Software can apply artificial intelligence to help us use our human intelligence more effectively.
Today, what the software companies are promising is automated intelligence, that is a computerised process of applying machine learning to the data that they can now conceptualise how to process and the data they can technically (and conceptually) work with.
Machine learning, being the computer capturing the changes you might otherwise make and observing the anomalies that you would normally observe.
Remember that computers process electronically 1,000’s of times faster than any biological (read human) process so it will be quicker and better to use the computer.
The Future and Maybe Even Today
Where automated intelligence should take us is using our data more extensively in conjunction with the data available to us through aggregation and the internet: Big Data. In effect next generation computer processing to analyse data and information effectively and efficiently and certainly more rapidly than we are able to perform as humans alone.
We already are seeing examples of the computer performing tasks we would do if we had the time and budget. We are seeing an emergence of “Assurance” dashboards. My concept of “Assurance Dashboards” is more than the directory of clients and where their bank feeds or purchase invoice in-tray is up to.
The first of these “Assurance Dashboards” was launched by MYOB many years ago in the form of the Company Data Auditor and its associated checking and reporting.
This form of automated review uses a process of checking what we would otherwise do manually: reconcile the control accounts, check for future dates, changes that seem wrong, Coding to accounts where the GST is different to normal or checking that GST is 1/11 and reporting to us.
Another “Automated Intelligence” service from MYOB is the ABN checking. When you enter the ABN in the supplier card it uses Big Data concepts to go and look up your data, the entered ABN, and compares that to the governments data, the ABR, and provides an integrity check and GST registration status.
These are examples of the early steps towards automated intelligence. (Artificial intelligence, I believe, is a whole different extended development).
Xero have recently brought us another development in the assurance dashboard concept
Who did what on the file when? Did someone work in a part of the file that they shouldn’t have? Did any bank accounts get changed? Did any bank transactions get deleted etc.
Third party products are also entering the solution space with fraud detection services: Has your “Cloud” based software been accessed from somewhere strange? Has a new user been added? Have your bank details been changed?
- These are what I would call first generation machine learning techniques.
- They are the things that we would check if we had time.
- They are (some of) the processes that we would follow and audit and review to provide 100% certainty of security and accuracy.
- They are good and I am glad they are provided to us.
But it isn’t enough.
The Promise That I Don’t Yet See
Many claim that with access to big data, software will be able to aggregate multiple businesses information and provide meaningful benchmarks. Until we have some means of comparing apples with apples I do not see it.
While we still have humans involved in any form of running a business, allocating names, allocating expenses to parts of the chart of accounts we are going to have the same type of transaction in different spots in different businesses. We might get “intelligence” that the aggregating computer will see that Company A put Officeworks bills to Stationery and Company B put Officeworks to Office Expenses and when it is doing the aggregation it will move all Officeworks to the same place.
Hopefully you see the dilemma. Maybe it shouldn’t all be in the same place. An Officeworks bill to Company A might be just for stationery but for the transaction in Company B maybe it was the computer that should have gone to the asset account for Office Equipment (so it was in the wrong place, let alone the amounts that should have gone to owner's drawings).
But, what if the software does compare your plumber client to the data it has of other plumbers? What if it compares to other plumbers that are in the same geographical area?
What if you are able to report to your client that, on average, your client is charging each client $265.00 but every other plumber in the area is charging an average invoice of $315.00?
What if it was able to tell you that despite charging that increased amount on average they are also issuing 100 invoices a week compared to your client only sending 75 invoices?
All of a sudden we have useful information easily generated based on big data, if the computer using automated intelligence has enough ability and direction and process to use big data in an appropriate form to provide useful information.
This type of aggregation of data sounds great but also opens up cans of worms about the integrity, privacy and use of such information. A topic for another time.
More Use of Big Data
What if the software knows that your client is a plumber and observes that your client often buys from Reece Plumbing Supplies? What if the software watches the Reece Plumbing website or online store and notices a special price offered on an item your client purchases regularly? Wouldn’t it be great if your software sent you an alert that reported this opportunity?
Your client, Mr Plumber, buys 25 Gold Tap Type B from Reece Plumbing each week for an average cost of $35 each. Reece Plumbing are currently selling that tap for $25 each.
The Artificial / Automated Intelligence I Want……Soon
Strategy / Wish List #1 From Software - Automated Intelligence #1
Use and Check my Own Data
ICB seek a more complete set of data interrogation techniques to be performed by the software.
We seek that the software provide us with a report of anomalies:
- Postings that don’t feel right (amount, coding, timing, frequency)
- Supplier payments to a strange expense account
- GST treatments that are incorrect
- GST claimed from suppliers that aren’t registered
- Postings by anybody that aren’t the same as the norm.
We, the Certified Bookkeeper, become the expert user of that dashboard or report of “anomalies” and process and adjust. One could say we then apply “intelligence” following the detection of the anomaly.
Strategy / Wish List #2 From Software - Automated Intelligence #2
Report to us Actions That Might Not be Right.
We seek software that provides us with warnings about behaviour that appears wrong before it has an impact, for example:
- Invoices going out with different banking details
- Invoices going out with abnormal transactions
- Payments being made to new / unusual suppliers
- Duplicate bank account numbers: suppliers or employees
- Changed bank accounts
- Different users changing system information
Similar to #1, we, the Certified Bookkeepers, become the authorised receiver of this report and due to our intelligence we are then the expert user of this information.
Strategy / Wish List #3 From Software - Automated intelligence #3
We seek intelligence tools that obtain authorisation of changes and payments. We seek intelligence tools so that a manager may be able to observe (even have to authorise) a change to anything in the system.
This vision gives us, as bookkeepers, the tools to help business make money and not lose money. We as the software and Business System Integration Experts set up the software with the audit, verification parameters and receive the anomaly report for our action or follow up.
And there are more items to the wish list. As the technology steps forward and the software companies learn more about what we actually do to make their software work properly they can help us be more effective. They can apply artificial intelligence to help us use our human intelligence more effectively
Imagine the possibilities when the 2.4 million businesses in Australia are using Certified Bookkeepers who have embraced the best “intelligence” tools available to us from the software and we are the ones enabling the best business process, certainty, security.
If the Royal Mint can have its employees steal $1m due to unauthorised payments by an ex-employee, then maybe our clients need us monitoring and assisting their automated intelligence tools.
Artificial Intelligence that we can’t dream of because we can’t conceptualise what it may be.
A long term friend and great influence in our space recently reminded me that today’s technology can do things that the technology from even a few years ago could not. The Artificial / Automated Intelligence technical community keeps espousing that technology is developing at an exponential rate i.e., double the double the double rate in half the time, only to be outpaced in the next iteration.
What Might the Future of Real “Artificial” Intelligence Look Like?
I hope it isn’t as mundane as machine learning, i.e., it watched what you did last time and suggests you do it the same way this time, or even it reviews what you did most times and tells you the instances that are not the same.
I want more than big data doing some comparison of what we did compared to what it has matched us to from everywhere else and reporting its version of what we might have done right or wrong.
I want it to make the changes, report to us what it observed and changed and we are the experts reviewing the machine.
Is this death to bookkeeping as we know it? No! It is empowering and enabling the bookkeeper to do provide certainty and verification and explanation work that we would do but we can’t afford to. Lack of time and lack of clients being able to fund us using today’s technology.
Tomorrow, the technology will enable us to spend the same time doing everything we would like to be doing.
Updated: 11 February 2020