AI-Controlled Fighter Jets May Be Closer Than We Think — and Would Change the Face of Warfare

Unpiloted combat jets also offer potential financial savings. Fighter jets are expensive to build, operate and maintain, not least because of the training and equipment needed to support pilots. A 2011 study found that the life cycle cost of a surveillance drone was roughly half that of a comparable piloted platform. And cheaper aircraft are important because of the likely losses which will be inflicted on air forces in the event of a conflict with Russia or China.

Another advantage of fully autonomous aircraft is risk mitigation. As Nato militaries grapple with a shortage of trained pilots for potential conflicts between states, uncrewed systems offer a way to restore the balance without putting lives at risk of death or capture.

Therefore, one option for militaries is to expand the use of remotely piloted aircraft – drones similar to those deployed in Iraq and Afghanistan. Crucially, this would ensure humans maintain control over weapons use. The only difference with the present would be in making these systems the backbone of the fleet, rather than supplementary systems struggling to operate in hostile airspace. This would require upgrading them with state-of-the-art technologies like stealth. This helps fighters jets reduce their chances of being detected by the enemy’s radar and infrared (heat) sensors.

A step up from this would be autonomous combat aircraft, carrying the advantages of on- or off-the-loop technologies. The US Air Force’s Project Venom is training AI in modified F-16 jets for eventual transfer to drones. These drones will operate alongside crewed aircraft, as part of mixed human and machine teams. But if this AI software was retained on the F-16s (or transferred to more advanced fighter jets), it could produce a squadron of autonomous jets just as capable as those piloted by humans.

A more radical idea is to forgo traditional fighter jets altogether. Proponents of this vision imagine swarms of low-cost, expendable drones working together to overwhelm enemy defenses. While current drones have limitations in range, payload, and labor requirements, true “swarming” could change the equation.

Current Limitations
So what is stopping militaries from pressing ahead with these options? A few things. AI isn’t ready, yet. Machine learning – a subset of AI where algorithms learn from experience – underpins all this. But it still struggles with the inherent ambiguity and creativity of war. Simply putting tyres on an aircraft can thwart computer vision - the field of AI that allows computers to interpret images and videos. So training AI to operate in the full range of possible combat situations is a mammoth task. In the words of one air force commander, “robotified warfare…is centuries away”.

Another issue concerns communications, since remotely operated drone systems, especially interconnected, swarming ones, need data links. Given how much adversaries are investing in jamming these signals, designs may be pushed in opposite directions: either keeping a pilot onboard or embracing autonomy so the aircraft can keep fighting, even if it is cut off.

Yet the real limit may be a fear of crossing the Rubicon. While the US and its allies have a de facto “no first use” policy on fully autonomous weapons, the demands of warfare against an enemy willing to use such systems may erode these norms.

So, the navy’s statement is a warning: the age of the human fighter pilot might be ending. But it’s the next war that could make that decision for us.

Arun Dawson is PhD Candidate, Department of War Studies, King’s College London. This article is published courtesy of The Conversation.