For someone with a hammer, any problem looks like a nail – and as expected, the tech industry is hard at work hammering every nail he can find. But the analytical prowess of the modern data ecosystem is especially limited when attempting to address the problem of potential coronavirus treatments.
It is only to be expected – and of course praised – that companies with huge computing resources would try to somehow devote those resources to the global efforts to fight the virus.
In some ways, these efforts are extremely valuable. For example, one can apply Semantic Scholar's context-aware text analysis to the thousands of articles on known coronaviruses to make them searchable by researchers around the world. And digital collaboration tools available worldwide to research centers and health authorities are leagues that go beyond where they were during the latest health crisis of (or rather, approaching) this magnitude.
But other efforts can give a false sense of progress. One area in particular where AI and technology have made great strides is drug discovery. Countless companies have been founded and have attracted hundreds of millions of funding with a promise to use AI to accelerate the process of identifying new substances that could have an effect on a particular condition.
Coronavirus is a natural target for such work, and already some companies and research organizations are calling early numbers: 10 or 100 of such substances have been identified that may be effective against coronavirus. These are the types of announcements that are making headlines – “An AI has found 10 possible coronavirus drugs” and stuff.
It is not that these applications of AI are bad, but rather that they belong to a set with little useful results. If your big data traffic analysis supports a proposed policy to limit or undermine transport options that way, that's one thing. If your analysis yields dozens of possible actions, all of which can be a dead end or even be detrimental to current efforts, that's another matter entirely.
Because these companies are technology companies, and necessarily break up with their solutions once introduced. Each line of treatment requires a grueling battery of field tests, even to be ruled out as a possibility , let alone be effective. Even medicines that have already been approved for other purposes must be retested for this new application before they can be used responsibly on a large scale.
In addition, it is not guaranteed that the new substances that often result from this type of drug discovery process have a realistic production path, even on the scale of thousands of doses, not to mention billions. That's a completely different problem! (While it has to be said, other AI companies are working on it .)
As a lead generation mechanism, these approaches are invaluable, but the problem isn't that we don't have leads – it's the whole world who can now manage to follow up on the leads it started with. Again, this is not to say that no one should do drug candidate identification, but they should be considered for what they are: a list of tasks, with uncertain results, that other people should do.
Likewise, an “AI” technique where, for example, chest X-rays can be automatically analyzed by an algorithm is something that could be valuable in the future and should be pursued – but it is important to meet expectations in keeping in line with reality. Telelaboratories may be set up for this in a year or two. But nobody gets a coronavirus diagnosis from an AI doctor this spring.
Other places where algorithmic predictions and efficiency would be welcome in other days will reject them in an emergency where everything has to be checked intentionally and three times, not smart and new. The most attractive and popular approaches to fast-moving startups are rarely the right ones for a global crisis involving millions of lives and thousands of interlocking parts.
We're happy when a car manufacturer reuses its factories to produce masks or fans, but we don't expect it to discover new drugs. Likewise, we shouldn't expect those who are discovering drugs to be more than that – but AI has a reputation for being something like magic, because the results are somehow fundamentally superhuman. As has been noted before, better processes sometimes give you the wrong answer faster.
In general, work on the digital frontier of the biotech industry is indispensable, but despite the impending health crisis, it is unsuitable to help mitigate the crisis. But it should not be expected, neither among the lay public who read only the headlines, nor among the technotopians who find more promise than justified in such progress.
Tags: #Robotics, corona, COVID-19, Covid19, Kunstmatige intelligentie, wuhan