top of page

Humans, robots and parrots

At the end of last year, ChatGPT, the artificial intelligence (AI) tool that everyone is speaking about, took a UK accountancy exam. It didn’t pass, but it didn’t fail by much either. What does this mean for the future of our profession?

Although the AI tool showed a good understanding of the subject matter, it battled with nuances and industry-specific definitions, it had questionable referencing ability and it struggled to stay clear and on topic. For some, ChatGPT’s exam results are cause for alarm. For others, this just goes to show that the human brain is still superior. For a third group, ChatGPT is the most exciting thing since sliced bread. The truth, I suspect, is somewhere in the middle. Technology changes such as AI becoming more mainstream and accessible is undeniably going to change most, if not all, professions.

It is how humans use these tools that is going to be the source of real value creation. As ever, the garbage in, garbage out (GIGO) rule applies. How humans ask ChatGPT questions or give it instructions is going to directly impact the outcome. It has never been more true that parrot-fashion learning can only get you so far—usually just until the end of the exam you are cramming for. But a broader understanding of underlying principles and the logic of how things work means your knowledge can be applied beyond a specific exam or task.

In the case of ChatGPT, this means the difference between a quick, superficial query delivering an OK but probably flawed result, and a well-thought-out, context-driven query delivering a better result. (As long as you only need knowledge that predates 2021, which is, at the time of writing this, the limit of ChatGPT’s database.

As one of IDU’s developers says: ChatGPT can make a mediocre developer quite good, a good developer excellent, and an excellent developer a rock star. But it can’t make a mediocre developer a rockstar.

As leaders, this is a seismic shift in how we succession plan. We need to rethink what will mean to be an accountant and hire people that can adapt, learn and evolve their skills to keep pace. And as accountants, this reminds us how much and how fast the world is changing and, as I wrote about previously in last year’s April edition, that the skills we should be developing extend well beyond what it traditionally means to be an accountant. This is likely to include getting really good at how we train and instruct our AI-based co-workers because currently, commentators are describing them as inexperienced juniors with all the confidence of not knowing they’re wrong.

The rise of human-AI collaboration in accounting

The future of accounting lies in human-AI collaboration, not replacement. ChatGPT's performance in a UK accountancy exam revealed that while AI can process information quickly, it still struggles with nuances and industry-specific knowledge. Human accountants will shift from information processing to interpretation and decision-making, while AI tools handle repetitive tasks. Success in the future of accounting depends on human accountants' ability to effectively instruct and work with AI tools.

P.S. This sidebar was written by ChatGPT after it read the main article. What do you think?

As published in Accountancy SA May 2023


bottom of page