Defence deploys Watson for psyops analysis

Australia's Defence department is utilising IBM's Watson cognitive computing platform to comb through psychological operations data to work out the target groups Defence staff should focus their attention on.

Psychological operations spreads information to specific adversary, friendly and neutral people and groups in an attempt to influence their attitudes and behaviour in a way favourable to the instigator's political or military objectives.

Over the six months to March, Defence undertook a trial of the Watson cognitive computing engine to see how and if it could be used more widely throughout the business.

Defence's IT team fed unstructured and unclassified psyops data into the software to help with target audience analysis, which works out which groups would be most receptive to the propaganda efforts.

The pilot found that Watson could reduce the timeframe of traditional target audience analysis from between three and six months to just two weeks, Defence CTO Mohan Aiyaswami told attendees at this week's CeBIT conference in Sydney.

The IT team also discovered unexpected linkages and insights through the unstructured data analysis, like criminal records and strong religious affiliations of targeted individuals.

"Watson has been very helpful in doing this analysis. There's lots of unstructured data - it's like picking a needle in a haystack. If you can can understand lots of seemingly disparate bits of information and put them together to make sense... [that's] very important to us," Aiyaswami said.

"Machine learning and cognitive computing will be a very important part of our architecture going forward."

Defence is now standing up a dedicated instance of Watson to use more broadly throughout the business following the success of the pilot. It is yet to decide whether to host the software in house or with a data centre partner.

Once the department has its own instance of Watson, the possibilities will be endless, Aiyaswami expects.

He's already asking whether one cognitive engine is enough.

"There are lots of interesting ways forward," he said.

"I'm [currently working with IBM] on whether one is enough, because we can let each cognitive engine be a different individual - so we can have one cog that thinks like a man, another cog that thinks like a woman, whatever it is. So we have an ecosystem or a network of cogs thinking differently."

The department joins fellow government agency the Immigration department and enterprises ANZ Bank and Woodside Petroleum, among others, to adopt the cognitive computing technology within their operations.