Dr Catriona Wallace is a leading Artificial Intelligence expert. She is the Founder and CEO of highprofile AI FinTech company (ASX: FGO) Flamingo AI, which provides Cognitive Virtual Assistant platforms to help financial services companies improve online sales conversion rates. Dividing her time between New York City and Sydney, Catriona has received a slew of industry accolades, including the Advance Australia’s Technology & Innovation Award in 2017. Catriona has also been inducted into the Australian Business Woman’s Hall of Fame.
AI is spoken about extensively in the context of the financial services industry. Can you tell us more about what it really means?
AI is a concept that was coined in the 1950s to describe how human intelligence could be mimicked by computers. Despite this early start, over the last 50 years, the level of computational power required to run the algorithm that created an AI engine, or software, just hasn’t been there. However, right now we are at a stage where two important factors are converging – we have computational power and we have huge terabytes of data, so we can train the algorithms. So, for these two reasons – we are at a time where AI has finally come of age. Attention now is turning to a particular type of AI called machine learning, where the machine is able to improve at tasks with experience, under no supervision. AI can be loosely categorised into two camps – general AI (think Siri, Cortana) and Narrow AI (think AlphaGo). The reason why we are seeing failure in some banks with AI deployment is that people are trying to deploy general intelligence machines that are broad, but not deep in their capability. Beyond general and narrow Intelligence is ‘Super’ intelligence’ – where the machines actually do things beyond what humans can do. That level of AI is predicted to come in in 2030 and there’s a term associated with that which is ‘singularity’ – the hypothesised time when super intelligent machines improve themselves without human involvement. That’s the bit that many AI commentators, including Elon Musk, are worried about.
What is the status of AI within the banking sector?
A lot of the banks have gone out and brought the traditional AI engines and they are trying to get them to answer customer enquiries – and it’s often failing. Within the AI, artificial narrow intelligence is what the banks should initially focus on. Artificial narrow intelligence is when we really limit the domain that the machine has to learn in. Flamingo AI’s machine ROSIE, who is narrow AI, is provided to banks and insurance companies and she just has to learn about one product and one function at a time. It’s very narrow and she can learn within a couple of weeks all the questions a customer might have around that narrow case. Conversely, an artificial general intelligence engine would be something like IBM Watson, Amazon Alexa, Google Home or Apple HomePod, where they are expected to know a lot of subjects. This is still not completely mature, so we are probably about three years away from the general intelligence machine being able to really mimic human functions and do effective tasks. In the banking sector, there’s a Gartner statistic suggesting 40 per cent of banking and finance jobs will be replaced by machines in the next 10 years. Within the next three years, 30 per cent of customer interactions will be performed by machines. The transformative power of AI is coming like a freight train and will be a huge change, described as the 4th Industrial Revolution or the 3rd Wave of technological change. The big AI commentators compare this to the same level of change experienced when electricity was introduced to the industry. The future of banking should be not only employees and machines working closely together to deliver great customer experiences, but it should also be customers using machine learning and AI to interact with their bank. Machine-to-machine and customer-machine to banking-machine can already happen for those organisations who are progressively thinking about this.
In terms of AI implementation, how advanced are Australia’s financial institutions compared to the US?
Because I work predominately in the US, it is obvious to me that the level of education, research and knowledge of AI in Australia is very poor in comparison. There is a very different conversation that we have in the US than in Australia and it is a real concern. There is not enough knowledge, conversation or debate about the risks and benefits of AI in the Australian market. The average budget a tech company will spend on AI is around $10 million, so it is a significant investment. AI is the fastest growing tech sector in the world at the moment, growing five times the annual compound growth-rate each year. There definitely needs to be a discussion at the board and executive team level and an AI strategy should underpin not only their digital strategy but also their business strategy. It needs to be one of the top three strategic imperatives for banking going forward. The very best way to start is to undertake experimentation in small proof of concept and pilots of no more than $100,000. If you can’t trial and test an AI product under $100,000 then there’s a problem. There is no need to be spending millions of dollars on AI. Banks need to start very small and prove out commercial results first. If there are no commercial results, then don’t proceed.
Have Australian banks effectively grasped AI?
Australian companies are still really in the early stages of testing and trialling. We know that the banks are trialling AI to some extent, but a lot of them are using contact centre vendors or chat bots which is insufficient. What we now need is a much better education dialogue around AI strategy for banks where it’s not just the large tech vendors that they consider working with. There needs to be a lot more engagement with the new AI machine learning companies who really are the ones trying to solve problems with AI.
These insights are drawn from the TAS Banking Industry Report 2018. Read the full report here.