SAN FRANCISCO, CALIFORNIA - Search for artificial intelligence (AI), machine learning (ML) or deep learning (DL), and COVID-19 in Google Scholars today, and you'll get over 1.5 million responses. Exactly how many of these articles have had any impact on frontline clinical medicine, disaster planning, or the pandemic response in San Francisco, in the United States, arguably one of the wealthiest, most highly resourced, and tech-savvy cities in the world?  None.

NLP word cloud of SFEPA national survey of 160 ED physicians in 16 cities on March 2, 2020, “What medicines or supplies do you anticipate running out?” (Picture supplied by author)

Four Reasons Why AI Failed Pandemic Playbook Planners

1. Even before the pandemic onset, most urban hospitals in the US struggled daily just to keep their emergency and critical care departments from being overwhelmed with daily patient loads. 

Superb emergency departments, which would later become critical epicenters of acute care delivery, such as Mt Sinai Hospital in New York, were labeled ‘war zones’ in the local press - a full three months before their first COVID-19 patient even arrived. Dump a pandemic on an already overstretched acute care grid, and it's easy to see that there wasn't the leadership bandwidth for AI innovation when the daily focus was to just keep on the lights.¹

2. A lack of understanding of AI capabilities from the boardroom to the executive suite to the frontline, combined with a real dearth of available AI talent across all industries, and it's clear why there is such a huge gap between ‘wanting AI’ and actually ‘doing AI.’