Digital Financial Reporting

The AI Ladder, by Rob Thomas and published by O'Reilly Media, is by far the best resource that I have run across related to getting your head around artificial intelligence.  Here is a summary of why AI projects fail:

• Lack of understanding. 81% of business leaders to not understand AI.
• Bad data. Not having a handle on your data is completely paralyzing.  Your AI is only going to be as good as your data.
• Lack of the right skills. The lack of the right skills on part of both business professionals and information technology professionals is problematic.
• Trust. Trusting the recommendations made by your artificial intelligence software is a must.  AI should not be a black box, business professionals need justification mechanisms that support conclusions.
• Culture. The Technology Fallacy points out that digital transformation involves changes to organizational dynamics and how work gets done.  AI will enable entirely new business models which were impossible in the past.

AI is hard work. Getting AI right involves the right tools, the right skills, and the right mindset.  Artificial Intelligence and Knowlege Engineering in a Nutshell can help you navigate AI and help you avoid the snake oil salesment.  If you understand AI, then you will likely understand the connection between AI and global standards like XBRL.

The Forbes article Demystifying Artificial Intelligence points out that AI is on the minds of 96% of business executives of leading corporations.  Another Forbes article, Deep Learning Must Move Beyond Cheap Parlor Tricks, warns that one should avoid creating a fragile house of cards. A third Forbes article, This Week in AI States: Up to 50% Failure Rate in 25% of Enterprises Deploying AI, points out that there is a 50% failure rate in the approximately 25% of organizations implementing AI globally do to lack of skilled staff and unrealistic expectations.

Yet, a PWC study in 2017 points out that:

“Global GDP will be 14% higher in 2030 as a result of AI – the equivalent of an additional $15.7 trillion. This makes it the biggest commercial opportunity in today’s fast changing economy”

AI is hard work as The AI Ladder points out.  It requires proper tools, proper methods, and the right mindset.

Here is, in my view, an example of AI done right.  A software engineer and I created an extensive working proof of concept of what amounts to an expert system for creating financial reports.  If you really want to understand the application, watch the set of videos in this play list.

How did we do it?  First, we did not make the "rush to detail" mistake that most people make.  We created a solid foundation then we built on top of that foundation.  I created a set of principles. With some help I created a theory. We created a model.  We put together a framework. We created a method.  We did the necessary testing.

All this resulted in our working proof of concept. This document, Guide to Building an Expert System for Creating Financial Reports, helps you to understand important details.  While we essentially created a good old fashion expert system, we used the right tool for the job.  All this very high-quality metadata will serve as the necessary training data to enable additional functionality using machine learning.

Don't misinterpret what you see.  As the Innovator's Dilema points out, "A disruptive product appears as if it's doing everything wrong. Large companies with sophisticated and demanding clients cannot adopt such a technology."

This could be the future of financial reporting. Yes, it has to work well.  It can.

A disruption is when new products and services create a new market and significantly weaken, transform or destroy existing product categories, markets or industries.

Maybe we will turn this into a product and get a piece of that $15.7 trillion.

If you are still confused about AI, read this.

###################################

Artificial Intelligence for the Real World, Harvard Business Review

Forbes article, Demystifying Artificial Intelligence, points out that AI is on the minds of business executives. While 96% of businesses are doing something with AI, most still seem confused about how to put it to use:

O’Reilly Media Founder and CEO Tim O’Reilly observes,

“Everyone is talking about ‘AI’ these days, but most companies have no real idea of how to put it to use in their own business”.

Everyone seems to want to rush to the "sexy" part and don't realize that 80% of their effort will be related to getting their data properly organized and sorted out. Not getting their data sorted out is why most AI projects fail.

O'Reilly Media publishes The AI Ladder to help organizations to help sort out their strategy and tactics. Here is part of the conclusion of that document:

Finally, AI is not magic. It’s hard work. It requires the proper tools, methodologies, and mindset, to overcome the gaps that companies are facing (data, skills, and trust) to truly embrace an AI practice and put it to work across your organization. AI is the biggest opportunity of our time, and yet there’s still a certain fear in the market that AI is going to replace jobs. However, the reality is this: AI is not going to replace managers. Rather, the managers who use AI will replace the managers who do not.

Time to up your digital maturity. A good place to start is Artificial Intelligence and Knowledge Engineering in a Nutshell.

Also, here is how I got my data in order.  And this is what I can do as a result.

The document, Auditing XBRL-based Financial Reports, proves that an audit strategy can be created for XBRL-based financial reports.

Here are the cliff notes:

A financial report is an allowed interpretation of an expression of the financial position and financial performance of an economic entity per some set of statutory and regulatory rules.  Here-to-for, that expression has been in a form that is only readable by humans. However, XBRL and other machine-readable formats change that, making those expressions readable by both humans and by machine-based processes.

