Nabugu - stock.adobe.com
How to use AI for disaster response
AI systems can be a dynamic method to automate disaster response, but they need to be well trained in interpreting disasters to be effective. Here are some tips to get started.
After many years of anticipation, artificial intelligence is now firmly embedded in the disaster response mainstream.
Sophisticated artificial intelligence (AI) tools can help organizations handle an array of critical tasks. These tasks include identifying, quarantining and removing potentially compromised data after a security breach, or building an emergency notification system.
Adnan Masood, chief architect of AI and machine learning at consulting firm UST, said AI can also help with IT service management automation, data recovery and disaster recovery. AI can help in a variety of disaster response situations -- but it has some limitations.
The big picture
AI is already used by disaster response teams to closely monitor the status of potential and ongoing threats. For example, AI-powered computer vision technology excels at several different types of Earth observation, including object detection and long-term area monitoring, as well as connecting AI models to climatic models.
"Automation with AI can make ... imagery analysis simpler and faster," said Appu Shaji, CEO and chief scientist at Mobius Labs, a visual intelligence algorithms development firm. He also said that collected data analysis is highly challenging and requires a massive amount of manual processing. "By detecting and analyzing how objects are moving, one can identify riverfront expansions or wildfire movements that can then be applied to tackle disasters and issue warnings."
On the downside, AI's analytical capabilities are currently restricted to after-event studies. "Although this is still extremely insightful, the goal is to predict potential disasters and calamities, which the technology cannot do at the moment," Shaji said.
Benefits of AI in disaster response
When it comes to disaster response, AI's key benefits are speed, scale and near-time information processing. "This helps organizations make the right decisions given the complexity of a disaster environment," said Dan Simion, vice president of AI and analytics for business and IT advisory firm Capgemini. "As time is critical ... decisions must be made instantly."
By rapidly absorbing and digesting large amounts of relevant data, AI can play an important role in identifying potential imminent threats, serving as a highly reliable early warning system in areas as diverse as weather, agriculture, financial markets and even geopolitics. "AI can bring together data sets and data sources from multiple sources and analyze the data as a single stream to provide predictions," said Chuck Everette, director of cybersecurity advocacy at cybersecurity firm Deep Instinct.
AI excels at suggesting appropriate incident response actions. Organizations may add an AI-driven automation response engine to a business continuity plan. Everette said this better predicts where resources and recovery efforts will be needed most. "It will help make decisions and identify critical paths to getting operations back up and running as fast as possible," he said. "AI is an invaluable source to provide rapid recommendations for the most expedited recovery possible."
Limitations of AI in disaster response
AI's most challenging limitation is that it can only learn from the data it receives. This makes AI totally data dependent. "Think of AI mathematical algorithms as the recipe and the data are the ingredients, so if either is bad or corrupt so is your AI platform," Everette said.
AI also can't think for itself; human research and creativity are still required. "Due to the large amounts of data needed -- as well as extensive hardware, storage and power requirements -- AI training and use continues to be quite expensive," Everette said. More data equals more space and, in turn, requires more computing power. "This all adds up and increases the cost overall," he said.
To ensure maximum performance and reliability, use a well-trained AI system. When building an AI model, the system should be able to recognize what a potential disaster might look like so that it can differentiate disorder from a normal situation. However, that's easier said than done. "The drawback of AI in disaster response is that it's very challenging to properly train the models to accurately interpret a disaster," Simion said.
AI's role in business disaster response will continue growing, Simion said. "The more accurately the technology can predict outcomes, the more AI will expand."
Additionally, as augmented reality and virtual reality technologies continue to improve, these tools can support advanced AI model construction. This enables enterprise DR teams to learn and train in a more visual, realistic way, Simion said.
Meanwhile, as AI technology keeps advancing, analysts generally expect that proactive analysis will become increasingly available and widely used. "This would mean accurately predicting disasters beforehand so that the appropriate cautionary warnings and preventive measures may be taken before the disasters actually strike," Shaji said. "However, considerable research and technology development is needed in order to make this a reality."