Artificial intelligence (AI) is now becoming a reality in Finance.  Silicon Valley companies are spending billions on making AI more intelligent than humans. AI is already impacting our everyday lives – through the adverts we see on Facebook, the shows selected for us on Netflix to the Google automated assistant being able to book meetings for us. 

It’s predicted that all but a very few jobs will be impacted by AI. In this age of AI, we will need to rethink of how we add value. Accountants, lawyers and tax experts are expected to be replaced by AI in the next generation or two – but there is hope, it’s known as Moravec’s Paradox.  As my former schoolmate, Elon Musk is quoted as saying, “Humans are Underrated”

Moravec’s Paradox

Moravec Paradox states that while "it is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility"...  In other words, computers by their nature can do certain things very well, but struggle with basic things a 1-year old may not.  

Elon Musk discovered this the hard way when he tried to have complete automation at the Tesla factory.  Although he had spent millions on a robot to cover the batteries with a fluff to reduce the noise – the machine kept on failing, because it couldn’t pick the fluff up as easily as you can with your hands.

So where do computers excel? 

One area is trial and error – they can rapidly fail at something until they figure out the best alternative.  DeepMind a London based startup acquired by Google in 2014 was able to demonstrate how a computer could maximize the score on a classic Atari game, Breakout.  The computer had taught itself to play the game and develop new tactics to surpass the human’s highest score.  Although this is a relatively basic application of AI, it demonstrates the key elements: Clear goals (gaining a higher score) and the ability for an instant feedback loop (have I exceeded the highest score).  

How AI will impact the office of the CFO

Unfortunately, reality is slightly more complicated than that especially in the world of Finance – success is not always that clear to define and the feedback loop may not always be that quick.  Today’s CFOs are required to be Oracles, predictors of the future. At FP&A we have learned that they’re required to be strategic so they can “divine the path”. The next role of the CFO is not as “accountant” and more a champion of data.

The real value of this big data and new technology is in enabling CFOs to analyze data in a meaningful way. FP&A surveys have found that new solutions are being architected from the ground up to capture more than traditional accounting transactions. They capture a broad set of financial and non-financial data and present insights via modern real-time dashboards. These dashboards may have KPIs, charts, reports, lists and even social feeds for analyzing the most important metrics to an organization’s success.

We’re seeing AI solutions emerging all over the financial workspace freeing up CFO’s time. CFO’s can now spend their time focused on the organization’s future versus only doing menial accounting and management.  An aware CFO will rise in stature, perceived as a valuable strategic resource based on data instead of a glorified bookkeeper.

Where AI starting to play a role in Finance

According to Forbes in May of this year, “To date, we’ve seen broad AI applied to big data and analytics. Now we can expect to see industry-specific solutions arrive in verticals such as finance and accounting. In 2018, we’ll begin to see CFOs take the leap to incorporate meaningful AI-based technologies into their day-to-day workflows. With this new bandwidth, we’re going to see CFOs rise from back office leaders within organizations to those who have a voice and a distinct point of view on the future of business.”

From my experience we are starting to see AI planning a larger role in the cash conversion cycle – specifically around Accounts Receivable (AR).  I was recently working on a proof of concept for a solution that helped to identify and optimize payment terms and discounts to ultimately improve the cash collection.  The results of the findings were that offering discounts drove improvements in cash collection, but the amount of the discount had a very little impact.

The unknown unknowns

The real advantage of AI in this scenario is to uncover the correlation in the unknown unknowns – these things we didn’t know we didn’t know.  As AI is able to look at data in many different ways and rapidly test correlations across a broad range of dimensions – it may discover something you didn’t even know you were looking for.

Some of the current challenge with AI in Finance

Sadly, in this proof of concept we were unsuccessful in deploying the full solution because of the following reasons:

1. Finance data is typically not available in the volume that would make AI a success as most external finance data is private and internal data is limited
2. Data can be messy – In our case the AR data was in 1 of 11 different solutions with different structures making it difficult to get an overall picture.
3. The current cost of AI is high, so the return has to be higher, making the ROI sometimes difficult to prove out

We continue to look at new solutions to mitigate these challenges, but overcoming these will require Finance teams taking a more direct role in managing their data all the way from data governance and master data management.

Where will AI have an impact in the future?

The quest for translating data into valuable insight is the new gold rush, making companies like Apple, the first trillion-dollar company, Amazon, Alphabet (Google) and Facebook the richest companies in the world.  Successful companies in the future will be those that can harness data to give them a competitive edge against their competition.

The real advantage is using AI in executive decision making

One area where I see AI playing a larger role in the future is in executive decision making.  Today executive decision making is done by a combination of gut feel and supporting data.

• Reduced bias – one of the biggest issues with data driven decision-making is there will always be bias – AI can help support decision making in a more unbiased, fact driven way
• What AI does well is identify things we never thought of – the unknown unknowns.

Next steps for Finance

The quality of AI will all depend on the quality and quantity of data. Being able to standardise your data across all your ERP’s and supporting solution will be key to the future.

Investing in the right talent within Finance: As the majority of internal data is touched by the Finance organization, building a team with data analytics, modelling and data science experience is key.  CFO’s should also take steps to make insight generation part of the core DNA of the Finance organisation.

Conclusion

AI is at an early stage and we are still learning how to take advantage of these new technologies – Please feel free to connect, to continue the conversation and share your experiences  
 
Bibliography:
“To be a Machine” by Mark O'Connell

The blog was published on LinkedIn.