Last updated: 29 Apr 2024 09:00 Posted in: AIA
AlphaPlus Assessment Managers, Gemma O’Brien and Priya Dutta explain the evolution of generative AI, and examines its implications for the accountancy profession.
Generative Artificial Intelligence (AI) has revolutionised various sectors, including finance and accounting. As the accountancy profession navigates through the era of digital transformation, the integration of generative AI technologies has become increasingly prominent.
This article explores what generative AI actually is and its relevance in assessing competence within the accountancy profession. We will examine its pros and cons, and discuss the international implications for the profession.
So what do we mean when we talk about AI?
In short, when we refer to AI, we are actually referring to the practice of getting machines to mimic human intelligence to perform tasks. Machine learning is a type of artificial intelligence. Through machine learning, practitioners develop artificial intelligence by using models that can ‘learn’ from data patterns without human direction.
The unmanageably huge volume and complexity of data that is now being generated has increased the potential of machine learning, as well as the need for it.
A key characteristic of machine learning is the concept of self-learning. This refers to the application of statistical modelling to detect patterns and improve performance based on data and empirical information – crucially, all without direct programming commands. The machine is formulating decisions based on experience and mimicking the process of human decision making to generate outputs.
One can think of building a machine learning model as similar to training a guide dog. Through specialised training, guide dogs learn how to respond in different situations. After the dog has been properly trained, the trainer is no longer required. The guide dog can apply its training to make decisions in unsupervised situations. Similarly, machine learning models can be trained to form decisions based on experience.
Until recently, machine learning was largely limited to predictive models, and was used to observe and classify patterns in content. A classic example of machine learning is when a program was shown a series of images of different breeds of dog. The program would identify patterns among the images, and then scrutinise random images for ones that would match the pattern – allowing it to identify the breed of dog for itself. It was largely (though not entirely) successful.
Generative AI was a breakthrough. Rather than simply perceiving and classifying images, machine learning is now able to create an image or text description of any specified item on demand.
Generative AI and machine learning
Generative AI systems fall under the broad category of machine learning. It is an umbrella term that describes algorithms that can be used to generate new content, including audio, code, images, text, simulations and videos.
A large language model works by receiving an input, encoding it and then decoding it to produce an output prediction. However, before a large language model can receive text input and generate an output prediction, it requires training, so that it can fulfil general functions and fine-tuning, which enables it to perform specific tasks.
Large language models are pre-trained using large textual datasets from sites like Wikipedia. Such datasets consist of trillions of words and their quality will affect the language model’s performance. At this stage, the large language model engages in unsupervised learning, meaning that it processes the datasets fed to it without specific instructions.
During this process, the large language model’s algorithm can learn the meaning of words and of the relationship between words. It also learns to distinguish words based on context. For example, it would learn to understand the different meanings of homonyms (words with the same spelling or pronunciation but different meanings) from their contexts – for example, ‘bat’ (an animal) and ‘bat’ (an item of sporting equipment).
Large language models can be used for a multitude of purposes, some of which are already ubiquitous. Take Google, for example: whenever you use the search feature, you are relying on a large language model to produce information in response to a query. The Google search engine is able to retrieve information, then summarise and communicate the answer.
Large language models can also be used for text generation. These are behind generative AI like ChatGPT and can generate text based on typed inputs from a user. They can produce an example of text when prompted. Large language models are always improving because they grow when more data and parameters are added. In other words, the more a large language model learns, the better it gets.
It is clear that the potential benefits are transformative and applicable across all of society. But what are the implication for the accountancy profession?
The challenges of AI
Large language models might give us the impression that they understand meaning and can respond to human inputs accurately. However, they remain a technological tool and, as such, face a variety of challenges.
Hallucinations have been talked about in depth in the media, with consequent doom-mongering. Simply put, a hallucination is when a large language model produces an output that is false, or that does not match the user’s intent – for example, claiming that it is human or that it experiences emotions. Large language models predict the next syntactically correct word or phrase, but they can’t wholly interpret human meaning. The result can sometimes be what is referred to as a ‘hallucination’.
