Euro-View: Amir Husain on the future of AI
Artificial intelligence (AI) is coming to the battlefield – and it is poised to be a revolution not only in military affairs, but in human affairs. Indeed, militaries will win or lose, and economies will prosper or perish, based on a nation’s capacity to harness advanced, synthetic decision-making technology.
The massive scale of distributed autonomy that AI enables will herald momentous shifts in age-old planning assumptions about the difficulty of coordinating forces, the high level of variability in the execution of orders and the logistical burden of supporting not just weapons but their operators. In a recent article that I and Gen. John R. Allen, former commander of US forces in Afghanistan, wrote for Proceedings, the US Naval Institute’s publication, we sketch out the contours of AI-fuelled conflicts of the future.
To communicate the magnitude of what is looming over the horizon, one has to re-define the term “hyper war” to encompass the incredibly tight observe-orient-decide-act (OODA) loops we likely face in future. Keeping too many humans – or even a human in too many places – in the decision-action loop would become a significant vulnerability. In the move to incorporate machine autonomy in battlefield systems, the latency of human decision-making may increasingly be seen as an Achilles heel.
The three principal global military powers – the United States, China and Russia – are already making heavy investments in weapon systems and intelligence, surveillance and reconnaissance (ISR) capabilities that leverage AI. Moscow is testing unmanned ground vehicles which, it claims, outperform human soldiers. Russia’s storied MiG aviation bureau is investigating AI control for its next-generation hypersonic fighter aircraft. Meanwhile, China has fielded what it claims is an AI-powered missile, with Russia having announced plans to do the same, specifically for long range air-to-air missiles.
And this is just scratching the surface. The Dubai air show of November 2017 showcased a vast array of Chinese unmanned aerial systems (UAS) that span the spectrum from high altitude long endurance platforms to tactical systems and easily deployed quad-copters. There is evidence that Beijing is also re-equipping old platforms such as its 1950s-era J6 Farmer with autonomy technologies to transform significant numbers of this otherwise-obsolete fighter aircraft into a platform with incredible relevance for future wars.
Many AI capabilities are being added to such platforms, including automated image recognition and the automation of ISR tasks such as pattern of life determination, or the analysis of a subject/target’s habits and behaviour. In fact, just this year, a Chinese company known as “Face++” developed what it claims to be one of the world’s best-performing recognition systems based on artificial neural networks and so-called deep learning technology. Chinese efforts – both in terms of AI algorithms as well as in fielding physical systems that can make use of them – can no longer be trivialised.
So, what is the West doing in response? Is America well positioned to maintain its edge in AI over the next few years? What about NATO’s ability to assimilate these technologies in its infrastructure? Sadly, my view is that without significant adjustments to our present course, we will – in all likelihood – fall behind.
At a recent AI conference in Washington hosted by the Center for a New American Security I asked Eric Schmidt, chair of Google’s parent company, Alphabet, how long he thought the US could maintain an AI edge over China. Five years was his shocking estimate.
And I agree with him. America is cutting science and technology spending from its new federal budget proposal. It is eliminating job permits for spouses of immigrant workers and thus effectively discouraging talented researchers from coming to the country. It has fostered an atmosphere of xenophobia that discourages foreign students while failing to address the prohibitive cost of higher education for most of its native sons and daughters. In all this, America is choosing to fall behind.
Moreover, the budgetary gap in governmental AI R&D investments has become nothing short of alarming. While China has committed the equivalent of $150 billion in government spending on AI for the next five years, Washington spent $1.2 billion in 2015 and a paltry $1.3 billion in 2016.
The situation in Europe, in many ways, is even direr. To highlight just one issue, NATO currently operates with a five-to-seven year acquisition cycle. With AI algorithms and technologies doubling in capability every few months, five years is an eternity. At the very minimum, a wholesale rethink of rapid acquisition for NATO is needed.
We live in what I’ve often referred to as “The AI Century”. This technology will be the defining one of the future, and unless there are immediate and significant corrections to the course in D.C., Brussels and allied capitals, the AI competition will be over before it begins. We can only hope that better sense prevails before this hypothesis becomes history.