Back in 2018, India was one of the first countries to develop a national AI strategy, as part of which an AI task force was formed specifically for defence-related projects, resulting in the formation of the Defence AI Council (DAIC) and the Defence AI Project Agency. As part of this, as many as 40 generative AI products have apparently been developed as part of the task force.
In March this year, Army Chief General Manoj Pande stated that natural language processing, vehicle tracking, facial recognition and other autonomous systems were things that the Indian Army was currently working on as part of their generative AI capabilities.
Similarly, during the SPACE (SPace, Aerospace and defenCE) Conference 2024, Aeronautical Defense Agency (ADA) director Jitendra Jadav said that research on generative AI use for military aircraft was ongoing, though there were still several challenges to tackle.
Challenges For GenAI Integration?
Apart from computational demands and software limitations, Jadav said that AI was good at knowledge-based and statistical reasoning, but still lacked when it comes to contextual reasoning. This is something that becomes a problem when talking about completely autonomous piloting.
“We did a lot of simulations and it is a very major challenge when there is an automated manoeuvring decision. It works in some scenarios, but it is very difficult to predict all the scenarios and make it work. So that is why fine tuning algorithms is one of the major challenges engineers face,” said Jadav.
“But when it comes to contextual reasoning, we found that human pilots perform significantly better than AI. A lot more work has to be done. Autonomous combat is very difficult because it cannot replace the skills of the pilot. Autonomous combat is totally different,” he added.
Despite this, generative AI has proven to be more cost-effective and efficient when it comes to the defence sector. From assessing vast swaths of data to providing real-time monitoring, the use of AI in general has found a permanent place in the sector.
Currently, Jadav said that the agency has already integrated the use of AI within several operations, including maintenance, optionally manned aircraft and manned-unmanned operations.
However, the use of generative AI will soon see the light of day, with the agency already working on inculcating this into their maintenance operations.
GenAI’s Place in Defence
Currently, the agency itself hopes to make use of AI as a preventative mechanism. In terms of vehicle management, the agency has largely made use of the data provided by their legacy aircraft to predict when functional aircraft begin to face fatigue-related issues, including structural and thermal fatigue.
“We are trying to do a complete documentation of the pilot flight manual. The whole thing will be used for maintenance and documentation of these systems. It will be like a chatbot that is given to the maintenance person on the ground. They can ask questions and receive answers on where the problem is and how to address them,” said Jadav.
Adding to that, Jadav said that they are planning to get out of the fatigue related issues by introducing integrated vehicle environment algorithms, where the fatigue cracks and all that thing will be well predicted, based on sensors. “And based on that, we can give a remaining estimate of the usefulness of each of the systems,” he said.
However, this will largely make use of data and physics-based algorithms, rather than generative AI.
But, as mentioned above, the agency is already experimenting with generative AI capabilities, like the use of a chatbot for maintenance issues. Similarly, echoing the initiatives of several other Indian defence agencies, the use of generative AI in virtual assistance and training is being explored, allowing for real-time assistance and training for pilots and maintenance workers.
When it comes to combat capabilities, while Jadav specifically mentioned that human pilots still manage to outperform AI pilots, the use of GenAI to augment judgments is actively being explored.
“Synergetic man-machine cooperation is very important while designing the architecture of artificial intelligence. So what we learn from ground systems and ground simulation of hardware, software, architecture, and standardisation is very important to get this artificial intelligence going on in the platform.
“We plan to use it initially as an advisory role and, after adequate training, push it into maintenance,” he said.
So it seems that the use of generative AI has already found a foothold in the Indian defence sector and will continue to expand. Specifically in terms of augmenting operations to make the sector more efficient, as well as proving to be more cost-effective and reducing the loss of human life.
“This is what sixth-generation fighters are planned to do, but we will do them phase-wise. But we are very confident in making use of AI to do so,” Jadav concluded.