An understanding of the processes by which an output is achieved helps to remove the fear from media reporting. Large language models are not gaining consciousness – they simply gathered the wrong data from which to produce an output.
Large language models do, however, present important security risks when not managed properly. They can leak private information, be used in phishing scams and produce spam. Users with malicious intent can reprogram AI to their ideologies or biases and contribute to the spread of misinformation. The repercussions can be far‑reaching, which is why public education about AI and its uses is crucial.
The data used to train language models will affect the outputs that a given model produces. As such, if the data represents a single demographic, or lacks diversity, the outputs produced by the large language model will also lack diversity.
Large language models are trained on trillions of datasets – some of which might not have been obtained consensually. When gathering data from the internet (known as ‘scraping’), large language models have been known to ignore copyright licenses, plagiarise written content and repurpose proprietary content without getting permission from the original owners or artists. Large language models have already run into lawsuits for violating intellectual property.
AI and the accountancy profession
In the realm of accountancy, competence assessment is crucial for maintaining professional standards and ensuring the accuracy and integrity of financial reporting. Generative AI has the potential to transform this process by automating tasks such as data analysis, report generation and risk assessment. For example, AI-powered tools can analyse financial data with greater speed and accuracy, identify anomalies or discrepancies, and generate comprehensive reports for stakeholders.
What does this mean for current accountancy students? Much as productivity tools in the early 1990s changed our approach to things like database management, generative AI will require a shift in mindset – and also a thorough re‑evaluation of curriculum content – to be assessed in order to determine competence. The knock-on effect on teaching approaches may mean that we have to rethink what competencies we are actually looking for in the accountants of tomorrow.
What are the pros of using AI in accountancy?
But what are the cons of using AI in accountancy?
International implications for the accountancy profession
The adoption of generative AI technologies in the accountancy profession has significant international implications, influencing regulatory frameworks, professional standards and workforce dynamics across different countries. International accounting bodies such as AIA will play a crucial role in guiding the integration of AI in accounting practices and ensuring global consistency in competence assessment.
Regulatory alignment: Countries will need to harmonise regulatory frameworks to address the ethical, legal, and technical challenges associated with the use of generative AI in accounting. This includes establishing guidelines for data protection, algorithmic transparency and audit trail documentation.
Professional standards: International accounting standards organisations such as the AIA must update existing guidelines to accommodate the use of AI in financial reporting, auditing and regulatory compliance. This includes incorporating AI-specific considerations into professional development programmes and certification requirements for accountants.
Cross-border collaboration: The globalisation of accounting services requires collaboration among stakeholders from different jurisdictions to develop common standards, share best practices and address emerging issues related to generative AI adoption.
Workforce development: The widespread adoption of generative AI in the accountancy profession necessitates investment in workforce development initiatives to equip accountants with the necessary skills and competencies to leverage AI technologies effectively. This includes training programs on data analytics, machine learning and AI ethics.
Conclusion
Generative AI has the potential to cause seismic shifts in the accountancy profession, offering opportunities to enhance efficiency, accuracy and innovation in financial management and reporting processes. However, its adoption also poses challenges related to ethics, job displacement and cybersecurity.
To navigate these complexities, international collaboration and proactive regulatory measures are essential to ensure the responsible and ethical use of AI in accounting practices.
By embracing generative AI technologies while upholding professional standards and ethical principles, the profession can adapt to the evolving digital landscape and drive sustainable growth. We are at an exciting time in the development of AI – and the profession must be proactive in shaping its own future.
Author Biography
As Assessment Managers at the educational consultancy AlphaPlus, Gemma and Priya provide bespoke services to clients, from assessment design, development of assessment instruments and management of large research projects, to data collection, analysis and reporting. They work with clients across all sectors to deliver data-driven improvement to learning and assessment programmes.
"By embracing generative AI technologies while upholding professional standards and ethical principles, the accounting profession can adapt to the evolving digital landscape and drive sustainable growth. It is an exciting time in the development of AI and we must be proactive in shaping its own future"
Gemma O’Brien and Priya Dutta, AlphaPlus Assessment Managers