'Threat assessment, intelligence analysis, (battlefield) situation review -- what has transpired in the last three hours -- was conducted with the help of AI.'
Home-grown military software applications and artificial intelligence tools were extensively employed by the Indian Army during Operation Sindoor to accelerate decision-making and enhance battlefield awareness, a senior officer said on Monday, adding that these capabilities will be further upgraded with a military-specific large language model (LLM) expected to become functional within six months.
Clarifying that the tools were not directed at any specific country but developed to give the army specific capabilities, Lieutenant General Rajiv Kumar Sahni, Director General, Electronics and Mechanical Engineers (DG EME), added that they are meant to be deployed along all of the nation's borders.
"That capability, in equal measure and when required, will be deployed along all our borders," Sahni said.
Speaking at a briefing in the national capital, the DG EME also underlined that these tools were indigenously developed and trained using data provided by the army to meet its operational and doctrinal requirements.
Underscoring that both the army and the nation are steadily upgrading their capabilities, General Sahni, who served as DG Information Systems before taking charge as DG EME, said, "Our military LLM will be fully functional, post testing and validation, in six months' time."
Revealing that a common operational, intelligence, and logistics picture was created for the army during Operation Sindoor using around 23 applications, General Sahni explained that this required collecting and processing a significant volume of information in real time.
"AI was used for multi-sensor and multi-source data fusion," he added.
Indigenous AI tools, including small language models, were used to collate and analyse the data.
"Threat assessment, intelligence analysis, and (battlefield) situation review -- what has transpired in the last three hours -- was conducted with the help of AI," General Sahni said, adding that heat maps were also generated at the joint operational control centre to support resource prioritisation during the May 7-10 conflict with Pakistan.
AI tools were also used in precision targeting during Operation Sindoor, with General Sahni citing one app -- developed by the army's Directorate General of Information Systems (DGIS) in collaboration with the India Meteorological Department -- that provided 72-hour meteorological forecasts covering areas deep into adversary territory.
"This helped artillery engage with precision at extended ranges," he said.
An electronic intelligence collation application, also developed by the DGIS, was employed during Operation Sindoor.
General Sahni explained that the AI model was trained using about 26 years of data collected by Indian agencies and armed services.
"All of the adversary's sensors, their frequency and signature, where these systems have moved, which units they are with... We can track all of that. And we have achieved over 90 per cent accuracy."
However, General Sahni underscored that the "human-in-the-loop" approach -- where humans remain actively involved in AI-enabled decision-making -- was maintained throughout the operation.
Responding to questions on Chinese support for Pakistan during Operation Sindoor and their respective AI capabilities, particularly that of Beijing, General Sahni said that the Indian Army was focused on building capabilities, without targeting a specific country, and was fully aligned with the Centre's IndiaAI mission.
"We are well on top of developments in AI," he added.
In March 2024, the Union Cabinet approved the national-level IndiaAI mission with a budget outlay of Rs 10,371.92 crore (Rs 103.71 billion).
Feature Presentation: Ashish Narsale/Rediff
