AI to save billions by avoiding risk, CIO News, ET CIO

By Dr. Anshu Sharma

The conventional disaster impact assessment methodology was formerly known as damage assessment. A disaster-damaged building would be accounted for with the replacement cost of the building and assets. A further refinement was the Damage and Loss Assessment (DALA) methodology, in which the added component of loss represented longer-term impacts.

With the building out of service until a replacement occurs, losses that will be incurred beyond the cost of replacing the building are included in this approach. Some losses, however, still remain invisible and unaccounted for. Suppose this building happens to be a school. According to the World Bank and others, each lost year of education can lead to a loss of seven to ten percent of each child’s lifetime productivity.

If it takes three years to replace the building (which is often the case, since a major disaster can destroy thousands of such buildings), some children may have already slipped too far to get back on track, and the loss could be much greater. Replace the school building with a factory, and the results might be similar. A few months of inactivity could mean far greater losses than visible damage to building and machinery. Closed for too long, the company risks no longer having the means to recover.

Growing Risks and the Economic Imperative

Concern for unseen losses has increased, even as India has made phenomenal progress in saving lives from disasters. While the cyclones of past decades used to grab global headlines for vulnerability and loss of life, India’s practice of timely warnings and evacuations in recent times is now cited by the United Nations as global good practice.

Casualties from these disasters have ranged from tens of thousands in a single disaster to double-digit loss of life. Contrary to this curve, economic losses have increased alarmingly. Two issues need to be considered when trying to stop this growing economic risk from disasters: first, we are increasingly building our infrastructure in places that are inherently risky and previously left uninhabited, and second , the majority of economic losses remain invisible. because they are private and uninsured (the proportion of insured catastrophe losses is still well below 10%, and often close to zero in developing and emerging economies, according to leading global reinsurance provider Munich Re).

The result is that our development path enters riskier areas as we progress, and the future risks we face are largely invisible. Climate change makes this even more pronounced.

Complexity of datasets and the role of AI

This understanding of the risks we face also raises the question of a very wide range of factors influencing these risks. Hazards such as floods, storms and earthquakes, some of which are becoming increasingly difficult to predict with climate change, are the first set of variables we encounter.

The next, equally diverse set is that of the vulnerability of our infrastructure, including buildings, neighborhood infrastructure, town and village infrastructure, and diminishing utilities and services to hazards. mainly because of their own weaknesses. Charles Richter, the famous scientist, once said that earthquakes do not kill people, but buildings. Our buildings and infrastructure are often too weak to withstand the earthquakes, storms, floods and fires to which they will be subjected. The final set of factors influencing risk are our own human shortcomings, in which we are unaware of the risks we face and therefore unprepared to act as we should in an emergency.

This is where AI comes in as an unprecedented tool, with its ability to quickly process very large and diverse data sets, to help us understand the completeness as well as the nuances of the risks we are at. confronted. Today we are able to assess very recent satellite or drone imagery, showing the reality on the ground from literally days ago, and identify specific risks that could lead to losses. This could include the depth of water we should expect at any location in an evolving flood situation, or the temperature levels during an anticipated heat wave, or the concentration of air pollution. , etc.

In addition to location information, this data also helps us assess the vulnerability of individual buildings. This lets us know which buildings will survive and which will not survive an impending disaster. Families may therefore be advised to evacuate, along with information on the most easily accessible safe places. Schools and offices may be advised to enhance preparedness levels to ensure occupant safety. Factories may be advised to protect their assets and supply chains from disruption. Government authorities may be advised to strengthen infrastructure to withstand future disasters.

Overall, AI-based risk assessment and disaster preparedness planning can help make decisions with very high accuracy and speed, both to protect lives from impending disasters and to make infrastructure safer for long-term resilience. AI tools are already available and affordable to make this change. It is now up to policy and practice adoption to ensure that the benefits reach people quickly. Faster than disasters inflict casualties.

The author is co-founder of SEEDS.