Single-entry accounting is how ‘everyone’ would do accounting. In fact, that is how accounting was done before double-entry accounting was invented. Double-entry accounting was the invention of medieval merchants and was first documented by the Italian mathematician and Franciscan Friar Luca Pacioli.

Double-entry accounting adds an additional important property to the accounting system, that of a clear strategy to identify errors and to remove the errors from the system. Even better, double-entry accounting has a side effect of clearly firewalling errors as either accident or fraud. This then leads to an audit strategy.  Double-entry accounting is how professional accountants do accounting.

An XBRL-based financial report is not only a machine-readable format; it also is a machine-readable logical system and has the potential to be a well-defined and fully expressed logical system.  A well-defined logical system, when fully expressed, will be properly functioning and demonstrably consistent, valid, sound, and complete.  These properties can be leveraged to offer a systematic audit strategy for XBRL-based financial reports.

Essentially, an XBRL-based financial report is a set of declarative statements provided in global standard XBRL format.  Logic programming software applications such as Prolog, Datalog, Clips, and Answer Set Programming can provide feedback as to whether these statements are consistent, valid, sound, complete and otherwise properly functioning.  Even XBRL processors and XBRL formula processors can effectively prove that XBRL-based financial reports are properly functioning to a large degree.

“It must be remembered that there is nothing more difficult to plan, more doubtful of success, nor more dangerous to manage than a new system. For the initiator has the enmity of all who would profit by the preservation of the old institution and merely lukewarm defenders in those who gain by the new ones.”  Niccolò Machiavelli

################################

Assurance on XBRL Instance Document: A Conceptual Framework of Assertions

Digital Auditing: Modernizing Government Financial Statement Audit Approach

xAudit: Auditing Representations in XBRL Based Documents

My testing of XBRL-based digital financial reports has lead me to the point where it is pretty obvious that there would be great benefits to business if there were a global standard machine-readable logic framework for business.

This is what I mean.  There are many domains that use logic.  Philosophers invented logic and tend to use propositional logic, electrical engineers use logic gates to design electronic circuit, designers of complex systems like nuclear power plants use Z-notation, computer scientists use predicate logic which is an extension of propositional logic, etc.

But why does business not really have a logic framework?  Perhaps it is not a fair or even appropriate comparison.  But it might be.  Business professionals use electronic spreadsheets for solving all sorts of problems.  The electronic spreadsheet is more of a presentation-oriented (sheet, column, row) modeling tool as contrast to a logic modeling tool.

Let us suppose that business professionals needed some sort of logic modeling tool.  What terminology might that tool use?  There is no real standard terminology for working with logic that I have seen.  However, the terminology can be created.  If I were to create standard terminology, the terms would be: theory, model, structure, term, association, assertion, and fact.  These are all statements. All these statements need to be complete, valid, consistent, and sound.  The system needs to be fully expressed.  Graphically, I created the following depiction: (here are the definitions of the terms used)

Now, this graphic communicates an idea, not a logical thing that can be measured. I would like to come up with a better graphic.

Think of this as a complete set of statements that have to be consistent and everything has to be valid and if you achieve this then the system is sound.  If all those four constraints are met and the system is fully expressed, then the system is functioning properly.  Two different people can look at the same set of statements and reach the same conclusion.

What is more, you can prove that the system is functioning properly, the full set of statements is like a “parity check” or a “check sum”.

I see this as having similarities to the double-entry accounting system.

Single-entry accounting is how 'everyone' would do accounting. In fact, that is how accounting was done before double-entry accounting was invented. Double-entry accounting adds an additional important property to the accounting system, that of a clear strategy to identify errors and to remove the errors from the system. Even better, double-entry accounting has a side effect of clearly firewalling errors as either accident or fraud. This then leads to an audit strategy.  Double-entry accounting is how professional accountants do accounting. Double-entry accounting was the invention of medieval merchants and was first documented by the Italian mathematician and Franciscan Friar Luca Pacioli.

Double-entry accounting is one of the greatest discoveries of commerce and its significance is difficult to overstate.  Which came first, double-entry accounting or the enterprise?  Was it double-entry accounting and what it offered that enable the large enterprise to exist; or did the large enterprise create the need for double-entry accounting?

The key idea here is the “clear strategy to identify errors and remove errors from the system”.

This is what I see some sort of global standard machine-readable logic framework for business doing.  If the logic of the system has no boundaries what-so-ever; then who is to say if something is right or wrong?  If you cannot determine if something is right or wrong; then how could it possibly be of any use?

Here is all my current brainstorming that I have done on this topic. More is still to come.

####################

Well-founded Semantics