We were excited to read ISCA’s white paper (below link) on AI and its applications to accounting.
Its reiteration of the need for accountants and their professional scepticism mirrors our perspective. AGI is not here yet, and many educated voices say it may never come. In the meantime, what is called AI today can be an augment to human work, slowly shifting our professional lives toward further efficiency. While complex activities rightly demand human involvement, many of accounting’s low-level tasks lend themselves to automation. We believe that a cautious approach to automating critical work like accounting is the key to the smooth adoption of AI.
Backbone’s philosophy on AI is currently centred around rules-based systems; repetitive task automation based on strictly defined steps will liberate many hours from accountants’ workdays. These rule inputs must be thoroughly verified and validated by human accountants before being implemented into systems. The clarity and simplicity of these first forays into accounting automation allow for a quality control that is difficult to achieve with neural networks, whose volumes of data are so large that how they operate exactly effectively becomes a black box. As such, we believe our systems will be much easier to subject to legal and professional standards, and can serve as the basis for more complicated AI automation efforts. An LLM might form part of the UI in a future accounting automation suite, but it would be stitched onto very reliable primary automation tools. Backbone will continue into the future to identify areas where AI performance will outweigh its risks, and integrate them into reliable accounting automation systems that meet the expectations of clients and stakeholders.
