I'm reading and hearing more and more about the arrival of AI and the application of Machine Learning.
As usual, separating actual from aspirational gets tough when we are riding Gartner's Hype Cycle. While we speculate on the “Black Swan” of Artificial Super Intelligence, ‘narrow’ (or ‘weak’ AI) is becoming a reality. The early applications of emerging AI/ML have included:
- Recommendation engines - others like you also bought or other movie titles that may interest you
- Natural Language Processing engines - that interpret and act upon the spoken word of which there are now many including Alexa, Google, Siri etc.
- Pattern or image recognition engines - that can replace hours of tedious human endeavour interpreting and taking action (perhaps via the internet of things, IoT) on visual stimulus ranging from saving wildlife to one day soon replacing the need for radiographers to interpret x-rays manually.
- Chatbots – aka IVR all grown-up, where today you can buy insurance entirely un-molested by actual human contact or begin any range of procedural activities online
What does this mean for you? In a personal context – an enormous amount. Already today you have probably touched a series of these services that simply did not exist five years ago, and that is just in the last few hours.
What about in your business? Well, in the vast majority, much of it will simply come to your business as pieces packaged into broader offerings. They will appear as naturally smarter and better ways of doing business. A smaller group of business will innovate internally around these services, and at the really big end of town, large amounts of money will be won and lost through productive engagement with these technologies.
If you’re reading this you are very likely a Finance person or someone who works a lot with groups of Finance people – you are reading a CALUMO blog post after all. So what does this mean for your team?
We think a lot about technological innovation and its effect on Finance teams and the work that they do – creating and using innovative and truly effective technology is at the core of what we do. We think the following list may be useful to you in how you consider the arrival of these new technologies in your Finance team and for the benefit of those that you serve:
- For now, AI and ML are only as good as the data. It's not quite garbage-in, and garbage-out but models require data to be fed, and in the majority, data is historical. Things that have not been important in the past will not appear in the algorithms used to project the future. We all know that business has many variables, which remain elusive to capture as interpretable data.
- The corollary of that is that while the practical application of AI/ML in your business may be a little way out, the timing is NOW to get your house in order regarding data. If you are remain mired in spreadsheets, suffering under too much friction in trying to get information or still doing manually what can be automated, you are making an avoidable mistake and creating a barrier to engagement with these technologies.
- Trust and explanation. To believe, to be motivated, to be moved – humans require both to varying degrees. Deep Learning (which is central to much of the current buzz about AI) is just not at all good at explaining itself or its logic right now and indeed may never be. How much of a brake that is going to be on widespread acceptance in business we are yet to see, but its fair to say that much of the way we do business today is founded on trust and a capacity to explain motivations, gives and gets, not summary answers. Imagine the Chairman of the Board attempting to explain to the stockholders the reason for an abrupt, unforeseen loss at whose heart was acceptance of the “new algorithm” for which there was no real understanding of its efficacy before its implementation other than being the latest algorithm.
- As it relates to prediction and projection specifically
- To imagine an AI-driven budget will replace existing budget processes is to misunderstand the critical elements of the process. We see the budget process to be as much about cultural alignment and commitment as it is about prediction. Well executed, the budget process creates the opportunity to think above the noise and exclusively about the future. To envision it and commit to creating it in concert with colleagues and partners. No amount of computer-based prediction is going to create the kind of alignment that an effective planning process can.
- The opportunity here is to allow a new voice into the conversation. In typical processes, we take our perspective and then we test it against things like past performance. With AI we get the opportunity to also compare our expectation with a new voice, the AI or ML based prediction, and explain any variance. This creates new ways for us to examine our assumptions
- Continuous forecasting processes can benefit greatly from predictive algorithms. Here we are making strategic re-forecasts of major drivers in order to better understand the likely future. The emphasis is on the efficiency of the process (so it can happen in short intervals) and on a limited range of factors. That said, human engagement in the process remains critical to its success. Arguably the techniques are not brand new however the tooling has become far more accessible, and this has been a core objective of the platform technology leaders like Google and Microsoft.
We are working with some of our clients and actively on the lookout for others to work with us in the application of Machine Learning to critical issues they have in their business. If you've got a question or would like to explore some ideas, please reach out to us.