The coming battle

How AI may sway spiritual feelings for political gain

Living in the secular West it is easy to believe that religion is a completely outdated idea, a quaint tradition that still persists in remote corners of the world, but one that will steadily wither as modern society progresses.

However, the news that Google is stepping up its battle against online extremism, through “the power of targeted online advertising” and machine learning, offers the potential to launch a new kind of religious warfare, a war of ideas, that may consume much of the next century.

Pitting AI against Extremism

Google is without any doubt at the top of the tree when it comes to combining online advertising techniques and artificial intelligence so the fact that it might use these skills to curb radicalisation of vulnerable people might be seen as only good news.

The global search giant announced four key initiatives for its YouTube platform that include: the use of technology to help identify extremist and terrorism-related videos, an increase in the number of independent experts in YouTube’s Trusted Flagger programme, a tougher stance on videos that contain inflammatory religious or supremacist content and an expansion of its counter-radicalisation efforts.

“This promising approach harnesses the power of targeted online advertising to reach potential Isis recruits, and redirects them towards anti-terrorist videos that can change their minds about joining,”
Kent Walker, general counsel at Google.

On the face of it, these seem entirely reasonable aims and while the story was mainly covered by the tech press, coverage has nonetheless been widely supportive. After all there are very few people, outside perhaps of ISIS training camps, that would argue that unprovoked killing of innocent people is a good thing.

The idea that we might use technology to identify and prevent the spread of violent ideology online therefore would seem only to be a good thing until we consider the potential battleground it sets out for our future.

Watershed to the century of religion

What makes Google’s recent announcement a watershed moment is the explicit move by the company to start manipulating ideological content through the use of artificial intelligence and machine learning algorithims.

While it has been obvious to most commentators that global platforms, like Google, long ago moved from being simple online social networks, the pretence that they are divorced from any ideology has still been maintained but as they start to tackle “inflammatory religious” content that separation cannot be maintained.

As Google now embarks on the identification and removal of religious and supremacist content it opens a path towards a future where machine learning algorithms may play an ever more important role in ideological preaching and evangelism.

To those in the secular West, religion is often seen as a concept that was surpassed with the arrival of the enlightenment. Post-Darwin, it is argued, we no longer need God to explain or permit either our existence or behaviour.

If asked to consider religion the average secular westerner might assume that globally religion will follow a similar path to that taken by Christianity in Western Europe over the last two hundred years — a steady decline from the religion of the state to a harmless past-time for a small proportion on a Sunday morning.

The likelihood of this secular vision of the future coming to pass however is in no way borne out by evidence. In fact research by the Pew Research Center suggests that by 2050 those unaffiliated with any religion will have shrunk from 16% in 2013 to just 13%.

In other words, if correct, the proportion of people ascribing to religious doctrine globally is set to increase not decrease. Even as things stand the religious believers outnumber non-believers several fold.

Weaponized AI

Religious practice and extremist terrorism are of course not the same thing and Google’s attempts to tackle online extremism are not solely restricted to religious radicalization, however by stepping into the arena of ideological manipulation it opens the way for those whose motives are not so noble.

Already the statistical scale and the basic design of machine learning algorithms mean that humans can only understand the broad approach for their decision-making processes, not the specific reasons.

Google may design the machine learning system but the exact process a given system uses to convert a trainee ISIS member into a moderate Muslim or turn a white-supremacist into a multicultural liberal may well be opaque to the operators.

Using the awesome power of machine learning to identify those vulnerable to radicalisation might seem like a valuable tool but what happens when extremists themselves seek to use these same tools?

If Google can identify individuals vulnerable to extremist propaganda what is to say that evangelical hate preachers cannot harness the same tools, what is the likelihood that religious zealots might want to invest heavily in such systems?

In the right hands these tools may reduce extremism but what happens when fundamentalist religious groups seek that same power to evangelize their message, to spread the faith?

“We may be building the infrastructure of authoritarianism… you build the infrastructure and it gets taken over by the people with money, with power, with authority.”
Zeynep Tufekci, techno-sociologist.

By planting its flag in the sand in opposition to extremism, Google is undoubtedly providing a valuable service but at the same time it plants the seeds for the development of a weaponized form of AI designed to wage a war of ideology for the hearts and minds of potential converts.

What is the likelihood that an organised religious group, that genuinely believes in the reality of their theology, will invest in such means to reach potential converts?

Doubtless the infrastructure and development required would be immensely expensive but well-funded organisations, such as the Vatican bank or the Saudi state, have already shown an inclination to spend heavily to acquire new followers.

Throughout history religion has been used to control populations and for overt political means. The promise of a higher purpose has proven intoxicating to the point that peoples have overlooked vast corruption in organised religion. What new power structures may develop if the process of evangelism can simply be scaled up at the click of a mouse?

“Anyone who can appease a man’s conscience can take his freedom away from him… we shall keep the secret, and for their happiness we shall allure them with the reward of heaven and eternity.” 
~ Fyodor Dostoyevsky, The Grand Inquisitor

In a world where over 80% of the global population subscribes to a religion how long before religious organisations become adept at using machine learning algorithms to spread their faiths? What means are in place to prevent an arms race of religious proselytization?

If you enjoyed this, please consider sharing or follow me on Twitter or Medium.

(Originally published at Flux Magazine:

Lochlan Bloom is a British novelist, screenwriter and short story writer. He is the author of the novel The Wave and is currently writing a novel based on the impacts of AI.