Generative artificial intelligence is making inroads into the accounting industry, promising to greatly increase efficiency and productivity while offering real-time, deep insights that help improve performance. As firms deal with labor shortages and expand their services amid elevated client expectations, they are avidly exploring AI’s possibilities.
AI doesn’t come without caveats, particularly for accounting firms that work with highly sensitive personal and financial information of their clients. Although Gen AI’s potential benefits are considerable, firms should proceed cautiously and understand its impact on business.
For all of its potential, AI may not immediately solve all of the industry’s challenges. As the initial excitement subsides, it’s critical that IT teams ensure that any AI initiatives align with the objectives of their stakeholders — including the firm itself, clients and regulatory bodies.
The steps to implementing responsible AI
Building a responsible AI strategy starts with a clear understanding of the specific problems or opportunities the firm aims to address with AI, coupled with a commitment to educating leadership and employees on what AI can and cannot achieve. This foundation ensures AI is implemented and used thoughtfully, with resources aligned to deliver maximum impact.
Accounting firms also need a strong data and analytics strategy to ensure their data is well-structured before implementing AI. Structured data is the backbone of responsible AI, enabling faster, more accurate insights and transforming data into a powerful decision-making tool. Without it, AI risks stumbling on inconsistencies and poor-quality data, leading to misguided outcomes and wasted resources. In short, well-structured data unlocks AI’s full potential.
Once these fundamentals are in place, firms can assess their current maturity and readiness for AI implementation. Using a Capability Maturity Model specific to knowledge work automation provides a structured framework for this purpose, helping firms evaluate their competencies across five key considerations when adopting new technologies:
- Information strategy;
- Governance/resourcing;
- Technology/IT infrastructure;
- Level of automation; and.
- End-user capabilities.
By using the model, firms can identify their capability levels in each category, ranging from beginner to advanced. For example, in the area of information strategy, a firm with minimal IT and business alignment may be considered a beginner, whereas one with integrated alignment across IT, business and executive functions may be classified as more advanced.
Responsible AI will prioritize safety, transparency and trustworthiness. Firms need to strike a delicate balance between innovation and security, which first requires a thorough evaluation of data connectivity, curation, and confidentiality.
To properly incorporate responsible AI, there are five essential areas accounting firms should consider:
Protecting client privacy
Because safeguarding client information is the foundation of building trust with clients, privacy protections must be a top priority when accounting firms add solutions to their tech stack or develop new tools.
Firms can ensure they meet client expectations of confidentiality by practicing techniques like data minimization, ensuring firms handle the least amount of information required for a specific purpose. That can reduce the risk of data breaches, privacy violations and misuse.
Firms should also never share client information on public platforms like ChatGPT, which are vulnerable to cybersecurity threats that the firm has no control over.
Guarding against bias
An AI model trains by analyzing enormous volumes of data and applying what it learns to perform its tasks. Data scientists and developers need to be wary of the information they use to train and create AI algorithms. If biases exist in the training data, those biases will be replicated in the AI model’s work and generate unrelated or incorrect information.
For example, a model may be trained to scrutinize a particular account that has a history of misstatements while overlooking new accounts in the current year. Or it may apply a biased risk profile to particular groups of clients based on historical data rather than client-specific information. IT teams should scrutinize inputs and outputs regularly to detect biased results.
Promoting trust through transparency
AI’s performance should not be a mystery; the models used by accounting firms should be simple, auditable and explainable. Explainable AI methods and tools can show how AI arrives at its decisions, allowing humans to understand the outcomes or identify and address potential issues. Establishing this level of transparency will help foster and demonstrate trust and respect with customers, users, and stakeholders.
Enforcing accountability
Better transparency enables better accountability. A user or group of users — which can include developers, deployers and even end users — should be assigned to regularly monitor and audit the firm’s AI models. They should be able to explain the rationale behind the AI’s outputs and perform updates or make adjustments to correct issues or errors.
Redefining roles
The truth is that AI isn’t going to replace accountants, but it will redefine their roles. AI has the power to transform the way accountants work, freeing employees from mundane tasks to drive growth. Accountants need to grasp the power of pairing their expertise with AI and learn to work with it to improve performance and efficiency.
AI will need accountants to provide extensive monitoring and oversight. But by taking over a lot of routine tasks that accountants spend time on now, AI will allow them to focus on more complex high-level initiatives. In the process, AI will help alleviate the labor shortage and could improve firm retention.
Future-forward accounting firms can reap immense benefits from GenAI as they embark on their digital transformation journey. However, they need to ensure they protect privacy and security. Implementing AI within a capable knowledge work automation framework can, for example, help ensure that data remains confidential, stays within internal system boundaries and that employees have access only to the data they need.
Making sure AI models are trained on complete, bias-free data. Having accountants monitor AI’s outputs can maintain transparency and ensure efficient, effective use of the technology. AI is part of the path forward for the industry, but firms need to be sure they step carefully.