glass ceiling of our ignorance had been shattered as other worlds were revealed to us.  From then onwards, the desire to peer further out into space brought huge strides forward in science. 

Today, the old-fashioned image of the astronomer as a man (as the stereotype goes!) looking through his telescope at the stars has been replaced by billion-dollar space telescopes, sending back huge amount of data that require supercomputers to be processed and understood (and thankfully the gender balance in astronomy is slowly but surely improving). Digital technology from the early 1990s onwards produced an exponential increase in astronomical data. As a consequence, our data processing capabilities are – already today – the main hurdle to advancing our knowledge of the universe. Within our lifetime, the entirety of the visible universe will have been mapped out: we will have seen everything there is to see. The question will then be: what does it all mean?  

Despite huge advances in the last 20 years, astrophysics and cosmology still have profound questions to answer. Perhaps the most important conundrum is the fundamental nature of dark matter and dark energy, which together account for 95% of the contents of the cosmos, but which remain largely unknown. Resolving this mystery has the potential of changing our understanding of the nature of reality. Discovering life elsewhere in the cosmos would be of historical importance. 

None of the above can be accomplished without statistical and data analysis methods that have yet to be invented. Machine Learning and AI are rapidly becoming indispensable tools for cosmologists and astrophysicists. No human eye will ever inspect all the 50 billion galaxies in the visible universe, nor the 7,500 billion potentially habitable planetary systems: we need machines to do it for us. 

This is why PASSENGERS is so timely: with its interrogation of the increasingly important role of AI in our exploration of the cosmos, the show invites us to examine the role of information and data manipulation in the production of scientific knowledge. In which sense, exactly, will we “understand” the universe if conclusions on its physical nature will depend on the output of an inscrutable and unintelligible algorithm, such as a deep neural network? How can we trust such methods to conform to our physical intuition? Can we teach machines to appreciate “beauty” in scientific theories? Will a super-intelligent AI surpass the combined genius of Einstein, Hawking and Gödel, and if so will it bother to explain its insight on the nature of the universe to us? 

As machine learning and AI seep through all of our lives, increasingly nudging them in undiscernible directions without us even noticing, this artistic cross-examination of this technology reveals its pervasive – and often invasive – potential. 

Leave a Reply