Chapter 28. The BVR Revolution: AI That Sees First and Shoots First

Revolution in Beyond Visual Range (BVR): AI that sees first and shoots first Do you remember the scene in Top Gun where Maverick makes eye contact through the canopy right in front of the enemy plane? Forget it. That has now become history. In May 2025, an event occurred in the sky proving that fact. It was the largest air war between India and Pakistan since World War II. And the battle was not a close-quarters battle (dogfight) in which the two sides bit each other's tail and fired machine guns. The pilots did not see each other.

It was a Beyond Visual Range (BVR) engagement in which long-range missiles were exchanged at a distance of 100 to 150 kilometers. The Chinese-made PL-15 missile mounted on the Pakistan Air Force's J-10C and the European-made Mi Tier missile mounted on the Indian Air Force's Rafale clashed in combat for the first time. The PL-15 is a missile equipped with an Active Electronically Scanned Array (AESA) radar seeker and dual pulse rocket motors, and is said to have a range of 200 to 300 kilometers. Mid-route updates are possible via data link during flight, allowing tracking of even evasive targets.

Meteor maintains kinetic energy over long distances with its adjustable ramjet engine. It is designed to form a 'No-Escape Zone' so that once aimed, it cannot be avoided by maneuver. According to reports, during the encounter, a PL-15 fired by a Pakistani J-10C shot down an Indian Rafale. This battle clearly demonstrates the nature of BVR engagements. Rather than ‘seeing’ each other, pilots see stochastic shadows created by sensors. Both sides deployed airborne early warning and control systems (AWACS) to exchange real-time targeting data.

The flat terrain of the Punjab Plain and high-altitude AWACS support gave both sides a significant radar horizon advantage. Long-distance engagement was possible without being obscured by terrain. I also did BVR engagement training when flying the F-16. When a dot appeared on the radar, the biggest problem was identifying whether it was friend or foe (IFF). Non-Cooperative Target Identification (NCTR) technology existed, but it was not perfect. We often had to close within visual range, and at that point the advantage would disappear. However, with the intervention of AI, this uncertainty is decreasing.

AI mounted on 6th generation fighter aircraft seeks to solve the ‘identification dilemma’. AI instantly detects the subtle tremors of radar reflected waves, infrared waves from engines, and the patterns (fingerprints) of radio waves flowing from electronic warfare equipment.

Analyze. “This is a Sukhoi-57, with two wingmen, and is currently armed with four air-to-air missiles.” It takes less than 0.1 second to make this decision. What's even scarier is 'predictive shooting'. When AI outscored human veteran pilots 5-0 in DARPA's AlphaDogfight experiment in 2020, it wasn't just reaction speed that the AI ​​showed. It was a ‘prediction’. Before the human pilot even moves the stick, the AI ​​probabilistically calculates where the human pilot will move and fires the cannon in advance. When this technology is applied to BVR, it becomes terrifying.

Based on the enemy aircraft's current speed, altitude, aircraft type, and past tactical database, the AI ​​predicts, "This enemy aircraft will make a sharp turn to the left in 3 seconds to avoid the missile." And it instructs the missile on the optimal interception path. Rather than blindly chasing after it, guard the corner of the road and then attack. This dramatically increases the effective range of the missile. In 2023, DARPA's Air Combat Evolution (ACE) program tested an AI-controlled dogfight on the X-62A VISTA fighter jet.

In the 12th flight test, the AI ​​agent performed a one-on-one BVR engagement with a simulated enemy and even performed a within visual range (WVR) dogfighting maneuver. The AI ​​flew at 1,200 miles per hour (approximately 1,931 kilometers per hour), matching the speed and agility of human pilots. “We actually pitted the X-62 against a manned F-16,” said Lt. Col. Marianne Callen of the U.S. Air Force Flight Test School. They even engaged in head-on combat at a distance of 2,000 feet (approximately 610 meters) and approaching at 1,200 miles per hour. This was the first AI vs. human within visual range engagement in history.

AI has shown its own unique tactics. We used sensor data to make split-second decisions and adjust our tactics. Unlike human pilots, AI was able to quickly process large amounts of data. “The X-62A team demonstrated that cutting-edge machine learning-based autonomy can be safely used for dynamic combat maneuvers,” said then-Secretary of the Air Force Frank Kendall. What DARPA and the Air Force are emphasizing is not the dogfight itself. Dogfighting is the most difficult flying style. If we teach AI this, it will become easier to fly other missions autonomously. In BVR engagements, AI's strength is even more overwhelming.

In the race to detect first, fire first, and shoot down first, AI is faster than humans. DARPA's Artificial Intelligence Reinforcements (AIR) program has the explicit goal of tactical autonomy in multi-aircraft BVR air warfare.

BVR becomes even more lethal when networked cooperative engagement (CEC) is added. Picture the situation. I turn off the radar and take a detour to the enemy's flank. My stealth unmanned wingman, far ahead, detects an enemy aircraft. That information is fed into my missile in real time. I fire the missile without even looking at the enemy or turning in their direction. You can also shoot backwards (High Off-Boresight). The launched missile is not guided by your fighter jet, but flies to the enemy based on information sent from a drone or satellite in space ahead.

From the enemy plane pilot's point of view, death suddenly rains down from the sky before the RWR sounds. He is shot down without even having a chance to say, “Where on earth did you shoot from?” This is ‘silent assassination’. The U.S. AIM-260 Joint Advanced Tactical Missile (JATM) is being developed to counter the PL-15 threat. Range and performance are classified, but it aims to surpass the PL-15. Russia's KS-172 claims an engagement range of more than 400 kilometers and boasts a speed of more than Mach 5. The European Meteor uses adjustable ramjet propulsion to create an extended no-avoidance zone. This is a BVR arms race.

Amidst all this technological advancement, there is one thing we must not forget. Who makes the decisions? When the AI ​​says, “An enemy plane is certain, 99% chance of being shot down,” who has the authority to pull the trigger? This is why DARPA's ACE program places 'Trust' and 'Human-Machine Collaboration' at its core. No matter how smart the system is, if humans don't trust it, they delay making decisions. Delay means loss in BVR. AI does not replace ‘decisions’, but sorts the candidates for decisions in real time. “If you fire now, your chance of hitting is 45%. If you get closer for 10 seconds and shoot, it’s 85%.

However, if you get closer for 10 seconds, the enemy’s chance of a counterattack increases from 20% to 60%.” Humans can't do these calculations like they can breathe. AI does. The battlefield ultimately costs you that difference. The conclusion of the BVR revolution is simple and unpleasant. The ace pilots of the future are not the ones with the best flying skills. You will become someone who trusts and knows how to manage the best algorithms. The sky is getting wider, but the window for decisions is getting narrower. Whoever catches the narrow window first wins. The era of Top Gun is over.

The ruler of the skies is no longer the pilot with the best control. Whoever has the smartest algorithms and missiles that fire first from the greatest distance rules the skies.