Current AI systems deployed to combat coronavirus AI are still in development and “hype outstrips the reality”, but this pandemic can change the AI landscape, shared MIT Technology review:
“So here’s a reality check: AI will not save us from the coronavirus—certainly not this time. But there’s every chance it will play a bigger role in future epidemics—if we make some big changes”
Too much confidence in unproved AI algorithms can be harmful for the whole AI industry, with investors’ expectations crushing into the reality, not mentioning the opportunity costs, highlighted researchers.
At first, AI algorithms have proven to be very useful, helping government officials to take measures. BlueDot’s AI made its first alert on coronavirus on December 31, 6 days ahead of the US Centers for Disease Control and Prevention. HealthMap is now updating its coronavirus spread map in real time, using artificial intelligence.
Yet with the epidemic progress, AI becomes “less accurate”, mostly due to lack of data: “news sources and official reports offer inconsistent accounts”, shared MIT Technology review. For more accuracy, society might have to opt to the “controversial strategy” of sharing “more of our personal data with companies and governments”. Moreover, “finding agreement on international standards” for the pandemics will take considerable amount of time.
Data sharing could also improve the diagnostic AI algorithms, together with using few-shot learning and transfer learning models, but so far “we should be sceptical about many of the claims of AI doctors diagnosing Covid-19 today”, highlighted researchers.
In theory, AI could be used to develop treatments. Lot of companies are already sharing preliminary results: for example, DeepMind, Google’s AI unit, has openly released the AI-generated structure predictions of coronavirus, yet it is unlikely that results can be seen immediately:
“This is a game-changer for drug discovery, but it can still take many months before a promising candidate becomes a viable treatment”
Current barriers to successful AI application includes lack of talent, technological resistance and data, which needed to be brought to new level to battle next pandemics faster:
“Making the most of AI will take a lot of data, time, and smart coordination between many different people. All of which are in short supply right now”
So far, scientific conferences that bring together scientists to battle coronavirus, are being cancelled as well, reported Venture Beat. Yet many argue that instead of cancelling the scientific gatherings, they might have opted for “all-digital conferences”.
The first conference to innovate is the ICLR 2020, which opted for hosting “a fully virtual conference”, shared Sasha Rush, ICLR general chair from Cornell University. Such measures could not only help scientific community at hard times, but also bring more inclusion and sustainability for the future of AI research.