The AI Technology Landscape: What Auditors andAccountants Need to Know Today
Artificial Intelligence (AI) has rapidly become a buzzword,especially with the rise of technologies like ChatGPT. This AI tool has takenthe world by storm, introducing the public to a new type of AI that is bothpowerful and accessible. But what does this mean for professionals in fieldslike auditing and accounting? Let's explore the AI technology landscape andunderstand what auditors and accountants need to know today.
What is AI?
AI, or Artificial Intelligence, can be understood in two ways:by its objective and by its means. The objective of AI is to create machinesthat can perform tasks that typically require human intelligence. These tasksinclude learning, reasoning, problem-solving, understanding language, andperceiving the environment. The means to achieve these objectives involve usingalgorithms, data, and computing power to mimic human cognitive functions.
Specialized AI vs. Generalist AI
Before the advent of ChatGPT and similar technologies, most AIsystems were specialized. Specialized AI, also known as narrow AI, is designedto perform specific tasks. Examples include AI systems used for fraud detectionin banking or software that automates data entry in accounting. These systemsare excellent at performing their designated tasks but lack the flexibility tohandle tasks outside their specific domain.
In contrast, the emergence of ChatGPT represents a shift towardsgeneralist AI. Generalist AI, also known as general AI, can perform a widevariety of tasks across different domains. ChatGPT exemplifies this capabilityby understanding a wide variety of text, drafting marketing communications orpoems, solving simple math problems, and coding in various programminglanguages.
AI Performance and Generality Levels
This table, from the research paper "Levels of AGI:Operationalizing Progress on the Path to AGI" by the Google Research team,categorizes AI systems based on their performance and generality. It highlightsthe progression from narrow AI, which performs specific tasks, to the potentialfuture achievement of Artificial General Intelligence (AGI) and beyond.
The Complexity and Limitations of Generalist AI
What can be troubling is that ChatGPT, while capable ofperforming some tasks at an expert level—such as answering bar exam questionsor medical queries—is not a substitute for doctors or lawyers. In other tasks,like complex reasoning in business analysis, it is still at an emerging level.People sometimes imagine that feeding large quantities of data into this blackbox will yield comprehensive insights. However, this is not the case, and itwill not be the case until we achieve at least Level 3 Expert AI.
The explosion of AI technologies like ChatGPT has introduceda new era of generalist AI, which holds significant potential for the fields ofauditing and accounting. While it is essential to recognize the currentlimitations—such as its emerging level of complex reasoning in businessanalysis— the benefits are already concrete and transformative. For instance,an AI like ChatGPT can draft a letter in under two seconds and adapt to anywriting style or generate a program in 20 seconds based on a simple sentence.This capability is not yet AGI, but it is already changing the world. Byunderstanding the landscape of AI and utilizing the available tools,professionals can enhance their efficiency, accuracy, and strategicdecision-making capabilities. As AI continues to evolve, staying informed andadaptable will be key to leveraging its full potential.
