Artificial intelligence (AI) and other disruptive technologies came under the spotlight in a series of thought-provoking sessions at our annual Financial & Risk Summit in Toronto.
This year’s event — “Revving the Innovation Engine” — brought together thought leaders and practitioners from across financial markets and risk and compliance communities for a full-day series of interactive discussions.
It was also an opportunity to introduce members of our #TRFinRiskCanada40 list, which showcases Canada’s top social media voices in finance, innovation, and risk.
One of the highlights of the summit was the Reuters Newsmaker interview with Geoffrey Hinton, a vice-president with Alphabet Inc’s Google, who is often referred to as “the godfather of deep learning.”
He discussed how artificial intelligence is being used today by internet companies such as Google and Facebook, and also in a wide variety of other industries, from financial services and healthcare, to automotive and manufacturing.
Hinton spoke about the promise and challenges of AI, with a particular focus on neural networks.
“We need to move to a world where we have to be comfortable not knowing how and why machines make predictions,” he said.
Hinton led a group of scientists at the University of Toronto who developed some of the key algorithms that neural networks use to crunch massive quantities of data, training themselves to identify patterns so they can mimic the way the human brain would perform tasks.
Humans, however, can’t reliably recall all the minute mental steps that lead to their decisions. Thus, pinning down exactly how a given decision was reached by a neural network overseeing a business process may be nearly impossible to communicate to regulators.
When asked about the risk of technology leading mission-critical systems awry, Hinton reminded the audience that humans build the algorithms for these neural networks.
“You have to worry about people manipulating these things. There is a lot of work that remains to be done on how to avoid that kind of adversarial attack on these systems.”
In other words, risk is inherent.
AI and big data
The next session sought to separate the hype from reality in AI technology.
Delegates were told how AI needs to run massive amounts of data — at least tens of thousands of data points, but preferably many more — through key algorithms in order to function.
The monumental increase in the availability of big data has facilitated advances in cognitive computing over the last decade, particularly in areas such as machine learning, natural language processing, and predictive data analytics.
As a result, new business practices have emerged, such as the use of AI in robotic processes within the client workflow for wealth managers.
Will AI drastically alter the landscape of financial risk management? Big data may hold the answer.
Rahul Deshpande, Senior VP of Digital Integration Strategy at Mastercard, cautioned: “The way you handle your data is key when it comes to enterprise AI.”
Whether an enterprise is trying to build a self-driving car or predict market dynamics, it will need people who can assess technical, legal, and regulatory impacts of the AI systems they build.
New era of risk management
Big data has become an integral part of the framework of our evolving regulatory environment.
Accordingly, the financial industry will investigate ways to use the power of technology, such as AI, to manage risk and serve consumers better.
Will these much-hyped innovations become critical to the practice of financial risk management?
When considering the role of technology in compliance, panelists relayed that the primary focus should be on starting with simple, elegant solutions to the “easiest pieces” and “low-hanging fruit,” while holding off on more advanced projects.
They also noted that the advantages of cloud-based tools and vendor systems (vs. in-house software) were worth considering, but only in light of the enterprise’s needs and goals.
Panelists felt that compliance stakeholders — from regulators through to various internal teams and the client — may not yet be ready to take a full leap on technologies such as AI and blockchain.
Blockchain may, however, hold out promise for its ability to certify provenance and uniqueness of an asset.
Innovation in cloud computing
Beer spoke in detail about data security and ‘kill switches’ in the cloud.
There are many great tools to prevent loss and protect consumers, but there are also many opportunities for data privacy missteps. Often, technology services are “stacked on each other” (i.e. through an API) to perform different jobs.
However, Beer warns: “You own the customer experience. Even if someone else does the work for you, you are responsible for ensuring it got done.”