Finance has always utilized technology. From the humble abacus (2700 - 2300 BCE) to adding machines (1642 CE), to today's PCs. The last 40
Future Technology - Artificial Intelligence (AI)
In short, AI refers to a device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals. This definition is intentionally broad as AI could refer to a software application designed to perform bank reconciliations or Terminator robots. When we think about AI in terms of software, one guide is that the application can do things that it was not specifically programmed to do. It learns based on the outcome of previous events.
There are two categories of artificial intelligence:
1) Narrow AI
Narrow AI is focused on one very specific topic. Still, in its early stages, we can already see developments in narrow AI. AlphaGo Zero is a computer program that taught itself the board game Go by playing against itself. Within days it was better than the best Go-playing applications which themselves beat the best human Go players. Then in 4 hours it taught itself Chess and could beat the best Chess programs in the world (the ones that beat the best humans). Finally, it taught itself Shogi (a Japanese version of Chess played on a bigger board) in 2 hours and was better than the world's best Shogi program.
Some other interesting examples:
- Google's self-taught Go playing AI crushes the best human player,
- AI in Tesla will predict your destination,
- MIT's autonomous cheetahs figure out how to navigate obstacles entirely on their own,
- Boston Dynamics SpotMini locating, recognizing and opening a door,
- Amazon's use of AI to improve their business processes
2) General AI
General AI (sometimes called Artificial General Intelligence or AGI) is a hypothetical machine that exhibits behavior at least as skillful and flexible as humans across a broad set of topics. This is the Hollywood sci-fi that many people immediately think of.
Should AGI become a reality, it promises to change every aspect of our world fundamentally. Many experts in the field are in fact worried that it will lead to disaster. To learn more, I recommend an excellent Ted Talk by Sam Harris (below) or Nick Bostram's great book on the topic.
AI & the finance department
What will AI (narrow or general) mean for the finance department?
In the near term, repetitive and time-consuming tasks are being automated at an accelerating pace driving massive improvements in efficiency. These advancements tend to be less risky and relatively low cost and thus are immediately appealing. The benefits are enormous, freeing your team from the mundane, repetitive work allowing them to tackle the difficult, often more rewarding tasks that they may not have time to tackle today.
As the technology improves, analyzing data and making decisions based on this analysis are also likely to be automated. We expect that the move to rely on AI for these most complex and difficult tasks - the ones that require years or decades of experience in humans - will be initially slow and cautious.
Nearly all studies suggest that the entire finance function (like many others) will be entirely automated by AI exclusively or a human-AI hybrid.
The process has begun. Recent research indicates 46% of CFOs in large companies already use narrow AI in some role in their organization and another 30% are investigating its use.
Today's Technology - Robotic Process Automation (RPA)For decades finance departments leveraged spreadsheets and more recently databases. These technologies allow for simplistic automation such as calculations or manipulating manually defined groups of data. While basic, this automation has allowed finance departments to complete larger & more complicated tasks with fewer hours of investment.
RPA takes automation to the next level and can be considered a pre-cursor to AI. The three characteristics generally associated with RPA:
- It does not require programming skills on the part of end users,
- It does not require complex, disruptive integration with existing systems,
- It is designed to be managed and even implemented by a business user,
- data analytics & monitoring systems that check thousands of variables in the way it’s been taught, against the benchmarks that have been provided. When exceptions are identified appropriate individuals are notified automatically and escalations occur at pre-determined intervals.
- automation of reconciliations between any data sources with automated adjustments when common deviations (think bank charges in the context of a bank rec) occur.
- automated modification of language in your MD&A based on the significance of an identified variance. So if the variance between actual and budget is greater than a predetermined threshold, entire sections of the report turn on and standard analysis is performed,
- automated analysis of documents to find violations of integrity,
- automated balancing of amounts across large complex documents.