The “fragmented” coordination between relief actors in the Philippines following Typhoon Haiyan last month underscores the need for artificial intelligence to streamline disaster response, says a team behind such an effort.

The ORCHID project — a consortium of UK universities and private firms — aims to make this possible by combining human and artificial intelligence into an efficient complementary unit known as a Human Agent Collective (HAC).

The computer systems being developed can assume tasks such as directing surveillance drones, resource management and search planning, says David Jones, head of Rescue Global, the disaster response organisation responsible for testing the software next year.

“Coordination of such a large response [after a disaster] is so challenging without technological assistance that makes data more accessible,” he tells SciDev.Net while on mission in the Philippines.

“Bringing humans and artificial intelligence together is the only way to get the job done better.”

Computers’ data-crunching abilities mean they are good at making sense of the huge amounts of information generated during an emergency from local status reports, social media, and the array of organisations involved in the relief effort.

By collecting and analysing these data, HAC systems can flexibly implement a number of activities vital for disaster response, says Jones.

These include planning the flight paths of surveillance drones, verifying the authenticity of information coming in from social media, facilitating data sharing and organising human teams based on their skill sets and current needs on the ground.
Machines not only complete many of these jobs better than humans, but by taking on these complex calculations they allow experts to concentrate on more nuanced tasks such as analysing the content of photographs or video, and strategic planning.

For HAC systems to be successful, this division of labour must be accounted for and the right balance found between artificial and human input, says Sarvapali Ramchurn, ORCHID applications theme leader from the UK-based University of Southampton.

He believes that the field trials in the Bay of Bengal planned for next year will go a long way towards showing how effective HACs can be, and pave the way for a big impact in the future.

Although exact details of the tests have not been finalised, how the complex algorithms cope with analysing environmental data coming in from sensors and social media, as well as their ability to assess the quality of this information will likely be scrutinised, he says.

Judging the reliability of a source is particularly important in developing nations, where information coming in from the local population is primarily text-based and therefore much harder to verify than when images are available from smartphones, adds Ramchurn.

Andrej Verity, an emergency manager from the UN Office for the Coordination of Humanitarian Affairs, agrees that any HAC-based system must work well using basic information, such as text messages and emails, if it is to function accurately in less developed nations.

But he is sceptical such software will be readily adopted by disaster relief actors.

“All the big humanitarian organisations already have their own procedures and software systems, so trying to do something bigger and bring collaboration across organisations on a technical level is extremely difficult,” says Verity.

Ramchurn, however, believes disaster response agencies are growing increasingly open to new technology, and so it is not unreasonable to assume they will be willing to participate in HACs in the future.