How Excel users can evaluate and mitigate the risks of generative AI

In my Copilot workshops tailored specifically for Excel users, particularly those in finance and accounting, two recurring concerns stand out: the accuracy of AI-generated outputs and the potential exposure of sensitive data.

These concerns, while important, are not entirely new. Similar risks have existed for years with traditional digital tools like search engines and cloud services. Yet generative AI seems to provoke greater apprehension.

Why is this, and how can Excel users effectively navigate these issues?

Excel users, especially those in finance and accounting, are known for their caution and emphasis on precision. They require results to be accurate and verifiable because the stakes—whether financial, reputational, or regulatory—are often very high. Generative AI, with its complex algorithms and seemingly human-like outputs, can feel intimidating, unpredictable, and opaque.

However, by drawing parallels to existing digital practices and incorporating robust digital literacy strategies, Excel professionals can confidently evaluate and manage these perceived risks.

Probabilistic Algorithms: Already Part of the Excel Workflow

Probabilistic algorithms might sound novel or complex, but in reality, Excel users frequently employ statistical models and algorithmic logic:

• Excel’s built-in forecasting tools leverage statistical algorithms to predict trends and future values, relying on historical data patterns.
• Business intelligence dashboards and visualization tools within Excel and related platforms use statistical and predictive approaches to offer insights and recommendations based on historical and real-time data patterns.
• Excel’s pattern recognition tools like Flash Fill and Column by Example are powered by AI techniques that identify data patterns to automate data entry and formatting tasks.

Given this familiarity, Excel users are already equipped with a foundation that can help them understand and contextualize the capabilities and limitations of generative AI.

Understanding Generative AI Risks: Familiar Concerns in New Forms

Despite the underlying familiarity, generative AI introduces nuances that feel particularly unsettling to Excel users:

Accuracy of Outputs

Generative AI can produce seemingly convincing answers that are factually incorrect, often referred to as “hallucinations.” This phenomenon understandably alarms professionals accustomed to the strict accuracy requirements typical in finance or accounting contexts.

However, consider this alongside traditional Excel practices. Incorrect formula results or misleading interpretations due to data entry errors or formula logic mistakes are similarly possible. Excel users have long managed such risks by using validation checks, audits, and peer reviews. Generative AI outputs require similar verification processes.

Sensitive Data Exposure

Another significant concern is the risk of sensitive data exposure when sharing information with generative AI models hosted externally. Excel users frequently handle sensitive financial or personal data, historically safeguarded through internal controls, encryption, and strict file-sharing protocols.

Again, the concern is familiar: cloud-based storage and sharing solutions, online accounting software, and email have long presented similar risks. The introduction of generative AI doesn’t fundamentally change the nature of these risks but underscores the need for robust data handling and governance strategies.

Why Generative AI Feels Particularly Risky to Excel Users

Generative AI presents unique challenges that intensify the perception of risk:

• Opaque operations: Unlike traditional Excel formulas or functions, generative AI often provides no clear indication of the source of its information, making validation more complex.
• Human-like outputs: AI-generated insights and responses frequently seem authoritative and well-informed, encouraging users to trust them implicitly without adequate scrutiny.
• High data stakes: Finance and accounting professionals often manage highly sensitive and regulated data, amplifying concerns around data security and compliance.

These factors contribute significantly to the heightened anxiety Excel users experience when integrating generative AI into their workflows.

Practical Strategies to Manage Generative AI Risks

Fortunately, Excel users have powerful tools and methods to address and mitigate the risks posed by generative AI:

• Enhanced Digital Literacy: Invest time in understanding how AI models function, including their limitations, can significantly reduce uncertainty. Workshops and training sessions focused explicitly on AI basics for Excel users can demystify these technologies, enabling users to approach generative AI more confidently and critically.
• Advanced Verification Techniques: Power Query is particularly valuable for validating and preparing data before submission to AI tools, ensuring cleaner inputs and reducing the likelihood of incorrect outputs. Python integration in Excel can further bolster verification practices by allowing more sophisticated data checks and validations that traditional Excel methods might struggle with.
• Data Governance Protocols: Establishing clear guidelines on what data can be safely shared with AI tools, as well as mandatory verification procedures for all AI-generated outputs, can significantly mitigate risks related to data privacy and accuracy.

Moving forward with confidence

Concerns about generative AI are understandable, especially among Excel users whose professional standards demand accuracy and transparency. Yet, these concerns are not entirely new, and Excel users already possess valuable skills and tools that can effectively manage and mitigate these risks.

By recognizing the familiar nature of these concerns and proactively strengthening their digital and computational literacy, particularly with tools like Power Query and Python, Excel users can transition smoothly into integrating generative AI within their workflows. Instead of viewing generative AI as a threat, users can harness it as a powerful enhancement to their analytical capabilities.

In short, the adoption of generative AI, while requiring caution and thoughtful management, is fully achievable for Excel professionals. By building upon their existing strengths and practices, Excel users can confidently embrace generative AI, reaping its substantial benefits while maintaining the rigorous standards their roles demand.

What questions or thoughts do you have about integrating generative AI into your Excel workflows and how to mitigate the risks? Please leave them below. And feel free to get in touch directly if you’d like personalized support or guidance on your and your organization’s journey toward AI adoption in Excel.

The post How Excel users can evaluate and mitigate the risks of generative AI first appeared on Stringfest Analytics.