For military commanders on the battlefield, a split-second decision can mean the difference between victory and defeat. Access to accurate real-time data in a GenAI application is paramount, according to Guo Jing Hua, chief digitalisation officer of Singapore’s Digital and Intelligence Service (DIS).
“We are investing heavily” in this area, he explained, emphasising the need to amalgamate data from various sources in real-time to create a trustworthy picture of the battlefield and minimise civilian casualties.
Set up in 2022, DIS is the digital services branch of Singapore Armed Forces. In a rare glimpse into the way the Singapore military use GenAI, Guo explained the challenges the military has in using this emerging technology.
“How fast does it take to get data from sensors to weapons systems?” he asked, highlighting the critical role of speed in real-time data processing.
This urgency is evident in war games – as in actual battles – where data streams from drones and ground troops are fed to a command post, enabling commanders to swiftly identify targets and launch attacks with the support of AI.
Guo gave his comments at a recent panel discussion on data streaming platform for modern AI and the importance of real-time intelligence and innovation panel discussion, organised by data streaming vendor Confluent.
Real-time data accuracy is equally crucial in the business world. Banks, for example, remain cautious about GenAI due to potential risks such as data hallucinations.
Panellist Vivek Chari, head of data engineering at Standard Chartered Bank, acknowledged these concerns, stating that the bank prioritised addressing risks before deploying AI applications.
However, he recognised the potential benefits of GenAI in areas like language translation, where it can significantly reduce costs and improve efficiency.
In the bank’s retail and wealth management business, for example, letters from the chief investment officer have to be translated from English to Korean and other languages for customers in various countries.
Translation is big business, and it’s an area where the bank spends about US$5 million to US$6 million annually. Using GenAI has led to cost reduction and improvement in productivity.
There are also discussions in Stanchart, said Chari, to hire in-house copywriters instead of outsourcing content creation. They can leverage GenAI to generate more relevant content for the bank’s global audience, he added.
The panel discussion also addressed the importance of responsible AI. Panellists agreed that human oversight is essential to ensure the relevance and accuracy of AI-generated information.
A recent Confluent survey revealed that data integrity was a significant hurdle for GenAI progress. Suvig Sharma, area vice-president for Asean at Confluent, stressed that data streaming capabilities can significantly improve data integrity by providing accurate real-time data.
He cited the example of American bank Capital One, a Confluent customer that successfully identified fraudulent transactions in real-time using an GenAI application, thus saving each customer US$150 in potential fraudulent transactions.
Being able to process the data as it is received would help companies react in a more timely manner during operations, he added.
The debate continues on the optimal balance between AI automation and human intervention in decision-making. “Do we rely on a fully automated cycle or add a human in the loop?” asked DIS’ Guo.
It’s a complex issue that remains a topic of discussion for militaries and businesses worldwide.
