Common myths around AI

Understand the terms and acronyms before you get started

IP image

The world is abuzz with news on how artificial intelligence, also known as AI, is changing the landscape for many industries, and accounting is no exception.

However, the volume of terms, acronyms and technical words sometimes tends to discourage accountants from fully adopting these technologies, and in some instances deter them from even getting started. After all, we got into the profession through our love of numbers and helping clients, not to become technology experts.

Types of AI

Artificial intelligence is an umbrella term, meaning technology that can mimic human-like tasks. AI can imitate human-like thinking and decision-making due to its capabilities to learn, reason and problem-solve.

To dig a little deeper, AI is trained to use algorithms and identify patterns, which allows it to process large amounts of data.

Generative AI is another term you may have heard which is a specific type of artificial intelligence, designed to create ‘new content’ in the form of text, imagery or video.

Machine learning (ML) is often regarded as a refinement of AI techniques where the frameworks are designed to self-improve with time as new data is recorded. ML applications can also forecast future situations by making use of past events. Great when we are looking ahead for our clients.

You may have heard of ‘deep learning’, too, which is essentially a more complex version of ML.

A more task-orientated type of intelligence, as opposed to a thinking approach as above, is robotic process automation (RPA). Something we have seen in the industry for a while with the introduction of ‘rules’ in software. For example, it can allocate bank transactions to specific categories automatically.

One commonality across all types of AI is the need for training. This is important for the user to be able to get the most out of the system, but also training the system's knowledge base so it can draw down correct information and provide accurate responses.

Other terminology

Natural language processing (NLP) is the way we communicate with technology. The ease with which we can now do so, the same as we communicate with humans, is likely to be one of the main reasons for the rise in popularity. It means we do not have to be able to code in order to ask questions and get responses that feel as though the systems understand us.

It is also where the term large language model (LLP) comes from, systems being trained on huge data sets of human language. A great use case is giving a system, such as ChatGPT, a source of data such as a complex tax legislation document and asking it to explain in layman's terms.

Addressing the concerns

It’s easy to see why AI is poorly understood and how those using it might be inaccurate in which features they are coming into contact with.

In future articles we will have a close look at AI in relation to the work of an accountant, dispel some of the myths around it, discuss practical use cases and touch on how that will impact the profession in the fast-changing landscape.

Billie McLoughlin, 20:20 Innovation