Another summer comes to an end. It's encouraging to see air travel steadily climbing over 2020 levels and air navigation service provider projections slowly rebounding. While the end is finally in sight, the pandemic is still ebbing then peaking on repeat. We enter back into the world and our professional and social lives, comforted by successes and increased vaccination rates, only to fall back into an uneasy vigilance when new variants emerge. As I'm writing this in early August, the Delta variant is the newest and most dangerous threat to public health.In reading this issue and reflecting on the last 18 months, I had a thought. We can learn a lot about machine learning (ML) from studying virology (spoiler alert, I'm not a virol-ogist, so bear with me!). Back in July 2020, four months before the U.S. Food and Drug Administration granted the Pfizer/BioNTech vaccine emergency authorization, I read an article in Smithsonian Magazine entitled "How Viruses Evolve" that resonated with me. The author, Bob Holmes, noted that the virus's first major evolution took place when the virus jumped species. The second step, he went on to hypothesize, would be a bifurcation between two options for COVID-19. According to Holmes, "The outcome depends on the complex and sometimes subtle interplay of ecological and evolutionary forces that shape how viruses and their hosts respond to one another."
展开▼
