For four articles we've been sensible and realistic, forgoing the futuregazing to focus instead on the tools and technologies that AI's component parts – perception, cognition and action – are able to deliver today. But now it’s time to indulge.
Let's consider what a world built on AI could be, should the hype stand up.
Despite their evolving functionality, the AIs we already touched on in this series represent artificial narrow intelligence. They're tools designed to perform only one task as well as a human. AlphaGo, for example, can beat the world’s top human GO players (but it can't come collect you from the cinema).
From talking weather with Alexa to your bank notifying you of potentially fraudulent activity, many of us interact with narrow AI on a daily basis. Once self-driving cars become a mass market reality (the smart money says we're still a decade away), we will be handing over significant portions of activity to narrow AI.
In the short term, narrow AI and its evolution means financial institutions replacing tasks and thus jobs with robots. Customer facing roles will, in all likelihood, continue to be taken on by chatbots and other non-human interfaces, while algorithms will soon monopolise areas like quantification and redistribution, risk mitigation and value creation.
Today, 50 percent of Wall Street's trades are based on algorithms. And the percentage of insurance underwriting decisions that require manual, human intervention is decreasing every day as machines refine their data sorting abilities. For sure, tomorrow's FS workforce will need hugely different skillsets.
But to emphasise, this all still represents narrow AI. One step on, we have Artificial General Intelligence (AGI). Such units may soon be able, without getting tired, ill or bored, to perform any intellectual task as well as any human. Although there's a lot of fear tired up in a future made with AGIs – or strong AI – it still smells a lot like science fiction. Even those scientists and researchers who believe in AGI concede that we're probably 25 years away.
Nonetheless, an AGI world could see financial institutions wiped out: if everything is automated in smart, digital, AI-driven ways, human workers may not be necessary. Perhaps currency won't be necessary. If what's known as the post scarcity economy comes into effect, human needs may be met entirely by machines. With no need or desire to make money, we may be looking at a new socioeconomic system entirely.
Yikes, right? But if you thought strong AI looks like a step change for humanity, here's super AI?
Some theorise that AGI units will pave the way for Artificial Super Intelligence (ASI): AIs that perform tasks in ways that are superiorto humans. The theory goes that artificial narrow intelligence units are controlled by agents to meet confined and defined goals. But AGIs are themselves the agents: they're free to think and work towards open ended self-improvement goals; learning new tricks and functionality.
If we program today's narrow AI to self-improve – which is how most people think we will arrive at AGI – it won’t necessarily stop improving at human-level intelligence. AGIs will continue getting smarter. Exponentially so. This is where it gets weird.
Many AI researchers believe that the dawn of ASI equals a crossroads for humankind. Either a) all our problems will be solved, or b) armageddon is nigh.
It might sound weird, lofty, and like a James Cameron movie, but it's for precisely this reason that both Elon Musk and Stephen Hawking are pushing for immediate regulation and control of AI research; to build in the kind of safeguards that'll ensure ASI is able only to benefit humankind, not destroy it.
Compared to potential existential and humanitarian crises tomorrow, today's AI seems a little perfunctory and prosaic. Nonetheless, narrow AI is where the game will stay for the next generation. And it's ready to add efficiency, accuracy and new levels of performance to the financial zeitgeist. So with that, we'd like to invite you to join us at our London HQ for an AI workshop. Let's talk about this as humans, while we still can.
To register your interest, please email: firstname.lastname@example.org