100 days of coronavirus: Has artificial intelligence helped?

Crises historically have an accelerating effect on technology trends.

During the 1918 Spanish Flu, the New York Telephone Company advertised a home phone to bring “cheer and encouragement to those in quarantine”. The decade following the pandemic saw a phone line installed in thousands of homes across continents, altering daily life around the world.

Technological advancement is by no means a silver lining. Covid-19, the disease caused by the novel coronavirus, has sickened 1.3 million people and killed more than 70,000 across the globe, according to Johns Hopkins University.

But 100 days since the first case was reported to the World Health Organisation from Wuhan, China, one technology has emerged to underpin much of the response to Covid-19: artificial intelligence.

Containing the disease’s spread, rolling out basic chatbots to help screen potential cases and surfacing treatments by centralising research data - these are the proving grounds for AI amid the pandemic.

"AI is not at a stage of development when it is a panacea for all ills," Kay Firth-Butterfield, the head of artificial intelligence and machine learning at World Economic Forum, told The National. But there are several use cases where it is showing promise, she said.

Geolocation data on people’s smartphones has been used to track and trace those who have come into contact with a Covid-19 infected person, with programmes rolled out in China, Singapore, South Korea and Italy, among other places.

But nowhere is the undertaking more widespread than in India, where the People’s Curfew has the population of 1.3 billion on lockdown as officials there are concerned the peak of the outbreak is still weeks away.

The Indian government’s Bridge to Health app, introduced last week, uses a phone's Bluetooth and GPS systems to alert any app user in the country who has come in the vicinity of a Covid-19 infected person.

The alerts are generated by scanning through government-owned, location-specific patient databases. The alerts are also accompanied by instructions from the Ministry of Health on how to self-isolate, and the course of action in case one develops symptoms of coronavirus.

Besides the main tracker feature, the app also lets users take a quick test to check if they have matching symptoms to Covid-19. However, movement data needs to be tracked regularly to ensure the service works as intended.

For Mark Minevich, president of Going Global Ventures and AI expert, the “most impressive, ethical use of AI” is from SparkBeyond, a New York-based AI research firm. The company created dynamic maps for Italy and Argentina, predicting places where an asymptomatic Covid-19 carrier is likely to pass, at building-level granularity.

The company has helped officials in Italy understand what locations are associated with higher infection rates, like public parks or tourist hotspots.

"Countries across the globe are beginning to use SparkBeyond's predictive AI to facilitate three core activities," Mr Minevich told The National.

These are “identifying regions that have a higher risk of infection based on known cases of infection; how to prioritise the deployment of sanitisation resources, pop-up testing and police presence; and when and how to allow citizens out of lockdown and back to work, safely and responsibly”.

While AI has helped track the movement of people and make inferences to help contain the pandemic, it has also helped people in their homes when they’ve been worried about symptoms.

“Companies gambled by deploying immature chatbots,” Ms Firth-Butterfield said. “The result was an unprecedented adoption rate of chatbots that proved to be accurate, responsive and cheap.”

Chatbots, enabled by natural language understanding (a category of AI), have been deployed around the world to provide information about Covid-19, according to the WEF. Its AI team has concluded that they have demonstrated their usefulness in getting information out to vast populations and that people around the world are getting comfortable with using chatbots for health care. “The coronavirus epidemic is a microcosm of general healthcare problems, offering a chance for the chatbot technology to prove itself,” they found.

“Even if ethical use is side-stepped in the interest of quick deployment to meet the needs of the coronavirus epidemic, we must create ways to find our way back to ethical use, once the epidemic is over,” Ms Firth-Butterfield said.

AI has also enabled collaboration at an unprecedented scale.

“Covid-19 is first and foremost a humanitarian crisis,” Mr Minevich said. Two things are happening with “intelligent data” to help curb the pandemic: decision-making has become data-driven and “there is too much data circulating”.

To help connect the dots, US tech company IBM partnered with the White House to offer supercomputing power and help researchers working to fight the spread of the coronavirus pandemic. The supercomputer, Summit, is expected to assist researchers around the world to better understand the virus and build predictive models to analyse its progress as a disease. The machine can also help explore potential treatments or formulate a vaccine.

In a similar move, Group 42, an Abu Dhabi-based AI and cloud computing company, is offering its supercomputer, Artemis, free of charge to scientific researchers in any field that contributes solutions to the challenge of the current virus outbreak.

But this is not the time for AI to go mainstream, according to Mark Esposito, an economist and faculty member at Harvard University, whose work focuses on AI and the Fourth Industrial Revolution.

"The examples we have seen are still episodic and mainly deployed under a state of emergency," he told The National.

We will only know if these technologies are ready for widespread adoption - like the telephone in 1918 - once the crisis is behind us, he said, and normative measures are put in place to help prevent the next pandemic.

“Only then may we know.”