We are going through unprecedented times. Coronavirus has each one of us in the grip. At best we are quarantined at home, at worst we are fighting for our life, or for saving someone else’s life. We are dependent on technology more than ever: to be able to work so as to sustain the global economy, to be able to stay connected, while being fully isolated, to be able to find a cure to the virus, which has already claimed many lives.
All this will come to an end, but the world will never be the same afterwards.
In times like these, expectations from Artificial Intelligence (AI) are very high. More than ever, we want AI to be a true friend to us and help us through our difficulty. And it does.
AI is helping in biotechnology to accelerate finding a cure to the virus, by scanning through all potential assays efficiently. It is helping to understand the symptoms of the disease so that a warning can be sent out in time. It is helping to understand where the infected or potentially infected individuals are so that careful measures can be taken in those areas. Just to name the few.
All this is good. Is it?
China has been using drones to spray the public with disinfectants. Through thermal sensing, these drones have also been helping Chinese officials to detect people with higher body temperatures, which could indicate they have the virus. Israel has started using its people tracking system which is otherwise used by their military service to hunt down potential terrorists. This technology enables the tracking of the location and movement of people via signals sent from and to cellular phones and their interconnected antennas. It uses GPS, street cameras, as well as special thermal devices. It records phone and computer conversations, extracts voice samples and monitors chats and messages in social media (including Google, Facebook, Twitter, YouTube). All this information is then stored and can be used by the security service retroactively up to 14 days.
During extraordinary times like these, when all is about humanity’s survival, any method and technology that helps us win the battle seem legitimate. Our foremost priority is to protect human life. Hence, without thinking much, new AI technologies are born and existing ones are repurposed to help us win.
One question is still valid though: what happens to the world after coronavirus?
Powerful AI technologies, such as the ones discussed above are born out of urgency and necessity. Once born, they will not die. As much as they will create good in good hands, they will create evil in bad hands. In wrong hands, such as a despotic leader’s, same drones can be used to spray poison instead of disinfectants. They can be deployed to detect people based on thermal signals to take them away to unspecified places for unspecified purposes. What if the thermal signals were wrong? The same people tracking system can be applied to start reporting on anyone, anywhere, for any arbitrary reason.
It is hard to think about the “after”, while we are still in the survival mode, but we have to. We need to put thoughts into the AI we are creating now and we must ensure that it lands in the right hands. The harm that AI creates is as much our responsibility as the good it creates. We cannot “morally outsource” AI, saying “it was not us, but the machine”. Nor can we simply say “it was necessary then”. We as humans must claim the responsibility for the future consequences of AI that is being born now.
This is where Responsible AI should help us. Today we may not have all the answers, but we need to be able to ask ourselves the right questions. Responsible AI should guide us through identifying the right questions to ask so that we are not only reacting to the urgent threat we have, but we are also making conscious decisions, as much as possible, in designing our AI-based response as well as relevant guardrails to it. The mission of Responsible AI is to help us humans create AI systems such that they preserve human integrity and that they “do the right thing” at all times. “At all times” entails times like ours now. Responsible AI should not only be providing guidance towards building ethical AI systems under “normal circumstances”, but it should also be capable of showing us the right path under extraordinary circumstances.
Reflections 