The Next Renaissance is Upon Us
How neural nets will free up scientists to do what they do best: get inspired, experiment, and invent
I worry about Peter Thiel’s observation of a slowdown in technology progress over the past 50 years. During a recent trip to Florence, I was surrounded by expressions of Renaissance-era artists whose inspiration catalyzed experiments which yielded results that were quickly accelerated to the mainstream.
Galileo’s observation of pendulum motion was swiftly followed by timepieces becoming commonplace. His observations of basic motion were mathematically modeled by Newton, whose famous laws were thought to describe all physics until the introduction of quantum physics a century ago. Shortly after Newton, Bernoulli, Carnot, Gibbs, Thomson, and others developed the laws of thermodynamics, yielding heat engines that quickly gave birth to the industrial age. The Wright Brothers soon coupled these engines to their unique “stable” aircraft design to catalyze general aviation less than a decade later. Shortly after World War II, integrated circuits took compute, sensing and communications systems from warehouses to rooms to pockets over the course of 20 years. Finally, advances in chemistry, materials, and processes have made compute and communications affordable. With the Internet, billions of people are now connected, with access to the vast corpus of human knowledge.
Where do we go from here? What series of events will catalyze the next Renaissance to shepherd the next major advances in technology and culture?
The practice of science has become ever more narrow, cumbersome, and unapproachable. The complexity of science has made its practice very narrow in scope. An “expert” undergoes many years of training, only to be on the cutting-edge in a very specific field. Experiments are very complex, take many years to accomplish and are difficult to replicate. The arduous nature of doing science makes its practice less attractive for talented youth, prompting them to go to trading desks rather than lab benches.
With information becoming largely ubiquitous, engineers are redirecting their attention toward intelligence machines. There has been plenty of discussion around how “discovery” is different from “search,” with the latter being about seeking and the former more akin to invention — which we do as humans. Conventional “state machine” representations of compute are not conducive to machines that are expected to “think.” Recent work on deep neural nets have yielded algorithms that can train themselves to identify characters, objects and people in images, process speech and do natural language processing. These algorithms establish patterns in data as they train on many thousands of samples, and like the human brain, recognize those patterns and make distinctions.
The art, science, literature and culture that blossomed during the Renaissance was a self-reinforcing phenomenon that catalyzed modern technology and culture. Ruling aristocrats further heightened their stature by embracing and funding the very best talent — similar to how billionaires fund startups and research today. This collaboration generated fabulous works of art and advanced the cutting edge of science which served as the basis of new industries — further empowering their nations and people. The advent of neural nets and intelligent machines will liberate aspiring inventors from spending time learning specifics, building code to test hypothesis and taking arduous measurements, and instead spending their efforts on advancing science and advancing humanity. By leveraging “active” vs. “passive” tools, I expect amazing new discoveries to be made, turning science fiction into science fact.
Shahin is a partner at Lux Capital. Based in Silicon Valley, he invests in space, robotic, AI, transportation, VR and brain-tech companies; follow him on Twitter@farshchi