Next line are just thoughts and opinions about the current state of things. The profession is changing significantly. Large parts of tasks that human accountants used to do are already being done by machines, and more will follow. Categorizing, summarizing, searching, writing, counting, matching, reconciling, comparing – for these tasks, the technology is already there, and they are commoditized to the degree that very few jobs will remain in this area.

The next round of tasks to be affected in a near future in are those related to preparing calculations that need to be precise and reliable, more complex journals and accounting processes, the interpretation and application of accounting standards, drafting of financial statements and presentations. LLMs are navigating paragraphs of standards much faster than humans, churning out solutions based on all their training data. There are already solutions being developed for spreadsheets, and PPT decks.

So what is the role of the human? At this point, it is increasingly about the synthesis of various domains of knowledge, creativity in understanding individual circumstances and finding more efficient solutions, proactivity, and ability to identify opportunities and implement changes. It is about critical thinking and the ability to perform the remaining 30% of tasks for which AI is not there yet and only Artificial General Intelligence (“AGI”) will resolve them, and reviewing AI outputs. Being a genuine expert in your own discipline matters a lot, as it will be the edge cases and unique problems in accounting problems that will still require humans to resolve them – such as one-off, very specific transaction for which no large volume of past data exists. And equally important it is for accountants of today to understand they must become technologists, and know where to implement AI solutions, as well as where not to implement them, because they would not stand up to a cost-benefit appraisal. A lot will be automatized in the coming years, it will not be in a form of a one ‘bombastic’ change, but gradually.

Where is the future headed? No one truly knows. There is a more sceptic camp where they say we may hit the qualitative limits of AI, due to the finite availability of data and machine learning approaches that do not evolve qualitatively, and the progress will hit a plateau. The optimist camp will tell you the real revolution is yet to come. That current LLMs were trained on written data and the internet, and that soon, integration with other data sources from the physical world (visual, sound, touch, and so on), more compute power, synthetic data, and various new machine learning approaches being developed, experimentation with randomness, and the AI arms race between the US and China will all lead to AGI in the coming years.

We however know where we are now. We are in an inter-period between the PC spreadsheet accounting era of the 1980s to early 2020s, and potential era where the human cognition will be gradually phased off, similar to how the industrial revolution largely replaced physical labor. How long it lasts, nobody knows. This does not mean the volume of work will decrease. Technology may enable what is today annual reporting to be performed quarterly or even monthly, while also making possible far more granular disclosures and insights that would otherwise be too time-consuming to produce.