AI – Little I and a pain in the A! – part 1

AI is the new buzz word. All of a sudden you and I don’t have to be intelligent to appear intelligent. We use GPS (?) to navigate to an intelligent Chat (or something like that!) and accept the outcome as gospel. Where is this going?

Introduction

AI is here to stay, whether we like it or not. The challenge is how do we embrace this technology without abdicating our free will and ability to apply our minds?

This article deals with how AI is or can be used by taxpayers to manage their tax affairs. In a follow-up article we shall deal with how SARS applies AI in practice and the challenges already experienced in this space.

AI versus data analytics

AI is often confused with data analytics used in organisations to identify trends and “out of range” transactions for further investigation.

Data analytics have been used for many years by organisations especially in the field of large data sets, but its use is not uncommon in smaller data sets.

The purpose of data analytics is to identify transactions that fall outside predetermined parameters for investigation by intelligent (?) human beings. It is a tool for use by human beings.

Data analytics on its own is not AI as no action follows without the intervention of a human being. The technology merely identifies exceptions to be further investigated by intelligent human beings. The technology does not create new data or outcomes; it merely extracts data based on parameters determined by human beings.

And if the human beings do not set the parameters appropriately (unintelligently?) for the outcomes required, the information that will be provided will be as unintelligent as the creator thereof.

In simple terms; garbage-in-garbage-out!

AI versus automation

In large organisations it is necessary for data to be distributed to the appropriate users thereof to ensure efficiency of data processing and usage.

Historically (i.e. pre-AI) this has been referred to as project management of data.

The issues raised above with regard to data analytics, equally apply to data automation processes.

Data automation is the process of allocating data and tasks to predetermined recipients based on the outcome of (normally) data analytics, and allocating and monitoring the completion of tasks on a cradle-to-grave basis.

The automation processes are based on the parameters set by human beings in an environment that is closely controlled to ensure that parameters cannot be changed without human intervention and approval processes. The integrity of the automated processes is dependent on human intervention and control.

While we are sometimes made to believe that computer programs make decisions on its own, it does not (presently). It merely executes what it is being told to do.

So where does AI fit in …

AI

Presently AI is in its infancy, but be mindful not to kick a puppy when it is small. It will grow up and you will rue the day that you ignored the puppy.

The Achilles’ heel of AI is that it is dependent on access to a very large data set to enable it to consider the widest possible variety of options and then apply/recommend the most desirable option.

As data sets and industry knowledge grow, AI will in future be “empowered” with more and more decision making ability. The question is whether that decision-making ability will be an executive ability where execution of the decision/recommendation is done without human intervention, or merely making options available for consideration by humans?

Only the future will tell, but the world is certainly preparing for a decision-making era by computers (are we talking about the singularity?).

What is SARS doing?

Amendments to tax legislation have already been proposed to prescribe the future format of documentation. Standardised documentation will be an essential element of using AI effectively to interrogate data of third parties in the future.

The future is closer than we think …

Conclusion

In this article we provided a birds-eye view of what is currently (correctly or incorrectly) viewed as AI or the forerunner of AI and what the future may (will) look like.

In the next article we shall deal with specific taxpayer data interrogation already being done by SARS and the challenges experienced in practice.

Watch this space …