A ‘responsible AI’ can help Bangladesh fight Covid-19 effectively

The fight against Covid-19 pandemic needs the coordination of all possible frontiers including the efficient use of Information and Communication Technologies. The recent success in South Korea with Big Data analytics through Artificial Intelligence has helped the Korean government to 'Contact Tracing' by developing a digital map of the suspected Coronavirus bearer or infected persons.

Massive data collected by the Internet of Things and real-time processing of those data in compliance with transparency, accountability and human rights principles were key behind the success. That is why governments around the world are now opting Artificial Intelligence-based citizen monitoring to better track the outbreak.

The government of Bangladesh also has taken a similar initiative to track coronavirus cases and find out areas susceptible to contamination by using mobile users' information (the Daily Star, 29 March). It is a much-needed initiative by Access to Information (A2i) project of the government of Bangladesh in collaboration with the National Telecommunication Monitoring Center, Ministry of Health, Ministry of Home Affairs and mobile operators (Somoy News TV, 31 March).

Although the use of AI to contain Coronavirus is a very new thing, it will give more efficiency to government efforts to trace back coronavirus infected persons and people exposed to them. AI can also be used to analyze and understand the long-term impact of this crisis on society and economy, the priority list for the policymakers, to develop coronavirus-related research and helping with diagnostics etc.

But we have to remember that AI is a mere technology, which can be useful to contain a pandemic subject to proper knowledge and creativity of the humans who use it. Identifying new and innovative ways is the key to leverage what AI can do to deal with Covid-19 pandemic. Machine learning through AI works by identifying patterns of already existing data, which can detect the pattern that is overlooked by a human if used wisely.

So, AI systems need a lot of data (Big Data) and machine learning implicitly assumes that what has worked in past (past data) will still work to analyze the future situation. How does it relate to fighting Covid-19 crisis? This global pandemic is different from past global outbreaks and all the learnings of the past may not work today.

Our traditional assumptions about cause and effect of a global pandemic, based on past data, may no longer work effectively to contain Coronavirus. However, humans have one advantage that they can learn lessons from one situation and apply the abstract knowledge to make best guesses on future events. AI systems, in contrast, have to learn from scratch even if there is a slight change in the given data.

Therefore, data quality is the key to the proper function of AI-based Big Data analytics. Confidentiality, integrity and availability are the three pillars of data quality. It is also important to maintain some checks and balance, e.g. robustness, transparency and respect for human rights and principles for proper analysis of those data gathered.

Special focus on maintaining the privacy of personal data through anonymisation and encryption reduces the scope of abuse. As Bangladesh does not have any dedicated Data Protection Act, despite some data protection related rules scattered in local ICT related laws and policies, it is not clear under which specific law(s) government will handle these big data analytics efficiently and transparently in order to deal with Covid-19 crisis.

Is government complying with the National Cybersecurity Strategy 2014 in order to ensure the cybersecurity of this whole data collection process? Is government only limiting itself to collect data through SMS sent to mobile phone users, or is government also taking the help of another medium such as Open-source intelligence (OSINT) collected from social media? Does the government have any plan to crossmatch this Big Data with the data stored by its different agencies, e.g. national household data collected by the Bangladesh Bureau of Statistics or NID data? Is the government collecting the data of around one million Rohingya Refugees residing in Bangladesh also as they are more vulnerable to this crisis?

As more and more data collected from various sources increases the chance of accuracy in AI projection, the government might need to collect, coordinate and cross-match personal data from as much as sources it can keep those data anonymous. However, to what extent personal data is being gathered and anonymized, it should be notified to the concerned people as per the Action Plan 2.3 of the National ICT Policy 2018. It is also a legitimate expectation that the government will limit the application of this Big Data only to the management of Covid-19 situation, not for any other purpose in future.

Thus, fighting Covid-19 needs an inclusive approach of all the concerned sectors such as health, transport, local government and social welfare etc. Are specialized government agencies such as the Department of Disaster Management, Bangladesh Bureau of Statistics, Ministry of Social Welfare etc. are working together with A2i in this process?

Lastly, it is not the government's burden only to fight this pandemic. Apart from mobile phone operators, civil society organizations and academia can also contribute their skills to government-led technological initiatives for better management of this crisis. So, it is wise for the government to include other non-state stakeholders in its technological efforts to map out the spread of coronavirus across the country, if they are not included already. This is how, through an inclusive approach, Bangladesh can ensure a 'Responsible AI' based Big Data Analytics to fight Covid-19 crisis efficiently.

Md Saimum Reza Talukder, is a senior lecturer of Cyber Law at Brac University