By David Frigeri, partner and Chief AI Officer at EisnerAmper
Responsible artificial intelligence (AI) in the accounting sector focuses on strengthening your colleagues’ and clients’ trust in the technology. Building this trust directly affects the success of AI at your firm.
Integrating AI Responsibly
When adopting AI, there are a few facets on which to focus. First, embed the responsible use of AI in your culture. Train and enable your colleagues to understand the importance of maintaining data privacy, fact-checking AI results and understanding that AI is an efficient tool that can deliver great work.
The successful adoption of AI often does not just reduce risk, it also increases ROI. For practitioners to use AI, they must trust it. So, there is also a pragmatic business aspect to using AI responsibly, along with the ethical aspects. 
The responsible use of AI applies to all technology, including your client serving software vendors. For example, accounting firms should require all major software vendors to answer a rigorous questionnaire based on the NIST AI Risk Management Framework.
Ethical Considerations
Accounting firms face three ethical challenges when adopting AI: bias, fact-checking and traceability. AI has biases when it bases an outcome on previous data. For example, using AI to assist with tax returns can have biases based on the zip code of the individual’s return. This makes it imperative to look beyond the given answer and determine if other data sources interfere with the models’ response, such as census data or past data entries on related subject matters.
Before relaying any information to a client, fact check the results first. Rather than blindly trusting the AI information, trace where the response originated to best understand your given results. This helps determine your answer’s accuracy and credibility while strengthening your AI responsibility and knowledge. Although AI technology is quickly progressing in terms of sophistication and accuracy, it still is not without faults. A key part of responsible AI use is understanding the technical terms around traceability to understand how the model came to a certain conclusion, especially as the information pertains to clients or stakeholders.
Measuring AI Success
There are different ways of defining success in AI implementation. One is to look at ROI; the other is to measure the response’s accuracy and specificity. Accounting is the business of accuracy, so it is important that AI identifies discrepancies between two different fields. When building AI, require everything to have that citation associated with it so clients can verify the accuracy and relevancy themselves.
It is important to understand the duality of AI knowledge and ethicality. Align AI responses with client’s business goals and reality by utilizing a structured method (through guardrails) that monitors AI-generated responses to review accuracy and relevancy.
Inbound guardrails prevent users from putting sensitive information into any ChatGPT. Outbound guardrails require AI responses to always include a citation, so if they do not include one, then the model automatically reruns until it produces a citation. This client-centric approach allows you to understand your objectives, opportunities, and obstacles—helping to define tailored guardrails.
The Future of AI
To fully leverage data’s potential, firms should treat data as a first-class citizen, meaning they should invest in consolidating data physically or logically, to make sure it is professionally managed, high quality and complete.
Organizations have also started investing in their people. Everyone has different levels of AI comfort, making it important for firms to meet their colleagues where they are and build AI confidence. Firms should offer training sessions and learning opportunities and work with a third-party advisor if necessary. People will gain confidence by enhancing employee or stakeholder knowledge around AI, ultimately increasing their comfort. This will have a direct, positive impact on your ROI, making investing in people key to long-term success.
With AI, you will see automating or significantly augmenting the more menial tasks that professionals do on every client engagement. This will then open the door for practitioners to spend more time building client relationships, solving complex problems, developing new skills and investing in their personal and professional development. These skills will come to the forefront as AI opens the door for us to spend our days with clients directly and provide greater value at that point of contact.
Efficiency improves with the technological advancements of AI, putting more available time on your calendar every day of the week. The question then shifts from “How will I get everything done?” to “What am I going to do with that extra time?” This could mean using that time to create new, innovative services for clients to help develop their businesses and co-envision their long-term growth.
Accounting firms should help their clients proactively address AI considerations to set them up for success with AI. First, identify your firm’s key differentiators in clients’ minds, then build an AI strategy that fully capitalizes on these strengths. If clients value the breadth of your services, use AI to identify additional areas where you can assist them effectively. If they value your responsiveness, craft AI solutions that enhance your speed and availability even further. Whatever your AI strategy, deliver it responsibly.
