Wednesday, June 29, 2016

Why AI revolution could defeat US air superiority and hence strategic hegemony

Pictured The Russian Brahmos II hypersonic cruise missile now under development by Russia and India.
The Bhahmos I is a supersonic anti ship cruise missile already deployed

The news that the Alpha AI programme can defeat human fighter pilots in simulated air combat ( BBC) shows that within the lifetime of the F-35 (the hideously expensive, frequently delayed, fighter bomber tasked with replacing half the US arsenal) unpiloted combat aerial vehicles will become the 'top guns' of air combat maneuver.

Not only will UCAVS be able to pull Gs no human body could withstand (which they can do now) they will also be able to outfly human rivals tactically. Add to that the ability to build vehicles without the need for a cockpit and all that opens up for stealthy design and you have a potentially new calculus for air superiority.

Fascinatingly the Alpha AI jet pilot programme does not require a huge computer to run on. In fact the average 2015 desktop will do. Or a high end cellphone. But most important Alpha is a learning machine. It doesn't just run through a bunch of routines. It accepts feedback and recalibrates. It can practice against people or it can practice against itself. Moreover it can practice against itself without burning any fuel or doing anything to the environment.

If you put AI at the core of your air combat vehicle you change the entire basis for air warfare. For a kick off you can afford to lose "pilots". Training is no longer an investment of thousands of flight hours and years of professional development. It's a case Ctrl-C Ctrl-V. Done! Each one is simply a software instantiation you can clone in seconds.

Instead of planes like the F-35 all you need is a Matroyschka doll of missiles. The long range "bus" rocket delivers a pair of Air Combat Manouver drones to the combat zone whether it is 2,000 or 12,000 miles away at hypersonic speeds. The air combat manouver drones use missiles to achieve air superiority until it's time to self destruct. Then they find targets on the surface and fly into them at 8,000km/h. Boom! The obvious targets: other air superiority asset bases - air force bases and carriers.

The ACM drones can be scramjet or rocket powered, carbon-fibre stealthy and a fraction the size of conventional aircraft. Delivered by missile they could easily be far faster than any conventional aircraft. Conventional fighters would be outperformed, outfought and ultimately uneconomic.
Transport aircraft could also provide a "mother ship" role for ACM drones for situations where an IFF problem might arise.

Consider the economics of this. Right now the F-35 is the most expensive weapons system programme in history, having cost US$1.5 trillion. It is $163 billion over budget and 7 years late. Nice for Lockheed Martin but not so great for its customers. When it's finally delivered the F-35 will have a max speed of almost 2,000km/h, a range of 650nm and be able to pull 9Gs. It will be able to carry 6 missiles and have a gun with about 3 seconds of firing ability.

Most of the time the F-35 will be flown in training missions. That is it will be used to programme its pilots. Only America really has the economic power able to sustain a large full time air force that trains so much. But that is a very slow and very expensive upgrade cycle compared to an AI system which never stops training and has uniform and immediate performance duplication.

Building UAVs and rockets is surprisingly cheap. The Brahmos I missile costs $2.1m each but if the Brahmos II with AI ACM took down a plane and a ship the ROI is over 100:1  Larger nations with reputations for military innovation such as Russia and Israel will see the implications in Alpha immediately. The technology is not beyond nations such as China, Iran, Turkey, India, Germany, France, Sweden, Japan, Korea, Australia, Singapore etc either. Even nations as small as New Zealand can build fast rockets.

Obviously the US will want to invest the most in this technology, just as it does in many things. But US military investment - as we have seen with the F-35 - is neither efficient, nor all that nimble. The US has nothing to match BrahMos even now. The Americans are good at reliable - even downright slow - programme development. But US manufacturers thrive on the political infighting over specifications changes, mission changes etc. The list of programmes US services have started but been forced to kill because of budget blowouts is an embarrassingly long one. This means that American programmes are wildly too expensive and embedded in the institutional excesses of their military agencies.

But what happens when America wakes up one day to find that Russia has an ICBM deliverable airforce of hypersonic robot fighters able to wipe out its carrier and ground wings in the air? Or over the next fifteen years a dozen countries develop hypersonic ICBM delivered AI fighter swarms? Yes, laser weapons are a potential defence for her carriers and air bases but the ability to contest the skies themselves will hugely change the status of the US as the nation with the world's greatest air force. Her ability to defend will remain but her ability to strike at will be drastically curtailed.

It is too early to say what the future holds exactly, but one thing is for sure. The F-35's viable lifetime just got a lot shorter.