There is a tremendous amount of hype and misinformation when it comes to what software is capable of doing and what is necessary to make software actually achieve the desired results. A really good way to cut through this hype and misinformation is to dig into the details just a little to understand how software actually works.
In that spirit I created the document Comprehensive Introduction to Intelligent Software Agents for Professional Accountants (DRAFT). Have a look at the document; if you have any feedback or questions please let me know.
Here is an excellent NPR story related to a different domain (i.e not financial reporting) that can give you an idea of the sort of hype and misinformation you can run into. That domain is "autonomous vehicles". Even the headline is misleading, Uber to Roll Out Self-Driving Cars in Pittsburgh. Listen to the story.
Notice first that the "self-driving" cars are really not autonomous. Uber will have a human in each car. What's the point? Are they saving anything by investing in creating the driverless car only to keep a human in the passenger seat as a safety net if something goes wrong? Sure they are. They are experimenting and trying to perfect the system. If they can make that work, Uber can save a lot of money in driver salaries.
Second, notice how you have three different predictions as to when this will actually work: couple of years, 10 years, 60 years. Who is right? Well, that depends how you define "autonomous vehicle".
On the other hand, go test drive a Tesla. Use the driver assist feature. Notice how useful that feature is. But then again, if you don't employ the feature properly, bad things happen. The point here is that setting the correct expectation, or goal, as to what software can do for you is important.
Finally, information technology people like to push things like "machine-learning" and "deep learning" and the power of "neural networks". But these people don't understand how domains such as financial reporting actually work. Not only can these sorts of approaches very expensive, you have to be careful about the results that you actually get. I repeat this statement from the blog post above, note the statement "high tolerance for error":
What Applications Should Neural Networks Be Used For?
Neural networks are universal approximators, and they work best if the system you are using them to model has a high tolerance to error. One would therefore not be advised to use a neural network to balance one's cheque book! However they work very well for:
•capturing associations or discovering regularities within a set of patterns;
•where the volume, number of variables or diversity of the data is very great;
•the relationships between variables are vaguely understood; or,
•the relationships are difficult to describe adequately with conventional approaches.
Be very careful when the snake oil salesman knocks at the door. The best defence is being an informed buyer. Reading that introduction to intelligent software agents can help make you an informed buyer, rather than an ignorant victim of hype and misinformation.
Now, this is not to say that these technologies don't have applicability to financial reporting. That is not what I am saying. They do. In terms of analyzing a financial report, they can be very helpful. But when creating a financial report, they are not necessary.
If you really want to understand this stuff, please read the following documents in addition to the one above:
- Conceptual Overview of an XBRL-based Structured Digital Financial Report: this document provides the vision.
- Knowledge Engineering Basics for Accounting Professionals: this helps you understand important moving pieces and the capabilities of computers; this is an outline at the moment but will be turned into a narrative (check back).
- Comprehensive Introduction to Business Rules for Professional Accountants: this points out the important role business rules play in making software perform work.
- Financial Report Semantics and Dynamics Theory: this helps you understand the conceptual model of a digital financial report.
- How XBRL Works: this video helps you see, in detail, why structured information makes all this work.
Over estimating the capabilities software can deliver is a mistake. But it is also a mistake to under estimate the work software applications will perform. As structured information is use more and more, understanding all of this will be of increasing importance.