What was the mindset required a century ago to drive scientific and technological progress?

In the 20th Century, it required physicists like Albert Einstein, Max Planck and Niels Bohr to establish new theories and formulae. They tried hard to find a “World Formula” that would allow calculating all the world’s scientific problems through a single set of equations representing a unified theory. Even though some of Einstein’s equations may look simple, e.g. the General Relativity requires deep understanding of advanced math.

One century later, a lot of today’s technological progress is fueled by Artificial Intelligence, with deep neural networks being fundamental building blocks. Core algorithms are based on Calculus, Linear Algebra and Numerics. Although their equations may look complex, you can already get a good grasp of them with high school or college level math, among them:

Calculus: Minimization using derivatives and chain rule

Linear Algebra: Systems of equations and matrices (vectorization for efficient parallel processing)

Numerics: Sequences and convergence for effective numerical optimization

In neural networks large combinations of relatively simple (linear and non-linear) functions feed on huge amounts of data, optimize myriads of parameters and are powered by the incredible data storage and processing capacities of today’s computers. Another layer of hyper-parameters governing the optimization (“learning”) of these parameters make them even more adaptable and customizable to almost any calculation task.

What is the mindset required in today’s AI Age to drive technological progress?

Rather than innovation through a “formulae” mindset, more than ever, technological progress can be driven by an experimental mindset: Learning through rapid cycles of experiments and feedback, e.g. optimizing learning parameters and hyper-parameters of an AI model. And experimenting on how AI applications could automate routine work. And experimenting on how whole organizations—and even whole industries—could be reshaped by Artificial Intelligence. So, in this day and age, an experimental mindset may trump a brilliant mind reluctant to make experiments and mistakes.

Back in my study days at the University of Göttingen (boasting world-famous mathematicians and physicists like “Carl Friedrich Gauss” and “David Hilbert”), pure math and theoretical physics enjoyed the highest reputation, whereas applied math were regarded as secondary by many academics. Yet, it is applied (numerical) math and computer science with their experimental approaches that are transforming today’s age!

What mindset is driving your progress? – I’d love to discuss and explore this with you. 😊

© Burkhard Wilmers

## コメント