COVID-19 Hangover — Part I
or The Better Day After With AI/ Climate Crisis
Photo by Jonas Weckschmied on Unsplash
This is the third worldwide crisis I experience as an adult. During the first one in 2000, the Dot-com bubble burst, I was serving at the IDF Intelligence Computing unit and deciding on what academic route to choose. Economics looked like a better ROI alternative, but I went with my heart and took Physics and Computer Science B.Sc. 2008 Sub-Prime caught me in a multinational software corporation and showed me how financial and high-tech worlds interconnected. And I added an MBA to my resume, which opened many professional opportunities down the road. 2020 COVID-19 surprised me with its intensity: Ebola, SARS, and other pandemics seemed to have more local short-term impact. It put everything on hold and make me stop and reflect on the crazy race of the past years and where the future can take us. Given my background and the span of interests, I look for the ways how AI and ML could give us the opportunity for a significant leap forward.
This part of my “COVID-19 Hangover /The Better Day After With AI” series is dedicated to the implication of the recent developments on the Climate Crisis. In Part II I’m covering the inventions and trends in personalized and telemedicine. In Part III I'm sharing my thoughts on Remote.
Climate Crisis and its denial was the topic of 2019. Greta Thunberg made it to the cover of TIME as a person of the year after sounding concerns of the young generations that we are at the point of no return of destroying our planet and if we do not take immediate action, there will be no way back. The opponents of the crisis argument stand by the cyclic nature of Earth, bouncing from the ice age to the warmer periods depending on the Sun activity and other factors. The climate academic and industrial activity is mostly focused on simulations and weather predictions, based on the seasonality and geo factors, like tectonic movements causing over 10,000 earthquakes a year.
How AI and ML could contribute? If we had enough “labeled data” (e.g. if an event happened or not after the previous event) we could run forecasts rather than simulations. The problem is that we do not have an “Earth B” planet where we could have different events happened than the ones on our home planet. COVID-19 travel and manufacturing shut down poses a one in the lifetime chance to see the effect of the lower consumption and thus lower pollution. In a few months, we will see the effect on the wildfires and hurricanes which occurred with growing frequency in the past years.
In an attempt to fight the pandemic, more and more data sources become publically available. ML can find correlations between thousands of factors that we didn’t encounter in the simulations, like oil and mineral mining, stock prices or consumption of specific goods and agriculture produce (is it correct that avocado toasts trending on Instagram causing a drought in Peru?).
Finally, the hardware race between GPUs, FPGAs, custom ASICs will result in better and cheaper infrastructure to run the simulations and projections with more variables in shorter times eventually making them more accurate and timely. When Quantum Computing kicks in, we will plan how to switch to prevention rather than reaction due to the increased capacity, parallelism and time of computations.
