• Post on Twitter
  • Share on Facebook
  • Post on LinkedIn
  • Post on Reddit
  • Copy link to clipboard
    Link copied to clipboard
Know

For Better or Worse, Computer Algorithms Are on the Front Lines of the COVID-19 Pandemic

2 min read
Apr 24, 2020

Would you let a computer algorithm make a life or death decision about your health? Many health professionals were skeptical about the widespread use of artificial intelligence (AI) in healthcare before COVID-19. Many still are. But the pressures the pandemic is putting on health systems have led to more experimentation. This goes for adjacent fields as well. From coronavirus education campaigns to treatment decisions, AI algorithms are on the front line of this pandemic.

IBM’s Watson, a computer system originally designed to win the Jeopardy! game show, is available to answer your coronavirus questions. The system was adapted to be a chatbot capable of chatting on topics from the virus itself to school closures caused by the outbreak.

Even if you think misinformation about COVID-19 is a grave danger, this might seem like a trivial use of AI. But information about how the virus might spread has significantly impacted our reactions, as Axios pointed out recently. The predictions an algorithm makes could cause us to extend or shorten a lockdown. Some argue this is much better than the gut feel of a politician or even the informed opinion of a public health expert. Others warn that such disease outbreak simulations often miss the mark. Either way, it’s clear we need to make sure the results of such simulations are sensible and checked by multiple experts.

A similar story is playing out in hospitals. The MIT Technology Review recently reported on the NHS (the United Kingdom National Health System) using AI to screen for COVID-induced pneumonia. In this case, NHS doctors used an already certified X-ray screening algorithm that had been updated to help detect COVID-19 cases. An early study found that the method worked with 95% accuracy. The potential is clearly there for such diagnostic tools, but there are also many risks. The same team reported being offered many other AI products that didn’t work at all. There is certain to be a learning curve as health officials learn how to evaluate and use the new methods.

The experiences we have now with AI will shape both public perceptions and business strategies for the near future. For businesses using or developing AI products, the results of these experiments are worth keeping an eye on.