WHY THIS MATTERS IN BRIEF
Algorithmic traders, like many of us, have a nascent reliance on being able to trust the information they ingest, but a new breed of AI is about to push fake news, and, by association our trust in each other and society, to the limits.
Firstly, I’d like to say a big thank you to Bartt and Libby for inviting me, again, to keynote at this year’s Battle of the Quants at London’s palatial like Royal Automobile Club which, asides from oozing racing history and paraphernalia from every corner, was also the first hotel I’ve been in that has a classic Ferrari sitting in the lobby. That said though it’s still hard to top the 62 ton tank I shared a stage with a few weeks ago when I was talking about the Future of Defence in Swindon.
Last time Bartt, Libby and I met I was their keynote at the Battle of the Quants in Frankfurt, discussing the Future of Automated Trading, and it’s clear from the audience and the attendance in London that Quant fever is catching and that everyone is drinking the cool aid. Step back fifteen years ago to when Bartt held his first Battle in New York and it was a different story and a stark reminder of how industries, often fuelled by new technologies, change dramatically.
Today’s Quant world is, arguably, like every other financial services sector, as almost everyone likes to say, “driven by trust.” Trust in each other and the trust that the data their models are analysing to find Alpha, or in layman’s terms returns, has been filtered, cleaned and validated, so it was with some fervour that I was pleased to present the “Future of Trust,” a showcase of how on the one hand new technologies like Blockchain are helping organisations trust in new ways, and how on the other Artificial Intelligence (AI), is about to, in one person’s words, not just about to bring the house of cards down, but set it on fire as well.
We’re all familiar with the role that, allegedly, fake news played in the recent German, UK and US elections, and the effect that it, again, allegedly, had on the independence vote in Catalonia in Spain, but compared to the tsunami that’s coming, where people with an agenda, whether they are foreign Governments or organised crime groups, can use advanced, democratised emerging technologies to disturb the status quo, and the markets for that matter, with much greater voracity than they could in the past, we haven’t seen anything yet.
During my presentation I lifted the lid on a new breed of creative AI’s and neural networks that can not only accurately reproduce anyone’s voice, or dynamically create images, individuals and objects that pass the canny valley test, but I also demonstrated how these new technologies, such as Generative Adversarial Networks (GANs) and Wavenets, have already been pulled together and combined with common a garden CGI techniques from the entertainment industry to create fake news videos of some of the world’s most powerful leaders that, frankly, had almost all, if not all the audience of three hundred or so professionals completely fooled.
Can you spot the fake?
Yesterday fake news was characterised by relatively crude, albeit effective, algorithmically produced blog posts and articles with photos, but tomorrow fake news is going to be characterised by videos, like the one I played of Obama reeling off fake news, that are indistinguishable from the real thing, and if you doubt me then take a look for yourself and answer me this question – if I didn’t tell you this was fake, would you believe it was the real thing? And that’s the issue. As Quants and their associated data providers suck up and analyse more data all it could take is for just one fake news item, like the one we saw recently courtesy of Bloomberg’s faux pas that wiped $22 Billion off Facebook’s share price, to go viral and be mistaken as a trustworthy element and it could have a ripple effect across the markets.
Furthermore, as the number of bots, for example, on Twitter, which now according to their latest SEC filing represent at least 9 percent of all the activity on the network, with some estimating that the real figure is closer to 15 percent, or 48 million accounts, explode, similarly Quant models that use Twitter as a tool to gauge sentiment could increasingly find those models, in their words “become increasingly biased,” but in the wrong way.
Society is about to undergo a radical revolution, and it’s one that only now the Governments around the world are trying to tackle, arguably at least four years too late, and we will soon find the fabric of society tested like it’s never been tested before, by fake news the likes of which we’ve never seen before, that looks real and convincing, and that’s produced automatically in massive volumes, and at scale, by algorithms run by people with “agendas.”
Get prepared for the tsunami and as difficult as it might be, remember that not everything you’ll see, in your Facebook newsfeed, or on your other feeds will be what it seems, and as for the world of algorithmic trading? Well, it’s time to get ready for data overload and what I called #FilterFailure.